IMAGE ENCODING / DECODING METHODS AND TOOLS

Abstract

The image encoding / decoding method and apparatus according to the present invention enables determination of an intra-screen prediction mode of a target block, generation of a prediction block from the target block based on the intra-screen prediction mode, and correction of the generated prediction block.

Description

The present invention relates to methods and apparatus for image encoding / decoding. Background of the Invention With the proliferation of the Internet and mobile terminals, and the development of information and communication technologies, the use of multimedia data has increased rapidly. Consequently, the need to improve the performance and efficiency of image processing systems to perform various services or tasks through predictive images in all types of systems has increased significantly. However, research and development efforts that can address this situation are still insufficient. Thus, in the image encoding and decoding methods and tools of traditional technologies, it is necessary to improve the performance for image processing, especially for image encoding or image decoding. The object of the present invention is to provide a method and apparatus for obtaining intra-mode prediction according to color components. The object of the present invention is to provide a method and apparatus for configuring reference pixels for intra-prediction. The object of the present invention is to provide an image encoding / decoding method and apparatus for modifying intra-prediction with arbitrary pixels. Brief Description of the Invention An image encoding / decoding method and apparatus according to the present invention can determine an intra-prediction mode of a target block, generate a prediction block from the target block based on the intra-prediction mode and modify the prediction block. In the image encoding / decoding method and apparatus according to the present invention, the intra prediction mode of the target block can be determined as a mode in a candidate group of prediction modes according to the state information of the target block. In the image encoding / decoding method and apparatus according to the present invention, when the color component of said target block is a luma component, a prediction mode candidate group comprising a directional mode and a non-directional mode can be referenced and when the color component of said target block is a chroma component, a prediction mode candidate group in which at least one of the directional mode, the non-directional mode, the color mode or the color copy mode is supported can be referenced. In the image encoding / decoding method and apparatus according to the present invention, the candidate group of prediction modes can be classified into a plurality of categories by considering a maximum number or a plurality of prediction modes capable of being included in each category. In the image encoding / decoding method and apparatus according to the present invention, the candidate group of prediction modes can be classified into a first category including non-directional modes and directional modes and a second category including color copy modes. In the image encoding / decoding method and apparatus according to the present invention, first information specifying one of a plurality of categories can be obtained and second information specifying an intra-prediction mode of the target block in the category according to the first information can be obtained. In the image encoding / decoding method and apparatus according to the present invention, the intra prediction mode of the target block can be determined from the determined category based on the first information and the second information. In the image encoding / decoding method and apparatus according to the present invention, the second information may not be obtained when only one prediction mode falls into a category according to the first information. In the image encoding / decoding method and apparatus according to the present invention, creating a reference pixel used for further intra-prediction may be included, but the reference pixel may have all or part of a plurality of reference pixel lines supported in the decoding apparatus. In the image encoding / decoding method and apparatus according to the present invention, performing at least one of a weighted filter or an interpolation filter for the constructed reference pixels may be further included. In the image encoding / decoding method and apparatus according to the present invention, modifying the prediction block can be selectively performed based on predetermined coding information and the coding information can include at least one of image type, color component, state information, coding mode, intra prediction mode, whether intra prediction in sub-block unit is applied or reference pixel line. According to the present invention, intra prediction mode can be effectively obtained. According to the present invention, the intra-prediction efficiency can be improved by selective use of predetermined reference pixel lines and filtering. According to the present invention, intra-prediction accuracy and coding performance can be improved by modifying the prediction block. Short Description of Image Figure 1 is a conceptual diagram of an image encoding and decoding system according to an embodiment of the present invention. Figure 2 is a block diagram of the components of an image encoding device according to an embodiment of the present invention. Figure 3 is a block diagram of the components of an image decoding device according to an embodiment of the present invention. Figure 4 is an exemplary diagram showing various partition shapes that can be obtained in a block partition unit of the present invention. Figure 5 is an example diagram showing the prediction mode in intra prediction according to an embodiment of the present invention. Figure 6 is a diagram of the arrangement of the target block and adjacent blocks according to an embodiment of the present invention. Figure 7 is a flowchart showing an intra-prediction modification method according to an embodiment of the present invention. Figure 8 is a diagram of the arrangement at pixels of the target block and adjacent blocks according to an embodiment of the present invention. Figures 9A and 9B are example diagrams of a modified method based on a compound reference pixel line according to an embodiment of the present invention. Figure 10 is a flowchart showing an intra-prediction modification method according to an embodiment of the present invention. Figures 11A, 11B, 11C, 11D and 11E are examples of diagrams at arbitrary pixels used to modify the prediction pixels according to embodiments of the present invention. Figures 12A, 12B, 12C, 12D, 12E and 12F are examples of diagrams where modifications are made based on arbitrary pixels according to embodiments of the present invention. Complete Description of the Invention An image encoding / decoding method and apparatus according to the present invention can determine an intra-prediction mode of a target block, generate a prediction block from the target block based on the intra-prediction mode and modify the prediction block. In the image encoding / decoding method and apparatus according to the present invention, the intra prediction mode of the target block can be determined as a mode in a candidate group of prediction modes according to the state information of the target block. In the image encoding / decoding method and apparatus according to the present invention, when the color component of said target block is a luma component, a prediction mode candidate group comprising a directional mode and a non-directional mode can be referenced and when the color component of said target block is a chroma component, a prediction mode candidate group in which at least one of the directional mode, the non-directional mode, the color mode or the color copy mode is supported can be referenced. In the image encoding / decoding method and apparatus according to the present invention, the candidate group of prediction modes can be classified into a plurality of categories by considering a maximum number or a plurality of prediction modes that can be included in each category. In the image encoding / decoding method and apparatus according to the present invention, the candidate group of prediction modes can be classified into a first category including non-directional modes and directional modes and a second category including color copy modes. In the image encoding / decoding method and apparatus according to the present invention, first information specifying one of the plurality of categories can be obtained and second information specifying the intra-prediction mode of the target block in the category according to the first information can be obtained. In the image encoding / decoding method and apparatus according to the present invention, the intra prediction mode of the target block can be determined from the determined category based on the first information and the second information. In the image encoding / decoding method and apparatus according to the present invention, the second information may not be obtained when only one prediction mode falls into a category according to the first information. In the image encoding / decoding method and apparatus according to the present invention, configuring reference pixels used for further intra-prediction may include, but said reference pixels may have all or part of a plurality of reference pixel lines supported in the decoding apparatus. In the image encoding / decoding method and apparatus according to the present invention, performing at least one of a weighted filter or an interpolation filter for configured reference pixels may be further included. In the image encoding / decoding method and apparatus according to the present invention, modifying the prediction block can be selectively performed based on predetermined coding information and the coding information can include at least one of image type, color component, state information, coding mode, intra prediction mode, whether intra prediction in sub-block unit is applied or reference pixel line. The present invention may be varied and modified and is illustrated by reference to different example embodiments, some of which will be described and shown in the drawings. However, these embodiments are not intended to limit the present invention but are interpreted as including all modifications, equivalents and substitutions that fall within the essential scope and technical scope of the present invention. Reference numbers of the same kind in the drawings refer to all elements of the same kind. Although the terms first, second, etc. may be used to describe various elements, these elements shall not be limited by these terms. These terms are used only to distinguish one element from another. For example, a first element may be called a second element and a second element may be called a first element as well without departing from the teachings of the present invention. The term and / or includes any and all combinations of any number of related items listed. It will be understood that when an element is referred to as being connected to or “coupled to another element, that element is either directly connected to or coupled to that other element or an intervening element. Conversely, when an element is referred to as being directly connected to or directly coupled to another element, no intervening element is present. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the present invention. As used herein, the singular forms a, an, and the are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will further be understood that the terms include and / or have, when used in this specification, specify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups. Unless otherwise specified, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those having ordinary skill in the art to which the invention relates. Terms generally used and defined in dictionaries should be interpreted as having the same contextual meaning as the technology concerned and unless expressly defined in the invention, they are not to be interpreted as idealized or overly formal. Generally, one or more color spaces can be configured according to the color format of the image. One or more images of a specific size or, one or more images of different sizes can be configured according to the color format. In the example, in the YCbCr color configuration, color formats such as 4:4:4, 4:2:2, 4:2:0, Monochrome (configured with only Y), etc. can be supported. In the example, for YCbCr 4:2:0, 1 luma component (in this example, Y) and 2 chroma components (in this example, Cb / Cr) can be configured and in this case, the configuration ratio of the chroma component and the luma component can be an aspect ratio of 1:2. In the example, for 4:4:4, they can have the same aspect ratio. When configured with one or more color spaces as in the example above, an image can partition into each color space. Such images can be classified into I, P, B, etc. According to the image type (e.g., image type, slice type, tile group type, tile type, brick type, etc.), image type I can be an image encoded alone without using a reference image, image type P can be an image encoded using a reference image, but allows only forward prediction and image type B can be an image encoded using a reference image and allows forward / backward prediction, but some of the above types can be combined according to the encoding arrangement (combining P and B) or image types in other configurations can be supported. Various encoded / decoded information generated in the present invention can be explicitly or implicitly processed. In this connection, it is understood that an explicit processing generates encoded / decoded information in sequences, slices, groups of tiles, tiles, bricks, blocks, sub-blocks, etc. to store them in a bit stream and decomposes the related information in the same units as the encoder in the decoder to reconstruct it into decoded information. In this case, it is understood that an implicit processing processes the encoded / decoded information in the same process, rules, etc. in the encoder and decoder. Figure 1 is a conceptual diagram of an image encoding and decoding system according to an embodiment of the present invention. Referring to Figure 1, an image encoding device (105) and decoding device (100) may be a user terminal such as a personal computer (PC), notebook, personal digital assistant (PDA), portable multimedia player (PMP), PlayStation Portable (PSP), wireless communication terminal, smart phone or TV, etc. or a server terminal such as an application server, service server, etc. and may include various devices equipped with communication devices such as communication modems, etc. for communicating with various wired and wireless communication instruments or networks, memory (120, 125) for storing all kinds of programs and data for inter or intra prediction for encoding or decoding images or processor (110, 115) for program operation and control by executing it, etc. In addition, an image encoded in a bit stream by the image encoding device (105) can be transmitted to the image decoding device (100) via a wired and wireless communication network, etc. such as the Internet, a wireless local area network, a wireless LAN network, a wibro network or a mobile radio communication network, etc. or via various communication interfaces such as a cable or a universal serial bus, etc. in real time or not in real time and decoded in the image decoding device (100). And can be reconstructed into an image and played back. In addition, an image encoded in a bit stream by the image encoding device (105) can be transmitted to the image decoding device (100) from the image encoding device (105) via a computer-readable recording medium. The image encoding means and image decoding means mentioned above may be separate means, respectively, but they may be constructed as a single image encoding / decoding means according to embodiments. In this case, multiple configurations of the image encoding means may be embodied so as to include at least the same structure or perform at least the same function as the multiple configurations of the image decoding means as substantially identical technical elements. Therefore, in the following detailed explanations on technical elements and their operating principles, etc., overlapping explanations on the corresponding technical elements will be omitted. In addition, since the image decoding tool corresponds to a computing tool that applies the image encoding method performed on the image encoding tool to decode, the image encoding tool will be mainly explained as follows. A computing device may include a memory storing a program or software module that embodies an image encoding method and / or an image decoding method and a processor connected to the memory for executing a program. In this case, an image encoding device may be referred to as an encoder and an image decoding device may be referred to as a decoder, respectively. Figure 2 is a block diagram of the components of an image encoding device according to an embodiment of the present invention. Referring to Figure 2, an image coding device (20) may include a prediction unit (200), a subtraction unit (205), a transformation unit (210), a quantization unit (215), a dequantization unit (220), an inverse transformation unit (225), a summation unit (230), a filter unit (235), an encoding image buffer (240) and an entropy coding unit (245). A prediction unit (200) can be realized using a prediction module, a software module, and can generate a prediction block in an intra-prediction method or an inter-prediction method for a block to be encoded. The prediction unit (200) can generate a prediction block by predicting the target block currently to be encoded in the image. In other words, the prediction unit (200) can generate a prediction block with the predicted pixel value of each pixel generated by predicting the pixel value of each pixel in the target block to be encoded in the image according to intra-prediction or inter-prediction. In addition, the prediction unit (200) can make the encoding unit encode information about the prediction mode by transmitting information necessary for generating the prediction block such as information about the prediction mode such as intra-prediction mode or inter-prediction mode to the encoding unit.In this case, the processing unit where the prediction is performed and the processing unit where the prediction method and concrete content are determined can be determined according to the coding settings. For example, the prediction method, prediction mode, etc. can be specified in the prediction unit and the prediction can be performed in the transformation unit. In the inter prediction unit, it can be divided into temporal prediction and spatial prediction based on the reference image. For temporal prediction, it can be a prediction method that finds Motion in an image that is temporally different from the current image and for spatial prediction, it can be a prediction method that finds motion in the current image (encoded region. A region that has been arranged adjacent to the target block) that is temporally the same as the current image. It can be integrated and managed by a list of reference images or can be managed by dividing the encoding mode. In addition, in the inter-prediction unit, it can be divided into translational motion models and non-translational motion models according to the motion prediction method. For translational motion models, prediction can be performed by considering only parallel translation and for non-translational motion models, prediction can be performed by considering motions such as rotation, distance, zoom in / out, etc. as well as parallel translation. When unidirectional prediction is assumed, translational motion models can require one motion vector, but non-translational motion models can require one or more motion information (e.g., one motion vector + rotation angle / scale factor, more than 2 motion vectors, etc. The following assumes the use of more than 2 motion vectors). For non-translational motion models, each motion vector can be information applied to a predefined position of the target block such as the top left vertex, the top right vertex, the bottom left vertex, etc.of the target block and can obtain the position of the target block region to be predicted by the corresponding motion vector in pixel units or sub-block units (integers above 2 such as 4 x 4, 8 x 8). For inter-unit prediction, some processes described later according to the Motion model can be applied generally and some processes can be applied separately. An inter prediction unit may include a reference image construction unit, a motion prediction unit, a motion compensation unit, a motion information determination unit and a motion information encoding unit. The reference image construction unit may include images encoded before or after the current image in a list of reference images (L0, L1). It may obtain prediction blocks from reference images included in the list of reference images and may include in at least one of the list of reference images as the current image configured with the reference image according to an encoding setting. In the inter prediction unit, the reference image construction unit may include a reference image interpolation unit and perform an interpolation process for decimal pixels according to an interpolation accuracy. For example, a DST-based 8-tap interpolation filter may be applied to the luma component and a DCT-based 4-tap interpolation filter may be applied to the chroma component. In the inter prediction unit, a motion prediction unit is the process of searching for blocks with high correlation with the target block through the reference image and can use various methods such as FBMA (Full search-based block matching algorithm), TSS (Three step search), etc. and motion compensation unit means the process of obtaining the prediction block by motion prediction process. In the inter prediction mode, a motion information determination unit can perform a process to select the optimum motion information from the target block and the motion information can be encoded with a motion information encoding mode such as a jump mode, a combined mode, a competition mode, etc. The node can be configured by combining the supported modes according to the motion model and the jump (translation) mode, jump (non-translation) mode, combined (translation) mode, combined (non-translation) mode, competition (translation) mode and competition (non-translation) mode can be examples. Several modes can be included in the candidate group according to the encoding settings. The motion information encoding mode can obtain prediction values ​​for motion information (motion vector, reference image, prediction direction, etc.) of the target block in at least one candidate block and when two or more candidate blocks are supported, optimum candidate selection information can be generated. In the jump mode (no residual signal) and the combined mode (with residual signal), the prediction value can be used as the motion information of the target block and in the competition mode, the motion information of the target block and the difference of the information value with the prediction value can be generated. A candidate group for the target block motion information prediction value can be adaptively and variably configured according to the motion information encoding mode. The motion information of blocks spatially adjacent to the target block (e.g., left block, top block, left-top block, right-top block, left-bottom block, etc.) can be included in the candidate group, the motion information of blocks temporally adjacent to the target block (e.g., left block, right block, top block, bottom block, left-top block, right-top block, left-bottom block, etc.) can be included in the candidate group. <tengah>in other images that respond to or correspond to the target block) may be included in the candidate group and mixed motion information from spatial candidates and temporal candidates (e.g., information obtained by averaging, mean, etc. from more than 2 candidates through motion information from spatially adjacent blocks and motion information from temporally adjacent blocks. Motion information may be obtained in the target block unit or in sub-block units for the target block) may be included in the candidate group. There can be a priority for configuring the motion information prediction candidate value group. The order included in the configuration of the prediction candidate value group can be determined according to the priority and the candidate group can be configured when the number of candidate groups (determined according to the motion information coding mode) is reached according to the priority. In this case, the priority can be determined in the order of motion information of spatially adjacent blocks, motion information of temporally adjacent blocks and mixed motion information of spatial candidates and temporal candidates, but can also be modified. For example, among spatially adjacent blocks, it can be included in the candidate group in the order from top-left - top-right - bottom-left - top-left block, etc. and for temporally adjacent blocks, it can be included in the candidate group in the order from bottom-right - middle - bottom-right block, etc. A subtraction unit (205) may generate a residual block by subtracting a prediction block from a target block. In other words, the subtraction unit (205) may generate a residual block, a block-shaped residual signal, by calculating the difference between the pixel value for each pixel of the target block to be encoded and the predicted pixel value for each pixel of the prediction block generated in the prediction unit. In addition, the subtraction unit (205) may generate a residual block within a unit except for the block unit obtained in the block partition unit described later. A transformation unit (210) can transform a signal belonging to the spatial domain into a signal belonging to the frequency domain and the signal obtained in the transformation process is called a transformed coefficient. For example, a transformation block having a transformed coefficient can be obtained by transforming a residual block having a residual signal transmitted from a reduction unit and an input signal determined according to a coding setting, which is not limited to a residual signal. A transformation unit can transform the residual block using transformation methods such as Hadamard Transformation, DST-Based Transformation (Discrete Sine Transformation), DCT-Based Transformation (Discrete Cosine Transformation), etc. and various improved and modified transformation methods can be used without being limited to it. At least one transformation method among the transformation methods may be supported and at least one detailed transformation method within each transformation method may be supported. In this case, the detailed transformation method may be a transformation method in which a portion of the basis vectors is differently configured in each transformation method. For example, for DCT, more than 1 detailed transformation method between DCT-1 to DCT-8 may be supported and for DST, more than 1 detailed transformation method between DST-1 to DST-8 may be supported. A candidate group transformation method may be configured by configuring a portion of the detailed transformation methods. In an example, DCT-2, DCT-8 and DST-7 may be configured as candidate group transformation methods to perform transformation. Transformations can be performed in the horizontal / vertical direction. For example, pixel values ​​from the spatial domain can be transformed into the frequency domain by performing a total two-dimensional transformation, a one-dimensional transformation in the horizontal direction using the DCT-2 transformation method, and a one-dimensional transformation in the vertical direction using the DST-7 transformation method. Transformation can be performed using a fixed transformation method or by selectively selecting a transformation method according to the coding settings. In this case, for the adaptive case, the transformation method can be selected using either an explicit or an implicit method. For the explicit case, each transformation method selection information or the selection information of the transformation method set applied to the horizontal and vertical directions can be generated in block units, etc. For the implicit case, a coding setting can be specified according to the image type (I / P / B), color component, block size / shape / position, intra-prediction mode, etc. and accordingly, a predetermined transformation method can be selected. Additionally, some transformations can be omitted depending on the encoding settings. In other words, this means that more than one of the horizontal / vertical units can be explicitly or implicitly omitted. In addition, a transformation unit may transform the information required to generate the transformation block to an encoding unit to encode it and store its information in a bit stream for transmission to a decoder, and a decoding unit of the decoder may decode its information for use in the reverse transformation process. A quantization unit (215) can quantize an input signal and in this case, a signal obtained in the quantization process is called a quantized coefficient. For example, a quantization block with quantized coefficients can be obtained by quantizing a residual block with transformed residual coefficients transmitted from a transformation unit and an input signal determined according to a coding setting, which is not limited to transformed residual coefficients. A quantization unit can quantize the transformed residual block using quantization methods such as Dead Zone Uniform Threshold Quantization, Weighted Matrix Quantization, etc. and various quantization methods improved and modified by it can be used without being limited to it. A quantization process can be omitted according to the coding settings. For example, a quantization process can be omitted (including the inverse process) according to the coding settings (e.g., the quantization parameter is 0. That is, a lossless compression environment). In another example, a quantization process can be omitted when the compression performance through quantization is not indicated according to the image characteristics. In this case, a region where the quantization process is omitted among the quantization blocks (M x N) can be all or part of the region (M / 2 x N / 2, M x N / 2, M / 2 x N, etc.) and the quantization omission selection information can be implicitly or explicitly specified. A quantization unit can transform the information required to generate a quantization block to a coding unit to encode it and store its information in a bit stream to transmit it to a decoder and a decoding unit of the decoder can decode its information to use it for the dequantization process. In the above example, it is explained with the assumption that the residue block is transformed and quantized by the transformation unit and the quantization unit, but the residue block with transformed coefficients can be generated by transforming the residue signal of the residue block and the quantization process can be omitted, only the quantization process can be performed without transforming the residue signal of the residue block into transformed coefficients, and both the transformation process and the quantization process can be omitted. This can be determined according to the coding settings. A dequantization unit (200) dequantizes a residual block dequantized by a quantization unit (215). In other words, a dequantization unit (220) produces a residual block having frequency coefficients by dequantizing a column of dequantized frequency coefficients. An inverse transformation unit (225) inversely transforms a residual block dequantized by a dequantization unit (220). In other words, an inverse transformation unit (225) generates a residual block having pixel values ​​by inversely transforming the frequency coefficients of the dequantized residual block, i.e., the reconstructed residual block. In this case, an inverse transformation unit (225) may perform the inverse transformation by inversely using the transformation method used in the transformation unit (210). A summation unit (230) reconstructs a target block by adding a prediction block predicted in a prediction unit (200) and a residual block reconstructed by an inverse transformation unit (225). The reconstructed target block can be stored as a reference image (or reference block) in an encoding image buffer (240) and can be used as a reference image when encoding subsequent blocks of the target block, other blocks or other images in the future. A filter unit (235) may include one or more post-processing filter processes such as a deblocking filter, SAO (Sample Adaptive Offset), ALF (Adaptive Loop Filter), etc. A deblocking filter may remove block distortions generated at the boundaries between blocks in a reconstructed image. The ALF may perform filtering based on a value obtained by comparing the reconstructed image after a block has been filtered through the deblocking filter with the original image. The SAO may reconstruct an offset different from the original image in pixel units for the residual block to which the deblocking filter is applied. Such post-processing filters may be applied to the reconstructed image or block. An encoding image buffer (240) may store reconstructed blocks or images in a filter unit (235). The reconstructed blocks or images stored in the encoding image buffer (240) may be made available to a prediction unit (200) that performs intra-prediction or inter-prediction. An entropy coding unit (245) scans the generated frequency coefficient column dequantized according to various scanning methods to generate a quantized coefficient column and encodes and outputs it using an entropy coding method, etc. A scanning pattern can be set as one of various patterns such as zigzag, diagonal line, raster, etc. In addition, encoding data including encoding information transmitted from each construction unit can be generated and output in a bit stream. Figure 3 is a block diagram of the components of an image decoding device according to an embodiment of the present invention. Referring to Figure 3, an image decoding device (30) may be configured to include an entropy decoding unit (305), a prediction unit (310), a dequantization unit (315), an inverse transformation unit (320), an adder-subtractor (325), a filter (330) and a decoding image buffer (335). Additionally, a prediction unit (310) may be configured to include an intra-prediction module and an inter-prediction module. First, when the image bit stream transmitted from the image encoding device (20) is received, it can be transmitted to the entropy decoding unit (305). An entropy decoding unit (305) can decode decoding data including dequantized coefficients by decoding the bit stream and decoding the information transmitted to each construction unit. A prediction unit (310) may generate a prediction block based on data transmitted from the entropy decoding unit (305). In this case, based on the reference images stored in the decoded image buffer (335), a list of reference images using a built-in configuration method may be configured. An inter-prediction unit may include a reference image construction unit, a motion compensation unit and a motion information decoding unit and some may perform the same process as an encoder and some may perform the derivation process in reverse. A dequantization unit (315) may dequantize the transformed quantized coefficients transmitted in the bit stream and decoded in the entropy decoding unit (305). An inverse transformation unit (320) may generate a residual block by applying an inverse DCT, an inverse integer transformation or a similar inverse transformation method to the transformed coefficients. In this case, the dequantization unit (315) and the inverse transformation unit (320) can be realized in various ways while reversibly performing a process performed on the transformation unit (210) and the quantization unit (215) of the image coding device (20) described above. For example, the same inverse transformation and process shared by the transformation unit (210) and the quantization unit (215) can be used and the transformation and dequantization processes can be reversibly performed with information about the transformation and quantization processes in the image coding device (20) (e.g., transformation size, transformation form, quantization type, etc.) A residual block after the dequantization and inverse transformation process can be added to the prediction block derived by the prediction unit (310) to produce a reconstructed image block. The addition can be performed by the adder-subtractor (325). A filter (330) may apply a deblocking filter to the reconstructed image blocks to eliminate blocking phenomena, if necessary and additionally use other loop filters to improve the image quality before and after the decoding process. A constructed and filtered image block may be stored in the decoding image buffer (335). Although not shown in the figure, a block partition unit may additionally be included in the image encoding / decoding tool. It can be partitioned into blocks of various units and sizes by the block partitioning unit. A basic coding unit (or maximum coding unit. Coding Tree Unit (CTU) can be the basic (or initial) unit for prediction, transformation, quantization, etc. in the image coding process. In this case, a basic coding unit can be configured with one basic luma coding block (or, maximum coding block. Coding Tree Block. Coding Tree Block CTB) and two basic chroma coding blocks according to the color format (in this example, YCbCr) and the size of each block can be determined according to the color format. And the coding block (CB - coding block) can be obtained according to the partitioning process. A coding block can be understood as a unit that is not partitioned into more coding blocks according to a fixed boundary and can be set as the initial unit for partitioning into lower units.In this invention, a block can be understood as a broad concept including various shapes such as triangles, circles, etc. without being limited to square shapes. It should be understood that the following mentioned content is targeted for one color component, but can be changed and applied to other color components in proportion to the ratio according to the color format (e.g., for YCbCr 4:2:0, the aspect ratio of the luma component and the chroma component is 2:1). In addition, it should be understood that block partitioning depending on other color components (e.g., in the case of depending on the result of the Y block partition in Cb / Cr) may be possible, but independent block partitioning of each color component may be possible. In addition, one common block partition arrangement (taking into account the proportions with the length ratio) can be used, but it is necessary to consider and understand that separate block partition arrangements are used according to the color components. In block partitioning unit, a block can be described as M x N and the maximum value and minimum value of each block can be obtained within a range. For example, when the maximum value of the block is set to 256x256 and the minimum value of the block is set to a 4x4 block, 2mx2n (in this example, m and n are integers from 2 to 8), a 2mx2m block (in this example, m and n are integers from 2 to 128) or a mxn block (in this example, m and n are integers from 4 to 256) can be obtained. In this case, m and n can be identical or non-identical and one or more ranges where a block for maximum value, minimum value, etc. is supported can be generated. For example, information about the maximum size, minimum size, etc. of blocks can be generated and information about the maximum size, minimum size, etc. of blocks in several partition arrangements can be generated. In this case, the former can be range information about the maximum and minimum sizes that can be generated in the image and the latter can be information about the maximum and minimum sizes that can be generated according to several partition arrangements. In this case, the partition arrangements can be specified by image type (I / P / B), color components (YCbCr, etc.), block type (encoding / prediction / transformation / quantization, etc.), partition type (Index or Type), partition method (QT, BT, TT, etc. in Tree method, SI2, SI3, SI4, etc. in Index method), etc. Additionally, there can be restrictions on the aspect ratio (block shape) that blocks may have and their boundary conditions can be set. In this case, only blocks below / less than an arbitrary boundary value (k) can be supported, and k can be set according to the aspect ratio such as A / B (A is the value longer than or equal to the width or height, B is the remaining value) and can be a natural number above 1 such as 1.5, 2, 3, 4, etc. As in the example above, boundary conditions on a single block shape in the image can be supported or one or more boundary conditions can be supported according to the partitioning settings. In short, whether a block partition is supported can be determined by the range and conditions mentioned above, the partition settings mentioned afterward, etc. For example, when the condition block that a candidate (child block) according to the block partition (parent block) is supported is met, the corresponding partition can be supported and vice versa, the corresponding partition can not be supported. A block partition unit can be configured to be associated with each construction unit for the image encoding and decoding device, through which the block size and shape can be determined. In this case, the block sets can be assigned differently according to the construction unit and prediction blocks for the prediction unit, transformation blocks for the transformation unit, quantization blocks for the quantization unit, etc. can correspond thereto. However, block units according to other construction units can be additionally assigned without being limited thereto. The present invention primarily describes a case where the input and output are rectangular in each construction unit, but input / output in other shapes (e.g., right triangle, etc.) may be possible in some construction units. The size and shape of the initial (or starting) block in a block partition unit can be determined by a higher unit. The initial block can be partitioned into smaller blocks, and when the optimum size and shape according to the block partition are determined, that block can be determined as the initial block in the lower unit. In this case, a higher unit can be a coding block and a lower unit can be a prediction block or a transformation block, but they are not limited to them and can be modified in various ways. As in the example above, when the initial block in the lower unit is determined, a partitioning process to find the optimum size and shape can be carried out like the higher unit. In summary, a block partitioning unit may partition a basic coding block (or a maximum coding block) into at least one coding block and a coding block may be partitioned into at least one prediction block / transformation block / quantization block. In addition, a prediction block may be partitioned into at least one transformation block / quantization block and a transformation block may be partitioned into at least one quantization block. In this case, some blocks may have dependent (i.e., defined by a higher unit and a lower unit) or independent relationships with other blocks. In an example, a prediction block may be a higher unit for a transformation block or an independent unit of a transformation block and various arrangement relationships may be possible according to the type of block. According to the coding arrangement, whether higher units and lower units are combined can be determined. In this case, the combination between units means that the coding process in the lower units (e.g., prediction unit, transformation unit, inverse transformation unit, etc.) is carried out by blocks in the higher units (size and shape) without partitioning from the higher units to the lower units. In other words, it can be that the partitioning process in a number of units is shared and the partitioning information is generated in one unit from it (e.g., a higher unit). In an example, (when an encoding block is combined with a prediction block and a transformation block), the prediction, transformation and reverse transformation processes can be performed in the encoding block. In the example, (when the coding block is combined with the prediction block), the prediction process can be performed in the coding block and the transformation and inverse transformation processes can be performed in a transformation block identical to or smaller than the coding block. In the example, (when the coding block is combined with the transformation block), the prediction process can be performed in the prediction block identical to or smaller than the coding block and the transformation and inverse transformation processes can be performed in the coding block. In an example, (when a prediction block is combined with a transformation block), the prediction process can be performed in a prediction block identical to or smaller than the coding block and the transformation process and reverse transformation can be performed in the prediction block. In the example, (when the combination is not performed on any block), the prediction process can be performed in a prediction block identical to or smaller than the coding block and the transformation and inverse transformation processes can be performed in a transformation block identical to or smaller than the coding block. The example above explains various cases of coding, prediction and transformation blocks, but is not limited to them. For combinations between units, fixed settings can be supported in the image or adaptive settings can be supported by considering various coding elements. In this case, such coding elements can include image type, color components, coding mode (Intra / Inter), partition settings, block size / shape / position, aspect ratio, prediction-related information (e.g., intra-prediction mode, inter-prediction mode, etc.), transformation-related information (e.g., transformation method selection information, etc.), quantization-related information (e.g., quantization region selection information, quantized transformation coefficient coding information, etc.), etc. As explained above, when a block with the optimum size and shape is found, its mode information (e.g., partition information, etc.) can be generated. The mode information can be stored in a bit stream with the information generated in the construction unit that has the block (e.g., prediction-related information, transformation-related information, etc.) and transmitted to the decoder and can be decomposed in the same unit in the decoder and used in the image decoding process. Next, the partitioning method will be explained and for ease of explanation, it is assumed that the initial block has a square shape, but since it can be equally or similarly applied even when the initial block has a rectangular shape, it is not limited to it. A block partitioning unit can support different types of partitions. For example, it can support tree-based partitioning or index-based partitioning, and other methods can be supported. A tree-based partitioning unit can specify the partition shape with various types of information (e.g., whether to partition, tree type, partition direction, etc.), and an index-based partition unit can specify the partition shape with predefined index information. Figure 4 is an example diagram showing the various partition shapes that can be obtained in the block partition unit of the present invention. In this example, it is assumed that the partition shape as in Figure 4 is obtained by performing a single partition (or process), but it can also be obtained by any number of partition moves without being limited to it. In addition, additional partition shapes not shown in Figure 4 may be possible. (Tree-based partitioning) In the tree-based partitioning of the present invention, Quad Tree (QT - Quad Tree), Binary Tree (BT - Binary Tree), Ternary Tree (TT - Ternary Tree), etc. can be supported. When one tree method is supported, it can be called as single tree partitioning and when more than two tree methods are supported, it can be called as multiple tree partitioning. QT means method (n) where a block is partitioned into two each (i.e., 4 divisions) in horizontal and vertical directions, BT means method (b to g) where a block is partitioned into two in horizontal or vertical direction and TT means method (h to m) where a block is partitioned into three in horizontal or vertical direction. In this case, QT can support 4-partition method (o, p) by dividing the partition direction into one of horizontal direction and vertical direction. In addition, BT can support only a method (b, c) with uniform size or only a method (d to g) with non-uniform size or can mix and support the two methods. In addition, TT can support only a method (h, j, k, m) with the arrangement that the partition is tilted in a certain direction (1:1:2, 2:1:1, etc. from left to right or from top to bottom) or can support only a method (i, l) arranged in the middle (1:2:1, etc.) or can mix and support the two methods. In addition, it is also possible to support a method (q) where the partition direction is partitioned into four each (i.e., 16 divisions) in horizontal direction and vertical direction. And, among the tree methods, it is possible to support the z-partition method (b, d, e, h, i, j, o) only in the horizontal partition direction or support the z-partition method (c, f, g, k, l, m, p) only in the vertical direction or mix and support the two methods. In this case, z can be an integer above 2 such as 2, 3 and 4. In this invention, it is explained on the assumption that QT supports n, BT supports b and c and TT supports i and l. Depending on the encoding settings, one or more tree partitioning methods can be supported. For example, QT can be supported, QT / BT can be supported, or QT / BT / TT can be supported. The above example is for the case where the basic tree partition is QT and BT and TT are included in the additional partitioning method according to whether other trees are supported, but various modifications are possible. In this case, the information about whether other trees are supported (bt_enabled_flag, tt_enabled_flag, bt_tt_enabled_flag, etc. These can have the value 0 or 1 and for 0, not supported and for 1, supported.) can be implicitly specified according to the encoding settings or can be explicitly specified in units of sequences, images, slices, tile groups, tiles, bricks, etc. Information about whether to partition (tree_part_flag or qt_part_flag, bt_part_flag, tt_part_flag, bt_tt_part_flag. It can have value 0 or 1 and for 0, it is not partitioned and for 1, it is partitioned.) can be included in the partition information. In addition, according to the partition method (BT and TT), information about the partition direction (dir_part_flag or bt_dir_part_flag, tt_dir_part_flag, bt_tt_dir_part_flag. It can have value 0 or 1 and for 0, <lebar horizontal>and for 1, <tinggi vertikal>) can be added, which can be generated when partitioning is performed. When multiple tree partitions are supported, partition information can be configured in various ways. The following will be explained by assuming a case where partition information is configured at one depth level (i.e., the supported partition depth is set to above 1, and recursive partitioning is possible, but for the sake of clarity). In example (1), confirming information about whether to perform a partition. In this case, when the partition is not performed, the partition is terminated. When partitioning is performed, confirm the selection information about the partition type (e.g., tree_idx. For 0, QT, for 1, BT and for 2, TT). In this case, additionally confirm the partition direction information according to the selected partition type and proceed to the next step. (If additional partitioning is possible because the partition depth is not reached the maximum, start over from the beginning, and when partitioning is not possible, the partition is terminated). In example (2), confirm the information about whether some tree method (QT) performs partitioning and proceed to the next step. In this case, when partitioning is not performed, confirm the information about whether some tree method (BT) performs partitioning. In this case, partitioning is not performed, confirm the information about whether some tree method (TT) performs partitioning. In this case, when partitioning is not performed, partitioning is terminated. When some tree (QT) methods perform partitioning, proceed to the next step. Additionally, when some tree (BT) methods perform partitioning, confirm the partition direction information and proceed to the next step. Additionally, when some tree (TT) partitioning methods perform partitioning, confirm the partition direction information and proceed to the next step. In example (3), confirming information about some tree methods (QT) performing partitioning. In this case, when partitioning is not performed, confirming information about whether some tree methods (BT and TT) performing partitioning. In this case, when partitioning is not performed, partitioning is terminated. When multiple tree methods (QT) perform partitioning, proceed to the next step. Additionally, when multiple tree methods (BT and TT) perform partitioning, confirm the partition direction information and proceed to the next step. The above examples can be cases where tree partition priority exists (examples No. 2 and 3) or does not exist (example No. 1), but can be variously modified. In addition, a case where the partition in the current step is not related to the partition result in the previous step is described in the above example, but the partition in the current step can also be set to depend on the partition result in the previous step. For example, in the case of examples No. 1 to 3, when partitioning of multiple tree (QT) methods is performed in the previous step to proceed to the current step, partitioning of the same tree (QT) methods can be supported also in the current step. On the other hand, when the partition of multiple tree methods (QT) was not performed in the previous step and the partition of multiple tree methods (BT or TT) was performed to proceed to the current step, the partition of multiple tree methods (BT and TT) excluding the partition of multiple tree methods (QT) can also be set to be supported in the next step including the current step. The above case means that the supported tree configuration for block partitioning can be adaptive, which means that the configuration of the partition information mentioned above can also be configured differently. (Assume that the example described later is example No. 3.) In other words, in the above example, when the partitioning of some tree methods (QT) was not performed in the previous step, the partitioning process can be performed in the current step without considering some tree methods (QT). In addition, partitioning information about the related tree methods (e.g., information about whether to partition, information about the partition direction, etc.) in this example <qt>, information about whether to partition) can be removed and configured. The above example is a case for adaptive partition information configuration in the case where block partitioning is allowed (e.g., block size is in the range between the maximum value and the minimum value, the partition depth of each tree method does not reach the maximum depth).<kedalaman yang diperbolehkan> , etc.) and adaptive partition information configuration can be possible even when block partitions are limited (e.g., block size is not in the range between the maximum and minimum values, the partition depth of each tree method reaches the maximum depth, etc.) As mentioned above, a tree-based partition in the present invention can be performed by a recursive method. For example, when the partition flag of the coding block that the partition depth is k is 0, the coding of the coding block is performed in the coding block that the partition depth is k and when the partition flag of the coding block that the partition depth is k is 1, the coding of the coding block is performed in the sub coding blocks N that the partition depth is k+1 according to the partition method (In this case, N is an integer above 2 such as 2, 3, 4). Such coding subblocks can be arranged as (k+1) coding blocks and partitioned into (k+2) coding subblocks in the above process and such hierarchical partitioning method can be determined according to partitioning settings such as partition range, allowed partition depth, etc. In this case, the bitstream structure for representing the partition information can be selected among one or more scanning methods. For example, the bitstream for partition information can be configured based on the partition depth order, or the bitstream for partition information can be configured based on whether to partition. For example, if based on partition depth order, it is a method for obtaining partition information at the next level depth after obtaining partition information at the current level depth based on the first block and if based on whether to partition, it means a method is preferred for obtaining additional partition information in the blocked partition based on the first block, and other additional scanning methods can be considered. The maximum block size or minimum block size can have a general setting regardless of tree type (or all trees) or can have separate settings according to each tree or can have a general setting to more than two trees. In this case, the maximum block size can be set to be identical to or smaller than the maximum encoding block. If the maximum block size according to the first predefined tree is not identical to the maximum encoding block, partitioning is implicitly performed using the second predefined tree method until the maximum block size of the first tree is reached. And, general partition depth can be supported regardless of tree type or separate partition depth can be supported according to each tree or general partition depth up to more than two trees can be supported. Alternatively, partition depth can be supported for some trees and partition depth can be unsupported for some trees. An explicit syntax element for setup information can be supported and some setup information can be implicitly specified. (Index-based partitioning) In the index-based partitioning of the present invention, the CSI (Constant Split Index) method, the VSI (Variable Split Index) method, etc. can be supported. A CSI method can be a method where k sub-blocks are obtained by partitioning in a predetermined direction and k can be an integer above 2 such as 2, 3, 4. In detail, it can be a partitioning method for configurations where the size and shape of the sub-blocks are determined based on the value of k regardless of the size and shape of the blocks. In this case, for a predetermined direction, one or two or more directions among the horizontal direction, vertical direction and diagonal direction (up-left -> down-right direction or down-left -> up-right direction, etc.) can be combined. The index-based CSI partitioning method of the present invention may include a candidate partitioned into z in one direction, horizontal or vertical. In this case, z may be an integer above 2 such as 2, 3, 4 and one of the width or height of each sub-block may be the same and the other may be the same or different. The aspect ratio of the sub-block may be A1:A2:...:AZ and A1 to AZ may be an integer above 1 such as 1, 2, 3. In addition, a candidate partitioned into x and y, respectively, in the horizontal and vertical directions can be included. In this case, x and y can be integers above 1 such as 1, 2, 3, 4, but when x and y are 1 at the same time (because a already exists), it can be restricted. Figure 4 shows the case where the width or height ratio of each sub-block is the same, but a candidate including different cases can be included. In addition, a candidate partitioned into w in one direction among several diagonal directions (up-left -> down-right direction) or several diagonal directions (down-left -> up-right direction) can be included and w can be an integer above 2 such as 2, 3. Referring to Figure 4, it can be classified into symmetric partition form (b) and asymmetric partition form (d, e) according to the length ratio of each sub-block and can be classified into (k, m) partition form tilted in a certain direction and (k) partition form arranged in the middle. The partition form can be specified by various coding elements including sub-block shape, etc. and the sub-block length ratio and supported partition form can be implicitly or explicitly specified according to the coding arrangement. Therefore, the candidate group in the index-based partition method can be determined based on the supported partition form. On the other hand, the VSI method can be a method where more than one sub-block is obtained by partitioning in a predetermined direction as the width (w) or height (h) of the sub-block is fixed and w and h can be integers above 1 such as 1, 2, 4, 8, etc. In detail, it can be a partitioning method for configuration where the number of sub-blocks is determined based on the size and shape of the block and the value of w or n. The index-based VSI partitioning method of the present invention may include candidates partitioned by specifying either a sub-block width or a height. Alternatively, it may include candidates partitioned by specifying both a sub-block width and a height. Since the sub-block width or height is specified, uniform partitioning in the horizontal or vertical direction may be permitted, but is not limited thereto. When a block is M x N before partitioning and the sub-block width is set (w) or height is set (h) or both width and height are set (w, h), the number of sub-blocks obtained can be (M*N) / w, (M*N) / h, (M*N) / w / h, respectively. According to the coding settings, only CSI methods can be supported, only VSI methods can be supported or both methods can be supported and information about the supported methods can be implicitly or explicitly specified. In this invention, it is described with the assumption that the method CSI supported. According to the encoding settings, candidate groups can be configured by including two or more candidates among the index partitions. For example, candidate groups such as {a, b, c}, {a, b, c, n}, {a to g, n} can be configured, which can be a case that the predicted block shape that will be generated significantly based on common statistical features such as block shape partitioned into two (2) in horizontal or vertical direction or partitioned into two (2) in horizontal or vertical direction are respectively configured as candidate groups. Alternatively, candidate groups such as {a, b}, {a, o}, {a, b, o} or {a, c}, {a, p}, {a, c, p} may be configured, which include candidates partitioned into two (2) and four (4) in the horizontal and vertical directions, respectively. It may be a case where the predicted block shapes that will be significantly partitioned in a particular direction are configured as candidate groups. Alternatively, candidate groups such as {a, o, p} or {a, n, q} may be configured, which may be the case where the predicted shape of the blocks to be partitioned is significantly smaller in size than the blocks before partitioning configured as candidate groups. Alternatively, a candidate group such as {a, r, s} can be configured, which can be a case where a non-square partition shape is configured as a candidate group according to the judgment that the optimum partition result that can be obtained in a rectangular shape by another method (tree method) in the block before the partition is obtained. As in the above example, multiple candidate group configurations are possible and more than one candidate group configuration can be supported by considering different coding elements. When the candidate group configuration is complete, various partition information configurations are possible. For example, index selection information can be generated in candidate groups configured to include unpartitioned candidates (a) and partitioned candidates (b through s). Alternatively, information indicating whether to partition (whether the partition form is a or not) may be generated and index selection information may be generated in candidate groups configured with candidates (b through s) that are partitioned when the partition is performed (if not a). Various configurations of partition information not included in the above description may be possible and except for information indicating whether to partition, binary bits may be assigned to the index of each candidate in the candidate group through various methods such as fixed-length binarization, variable-length binarization, etc. When the number of candidate groups is 2, 1 bit may be assigned to the index selection information and when the number of candidate groups is more than 3, 1 or more bits may be assigned to the index selection information. In contrast to tree-based partitioning methods, index-based partitioning methods can be methods for selectively configuring the predicted partition shapes to be generated significantly in candidate groups. And, since the number of bits to indicate index information can increase according to the number of supported candidate groups, it can be a method suitable for mono-hierarchical partitioning (e.g., partition depth limited to 0), rather than tree-based hierarchical partitioning (recursive partitioning). In other words, it can be a method that supports a single partition move and that can not additionally partition the sub-blocks obtained by index-based partitioning. In this case, it can mean that additional partitioning into smaller blocks within the same type is not possible (e.g., a coding block obtained by the index partitioning method cannot be additionally partitioned into coding blocks), but additional partitioning into blocks within different types is also set to be impossible (e.g., partitioning a coding block not only into coding blocks, but also into prediction blocks is not possible). Of course, it is not limited to the above examples and examples for other modifications can be possible. Next, a case where the block partitioning arrangement is determined based on the block type among the coding elements will be described. First, coding blocks can be obtained in the partitioning process. In this case, for the partitioning process, a tree-based partitioning method can be used and the resulting partition forms such as a (no splitting), n (QT), b, c (BT), i, l (TT), etc. In Figure 4 it can be displayed according to tree type. According to the coding settings, each tree type can be variously combined such as QT / QT+BT / QT+BT+TT, etc. The following example shows a process where the prediction block and the transformation block are finally partitioned based on the encoding blocks obtained in the process and assumes the case where the prediction, transformation and inverse transformation processes are performed based on each partitioned size. In example (1), the prediction process can be performed when the prediction block is set to the same size as the encoding block and the transformation and inverse transformation processes can be performed when the transformation block is set to the same size as the encoding block (or prediction block). Since the prediction block and transformation block are set based on the encoding block, no partition information is separately generated. In example (2), a prediction process can be carried out as prediction blocks set to the same size as the coding block. For the transformation block, the transformation block can be obtained in the partition process based on the coding block (or prediction block) and the transformation and inverse transformation processes can be carried out based on the obtained size. In this case, a tree-based partition method can be used for the partition process and the resulting partition forms such as a (without separation), b, c (BT), i, l (TT), n (QT), etc. In Figure 4 it can be displayed according to tree type. According to the coding settings, each tree type can be variously combined such as QT / BT / QT+BT / QT+BT+TT, etc. In this case, an index-based partitioning method can be used for the partitioning process and the results of the partitioning form such as a (without splitting), b, c, d, etc. In Figure 4 can be output according to the index type. According to the encoding arrangement, various candidate group configurations such as {a, b, c}, {a, b, c, d}, etc. can be possible. In example (3), for the prediction block, a prediction block can be obtained in the partition process based on the encoding block and the prediction process can be performed based on the obtained size. A transformation block can be set the same as the size of the encoding block to perform the transformation and inverse transformation processes. This example can correspond to the case where the prediction block and the transformation block have a mutually independent relationship. In this case, an index-based partitioning method can be used in the partitioning process and the results of the partition form such as a (without splitting), b to g, n, r, s, etc. In Figure 4 can be output according to the index type. According to the encoding arrangement, various candidate group configurations such as {a, b, c, n}, {a to g, n}, {a, r, s}, etc. can be possible. In example (4), for a prediction block, a prediction block can be obtained in the partitioning process based on the encoding block and the prediction process can be performed based on the obtained size. The transformation block can be set to the same size as the prediction block to perform the transformation and inverse transformation processes. This example can be a case where the transformation block is set to the same size as the obtained prediction block or vice versa (e.g. the prediction block is set to the same size as the transformation block). In this case, a tree-based partition method can be used in the partition process and partition forms such as a (no separation), b, c (BT), n (QT), etc. In Figure 4 it can be generated according to tree type. According to the coding settings, each tree type can be variously combined such as QT / BT / QT+BT, etc. In this case, an index-based partitioning method can be used in the partitioning process and partition forms such as a (without splitting), b, c, n, o, p, etc. in Figure 4 can be generated according to the index type. According to the encoding settings, various candidate group configurations such as {a, b}, {a, c}, {a, n}, {a, o}, {a, p}, {a, b, c}, {a, o, p}, {a, b, c, n}, {a, b, c, n, p}, etc. can be possible. In addition, candidate groups can be configured by the VSI method alone or by mixing the VSI method with the CSI method among the index-based partitioning methods. In example (5), for a prediction block, a prediction block can be obtained in a partitioning process based on an encoding block and a prediction process can be performed based on the obtained size. In addition, for a transformation block, a prediction block can be obtained in a partitioning process based on an encoding block and a transformation and inverse transformation process can be performed based on the obtained size. This example can be a case where the prediction block and the transformation block are partitioned respectively based on an encoding block. In this case, a tree-based partitioning method and an index-based partitioning method can be used in the partitioning process and the candidate groups can be configured the same as or similarly to example No.4. The above example describes several cases that can be generated depending on whether the partitioning process for each block type is divided, etc., it is not limited to it and examples for various modifications are possible. In addition, the block partition settings can be determined by considering various coding elements as well as block types. In this case, a coding element may include image type (I / P / B), color component (YCbCr), block size / shape / position, block aspect ratio, block type (coding block, prediction block, transformation block, quantization block, etc.), partition state, coding mode (Intra / Inter), prediction related information (intra prediction mode, inter prediction mode, etc.), transformation related information (transformation method selection information, etc.), quantization related information (quantization region selection information, quantized transformation coefficient coding information, etc.), etc. Figure 5 is an example diagram showing the prediction mode in intra prediction according to an embodiment of the present invention. Referring to Figure 5, 95 prediction modes can be supported for intra prediction and among them, 93 prediction modes are directional modes and 2 prediction modes are non-directional modes (DC, Planar). In this case, the directional modes can be classified by inclination (e.g., dy / dx) or Degree angle information). A non-directional mode can be a method that performs prediction by methods such as averaging, interpolation, etc. from reference pixels adjacent to the block and a directional mode can be a method that performs prediction by methods such as extrapolation, interpolation, etc. from reference pixels adjacent to the block. A direction mode can have vertical direction (up>down / down->up), horizontal(left->right / right->left), diagonal A (left-up->right-down / right-down->left-up), diagonal B (right-up->left-down / left-down->right-up), etc. Generally, when encoding is performed by raster scanning, etc., there can be adjacent blocks with left, top, top-left, top-right, bottom-left directions and non-directional modes that consider them as reference pixels or starting points or for prediction or prediction modes in vertical (top>bottom), horizontal (left->right), diagonal A (top-left->bottom-right), diagonal B-1 (top-right->bottom-left), diagonal B-2 (bottom-left->top-right) directions can be supported. When other scanning except for raster scanning is supported, the direction mode can be differently set. For example, when the left block is available and available, the horizontal mode can perform the right block no prediction by extrapolation in the direction (right->left). Alternatively, when both the left block and the right block are available, the prediction can be performed by extrapolation in a predetermined direction (left->right or right->left) or the prediction can be performed by interpolation of both blocks. In the present invention, the encoding is performed according to the raster or z-scan sequence and is described by assuming that the reference pixels are positioned in the left, top, top-left, top-right, bottom-left directions. In addition, blocks reconstructed from color spaces encoded using correlations between color spaces can be used for target block prediction and prediction modes supporting it can be included. For example, for a chroma component, a prediction block of a target block can be generated using a reconstructed block of a luma component corresponding to the target block. In other words, a prediction block can be generated based on a reconstructed block taking into account correlations between color spaces and can be included as an intra-prediction mode for the chroma component. A chroma component may have the same candidate group as the prediction mode candidate group of the luma component or multiple modes among the prediction mode candidate group of the luma component and additional prediction modes of the chroma component (color copy mode, color mode) may be included in the prediction mode candidate group. In this case, for the color copy mode, it may be a prediction mode associated with a data acquisition method for generating prediction blocks from regions positioned in another color space and for the color mode, it may be a prediction mode associated with a method for obtaining prediction modes from regions positioned in another color space. For the color copy mode and the color mode, m and n modes (m, n are integers such as 0, 1, 2, 3 or more) may be supported, respectively. When 1 color copy mode is supported, a predefined data acquisition method for generating prediction blocks can be predefined and when two or more color copy modes are supported, a data acquisition method for generating prediction blocks can be classified (e.g., referenced positions for obtaining correlations, etc.) area <1> left, <2> on, <3> left + top, etc.) and supported. When 1 color mode is supported, a predefined position for obtaining the prediction mode can be set in advance and when two or more color modes are supported, a number of positions (e.g., <1> middle, <2> top left, <3> top right, <4> bottom left, <5> bottom-right, etc. for the corresponding block) to obtain the prediction mode can be supported. All or some of the prediction modes described in the examples above may be included in the prediction mode candidate group of the luma component or the chroma component and other additional modes may be included in the prediction mode candidate group. The prediction modes can be all the candidates for the supported intra-prediction modes, and the candidate group of prediction modes can be configured with all or some of them. In this case, the candidate group of prediction modes can be configured according to the size, shape (aspect ratio), etc. of the block. For example, the number of candidate groups of such prediction modes can be specified according to the block size. In this case, the size block can be classified into one of more than two ranges divided based on one or more predetermined threshold sizes (A x B, C x D, etc.) and can be specified as one of 11, 35, 67, etc. and the number of candidate groups according to the classified range or whether color copy mode, color mode, etc. are supported and the number can be specified. In this case, the threshold size can be indicated as width (W), height (H), W x H, etc. and W and H can be integers above 2 such as 4, 8, 16, 32, 64, etc. Alternatively, according to the block shape (or block aspect ratio), the configuration of the candidate group of the prediction mode can be determined. In this case, the block aspect ratio can be classified as one of two or more divided ranges based on one or more predetermined threshold values ​​and according to the classified ranges, the configuration of the candidate group can be determined. In this case, such threshold values ​​can be indicated as W / H (or H / W), etc. and can have integer values ​​above 1 such as 1, 2, 4, 8, 16, etc. or decimal values ​​between 0 and 1 such as 1, 1 / 2, 1 / 4, 1 / 8, 1 / 16, etc. The intra-prediction of the present invention assumes a case where 95 prediction modes as in Figure 5 can be supported for directional and non-directional modes and color modes and color copy modes can be supported. In addition, it is assumed that for the prediction mode candidate group of the luma component, 67 directional modes and non-directional modes are supported and for the prediction mode candidate group of the chroma component, a total of 8, 4 directional and non-directional modes, 1 color mode and 3 color copy modes, are supported, but not limited thereto. A predetermined standard for configuration may exist for selecting a mode included in the prediction mode candidate group except for the aforementioned assumptions. The following shows an example for configuring candidate groups of prediction modes according to block aspect ratio. [Table 11 pred mode ratio (including) pred mode (except) 16 67 ~ 80 2 ~ 15 8 67 ~ 78 2 ~ 13 4 67 ~ 7 6 2 ~ 11 2 67 ~ 72 2 ~ 7 1 - - 1 / 2 -6 ~ -1 61 ~ 66 1 / 4 -10 ~ -1 57 ~ 66 1 / 8 -12 ~ -1 55 ~ 66 1 / 16 -14 ~ -1 53 66 The table is arranged based on a case where prediction modes No. 0 and 1 and prediction modes No. 2 to 66 are included in the candidate group of prediction modes when having a square block shape. By a prediction mode being added to or excluded from the candidate group of prediction modes (excluding non-directional modes) of a square block when having a rectangular block shape, a candidate configuration shown in the table may be possible, which may be an example arranged assuming that it is preferable in terms of prediction accuracy to configure a wider prediction mode in a longer block than in a shorter block, but may allow the opposite. A reference pixel used for intra-prediction can be configured in the reference pixel construction unit. In this case, a reference pixel can be managed in a temporary memory (e.g., Array. First, second array, etc.) and can be generated and deleted in each intra-prediction process, and the size of the temporary memory can be determined according to the configuration of the reference pixel. This is explained by assuming that the left, top, top-left, top-right and bottom-left blocks are used for intra-prediction based on the target block, but not limited to it and candidate group blocks from other configurations can be used for intra-prediction. For example, candidate groups of adjacent blocks for such a reference pixel may be an example of the case where a raster or Z scan is followed and according to the supported scan order, some candidate groups may be removed or configured by including other candidate groups of blocks (e.g., additional configuration with right, bottom, bottom-right, etc. blocks). On the other hand, a pixel adjacent to the target block can be classified into at least one layer of reference pixels and can be classified into ref_0, the pixel closest to the target block {the pixel whose pixel value difference with the boundary pixel of the target block is 1. p(1,-1) ~ p(2m-1,-1), p(-1,0) ~ p(-1,2n-1)}, ref_1, the next adjacent pixel {the pixel value difference with the boundary pixel of the target block is 2. p(-2,-2) ~ p(2m,-2), p(-2,-1) ~ p(-2,2n)}, ref_2, the next adjacent pixel {the pixel value difference with the boundary pixel of the target block is 3. p(-3,-3) ~ p(2m+1, -3), p(-3,-2) ~ p(-3,2n+1)}, etc. In other words, the reference pixels can be classified into a number of reference pixel layers according to the distance of adjacent pixels to the boundary pixels of the target block. The reference pixel lines supported here can be above N and N can be an integer such as 1, 2, 3, 4, or more. In this case, generally included in the reference pixel lines are the candidate groups sequentially from the reference pixel lines closest to the target block, but not limited to them. For example, if N is 3, the candidate groups can be sequentially configured as<ref_0, ref_1, ref_2> or candidate groups can also be configured by excluding non-sequential reference pixel lines such as<ref_0, ref_1, ref_3> ,<ref_0, ref_2, ref_3> ,<ref_1, ref_2, ref_3> or closest. Prediction can be done by using all reference pixel lines in the candidate group or by using multiple reference pixel lines (more than one). For example, according to the encoding settings, one of a plurality of reference pixel lines can be selected to perform intra-prediction using the corresponding reference pixel line. Alternatively, two or more of a plurality of reference pixel lines can be selected to perform intra-prediction using the corresponding reference pixel line (e.g., applying weighted average, etc. to each reference pixel line data) In this case, the selection of the reference pixel line can be implicitly or explicitly specified. For example, for the implicit case, it means that it is determined according to the coding settings defined by more than one or two combinations among elements such as image type, color components, block size / shape / position, etc. In addition, for the explicit case, it means that the information about the selection of the reference pixel line can be generated in block units, etc. The present invention mainly describes a case where intra-prediction is performed using the most adjacent reference pixel line. However, a reference pixel line referenced for prediction can be considered as one of the main coding elements in the present invention. In other words, the intra-prediction setting can be determined regardless of the selected reference pixel line, but its intra-prediction setting can be determined. The reference pixel construction unit of intra prediction in the present invention may include a reference pixel generating unit, a reference pixel interpolation unit, a reference pixel filter unit, etc. and may be configured by including all or part of the configuration. In the reference pixel construction unit, available reference pixels and unavailable reference pixels can be classified by configuring the availability for the reference pixels. In this case, a reference pixel is considered unavailable when at least one of the following conditions is met. For example, it can be considered to be unavailable when at least one of the cases where it is positioned outside the image boundary, a case where it does not include the same partition unit as the target block (e.g., units that can not be referenced to each other such as slices, tiles, etc. However, when even a unit such as slices or tiles, etc. can be referenced to each other, is excluded even though it is not the same partition unit) and a case where the encoding is not completed is satisfied. In other words, it can be judged to be available when one of the conditions is not satisfied. In addition, the use of reference pixels may be restricted by encoding settings. For example, even though they are judged to be available under the conditions, the use of reference pixels may be restricted according to whether constrained intraprediction (e.g., constrained_intra_pred_flag) is performed. Constrained intraprediction may be performed to prohibit reconstructed blocks referenced from other images to prevent propagation of errors caused by external factors including communication environment, etc. from being used as reference pixels. When constrained intra-prediction is disabled (e.g., for image type I. Or image type P or B, constrained_intra_pred_flag = 0), all candidate blocks of reference pixels can be available. Alternatively, when constrained intra-prediction is enabled (e.g., for image types P or B, constrained_intra_pred_flag = 1), it is assumed by the evaluation condition whether the reference pixel candidate block is used according to the coding mode (Intra_Mode, Inter_ModeD, Inter_ModeC), but the condition can be specified according to various other coding elements. In this case, Mode_Intra, Mode_Inter_D, Mode_Inter_C can be intra prediction, inter prediction referenced in other images (block matching, etc.) and inter prediction referenced in the current image respectively, and the settings where in the case of Mode_Intra, reference is possible, and in the case of Mode_Inter, reference is not possible, can be general, but not limited to it. Since a reference pixel consists of more than one block, it can be classified into three cases such as<semua tersedia> ,<beberapa tersedia> And<semua tidak tersedia> after checking the possible reference pixels. In other cases except for the case<semua tersedia> , reference pixels at unavailable candidate block positions can be filled or generated. When a reference pixel candidate block is available, pixels at corresponding positions can be included in the reference pixel memory of the target block. In this case, the pixel data can be copied as is or can be included in the reference pixel memory in a process such as reference pixel filtering, reference pixel interpolation, etc. In addition, when a reference pixel candidate block is not available, pixels obtained in the process to generate reference pixels can be included in the reference pixel memory of the target block. The following shows examples where reference pixels at unavailable block positions are generated using various methods. For example, a reference pixel can be generated using arbitrary pixel values. In this case, an arbitrary pixel value can be a single pixel value (e.g., the minimum value, maximum value, middle value, etc. within a range of pixel values) that falls within a range of pixel values ​​(e.g., a range of pixel values ​​based on bit depth or according to the distribution of pixels in the corresponding image). Specifically, it can be an example applied when all candidate blocks of reference pixels are not available. Alternatively, a reference pixel can be generated from the region where the image is encoded. Specifically, a reference pixel can be generated from at least one adjacent available block with an unavailable block. In this case, at least one of the methods such as extrapolation, interpolation, copying, etc. can be used. For the above example, a reference pixel may target a region centered around the target block and may be considered to include a region (a region that obtains prediction data or a region adjacent to it) that corresponds to the target block in another color space referenced in some prediction mode (e.g., color copy mode, etc.). Once the reference pixel configuration is complete, reference pixel filtering or reference pixel interpolation can be performed. For example, only reference pixel filtering can be performed, only reference pixel interpolation can be performed, or both reference pixel filtering and reference pixel interpolation can be performed. Reference pixel filtering can be performed before or after reference pixel interpolation, and can also be performed in combination with reference pixel interpolation at the same time. A reference pixel filtering can be a process performed to reduce the residual degradation in a reference pixel. A reference pixel filtering can be one of a number of filters that can be classified by setting the length and coefficients of various filter beats such as [1,2,1] / 4, [2, 3, 6, 3, 2] / 16, etc. In addition, a number of filtering can be performed by varying the type of filter. Whether reference pixel filtering is performed can be explicitly specified or whether reference pixel filtering is performed can be implicitly specified according to the coding settings. In this case, coding settings can be set based on the state information (size, shape, block position, etc.) of the target block, image type (I / P / B), color components (Y / Cb / Cr), reference pixel line selection information, whether intra prediction in sub-block units is applied, intra prediction mode, etc. Reference pixel interpolation can not be performed in prediction mode which only refers to pixels in integer units and reference pixel interpolation can be performed in prediction mode which refers to pixels in decimal units. The pixel position (i.e., which decimal unit is interpolated) at which the interpolation is performed can be specified according to the prediction mode (e.g., direction for prediction mode. dy / dx, etc.) and the reference pixel position and the prediction pixel. In this case, one filter can be applied regardless of the accuracy of the decimal unit or one of a number of filters (e.g., assuming a filter that an equation used to determine the filter coefficient or the filter beat length is divided) can be applied selected in decimal units. The foregoing may be an example where a pixel in integer units is used as input for interpolation of pixels in decimal units and the latter may be an example where the input pixels vary per step (e.g., for 1 / 2 unit, using integer pixels. For 1 / 4 unit, using integer and 1 / 2 unit pixels, etc.), but is not limited thereto and is described based on the foregoing in the present invention. A fixed filtering or adaptive filtering can be performed for reference pixel interpolation, which can be determined according to the coding settings. In this case, the coding settings can be set based on the target block state information, image type, color components, reference pixel line selection information, whether intra-prediction in sub-block units is applied, intra-prediction mode, etc. A fixed filter can interpolate reference pixels using a single filter and an adaptive filter can interpolate reference pixels using any of a number of filters. In this case, for adaptive filtering, one of a number of filters can be implicitly or explicitly specified according to the encoding settings. In this case, the filter type can be configured after more than one is selected from a 4-tap DCT-IF filter, a 4-tap cubic filter, a 4-tap Gaussian filter, a 6-tap Wiener filter, an 8-tap Kalman filter, etc. and the supported filter candidate groups can also be assigned differently according to the color components (e.g., some filter types are the same or different, and the filter tap length is short or long, etc.) A prediction block can be generated according to at least one prediction mode in a prediction block generating unit and a reference pixel can be used based on the prediction mode. In this case, a reference pixel can be used for methods such as extrapolation, etc. according to the prediction mode and can be used for methods such as interpolation, DC, or copy, etc. A process for selecting the optimum mode among a number of candidate prediction mode groups is carried out in the prediction mode determination unit. Generally, the optimum mode in terms of coding cost can be determined using block distortion {e.g., Distortion of the target block and the reconstructed block. Distortion. SAD (Sum of Absolute Difference), SSD (Sum of Square Difference), etc.} and the distortion rate that the number of bits generated according to the corresponding mode is considered. A prediction block generated based on the prediction mode determined in the process can be transmitted to the subtraction unit and the summation unit. A prediction mode selected in a prediction mode determination unit can be encoded in a prediction mode of an encoding unit. In a candidate group of prediction modes, index information corresponding to a prediction mode can be encoded or a prediction mode can be predicted to encode information about it. The former can be a method applied to the luma component and the latter can be a method applied to the chroma component, but is not limited to them. When a prediction mode is predicted and encoded, it can be classified into a number of categories (K. K is an integer such as 2, 3, 4, or more) and managed. For example, when classified into 2 categories, one category can consist of prediction modes that are predicted to be identical to the prediction mode of the target block and another category can consist of predictions that are predicted to be unlikely to be identical to the prediction mode. In other words, another category can consist of prediction modes that are not selected as modes that are predicted to be identical to the prediction mode of the target block. The above example is a case where a prediction mode is classified into 2 categories and can be classified into more categories. When classified into 3 categories, the first category can consist of prediction modes that are predicted to be very likely to be identical to the target block prediction mode, the second category can consist of prediction modes that are predicted to be likely to be identical to the target block prediction mode and the third category can consist of prediction modes that are predicted to be unlikely to be identical to the target block prediction mode. In other words, the second category can consist of prediction modes that are predicted to be identical to the target block prediction mode among prediction modes that do not belong to the first category and the third category can consist of prediction modes that do not belong to the first and second categories. For another example of classifying categories, they can be classified according to the prediction method. For example, they can be classified by the prediction method by extrapolation, prediction method by interpolation / average, prediction method by copy, etc. In this case, the prediction method by extrapolation can be a directional mode, the prediction method by interpolation / average can be a non-directional mode, the prediction method by Copy can be classified into color copy mode and various other classifications can be possible. Since the color mode gets the prediction mode in another color space, it can be distinguished from the directional or non-directional mode, but for ease of explanation, it is assumed that it is classified into directional / non-directional mode. When classified into 2 categories according to the prediction method, the first category can consist of directional / non-directional modes and the second category can consist of color copy modes. Alternatively, the first category can consist of directional modes and the second category can consist of non-directional modes. The aforementioned can be applied to the intra-prediction of the chroma component and the latter can be applied to the intra-prediction of the luma component. In the sections described afterward, it is assumed that when the color copy mode is included, it can be applied to the intra-prediction mode of the chroma component, but it should be understood that even when the color copy mode is not included, it does not apply only to the intra-prediction mode of the luma component. When classified into 3 categories, the first category may consist of directional modes, the second category may consist of non-directional modes and the third category may consist of color copy modes. Alternatively, the first category may consist of predefined modes (e.g., vertical, horizontal, diagonal A modes, etc.) among directional modes, the second category may consist of predefined modes (e.g., diagonal B modes, etc.) among directional modes and the third category may consist of color copy modes. In this case, diagonal A modes may consist of diagonal modes that use only integer pixels in prediction and diagonal B modes may consist of diagonal modes that use integer pixels and decimal pixels in prediction, but are not limited thereto. The description above is an example of a category configuration and can be classified according to various predefined standards without being limited to them. Furthermore, any number of predefined standards can be combined and applied to a category configuration. When prediction modes are classified into multiple categories, selection information about the target block prediction mode's category is generated, and candidate selection information within each category can be generated. When the candidate in a category is 1, candidate selection information can be omitted. It is possible to configure various syntax elements for category selection and candidate selection within a category. Next, a case where 3 categories are supported is assumed. * one syntax element on category selection is supported. The supported syntax element values ​​for category selection in this example can have values ​​0 to 2 (values ​​above 1). In the example, for the first to third categories, the binarization where the bit <0> , <10> , <11> defined can be possible. In other words, a single syntax element for category selection can be supported and can be configured in the type that the indexing and binarization for each category are applied. * A number of syntax elements are supported for category selection. The supported syntax element values ​​for category selection in this example can be 0 or 1 (also any value above 1). In the example, for one syntax element (the first syntax element), <0> can be assigned to the first category and the second category and <1> can be assigned to the third category. In this case, when the syntax element has a value <0> , another syntax element (second syntax element) on the subcategory selection that selects one of the first category and the second category can be supported and for the corresponding syntax element, <0> , <1> can be assigned to the first category and the second category. In the example, for one syntax element (the first syntax element), <0> can be assigned to the first category and <1> can be assigned to the second and third categories. In this case, when the syntax element has a value <1> , another syntax element (second syntax element) on the subcategory selection that selects one of the second category and the third category can be supported and for the corresponding syntax element, <0> , <1> can be assigned to the second category and the third category. The above example is an example of the configuration of syntax elements for category selection and various modifications and configurations are possible. Each of the above categories can include at least one prediction mode. The number of prediction modes in the first to third categories can be a, b, c, and a can be less than or equal to b and c and b can be less than or equal to c. In this example, it is assumed that a is an integer between 1 ~ 5, b is an integer between 3 ~ 8 and c is a number that subtracts the numbers a and b from the number of candidate groups of prediction modes. Variable-length binarization or fixed-length binarization such as truncated Rice Binarization, truncated Binarization, k-th Exp-Golomb binarization, etc. can be applied to the prediction modes belonging to these categories. In the example mentioned afterward, it is explained that the configuration and arrangement of categories are fixed, but adaptive arrangement can be possible according to various coding elements (e.g., number of categories, associated flag configuration, etc.). In this case, coding elements can be set based on color components (luma / chroma), block state information (e.g., size, shape, aspect ratio, block position, etc.), image type (I / P / B), whether intra prediction in sub-block units is applied, reference pixel line selection information, etc. In the example, based on the coding information, the number of categories can be determined to be either 2 or 3. Next, an example will be explained where the prediction mode is classified into a plurality of categories (3) based on the matching probability of the prediction mode and the prediction mode is encoded based on the category selection information and the candidate selection information in a category. However, an example where the standards for category classification according to the prediction method are combined for category configuration may be possible. In addition, when the second category between the first category and the second category is excluded or the explanation of the first category and the second category is integrated, the explanation configured with 2 categories may be derived, so that the detailed explanation is omitted. Figure 6 is a diagram of the arrangement of the target block and adjacent blocks according to an embodiment of the present invention. A target block can have a high correlation with its adjacent blocks, and can not only perform predictions using adjacent reference pixels, but also use the prediction modes of adjacent blocks to predict the prediction mode of the target block. Therefore, the prediction modes of blocks adjacent to the target block can be selected as candidates belonging to the first or second category. In this case, all or some of the prediction modes of the left, top, top-left, top-right and bottom-left blocks for the target block can be considered as candidates belonging to the (previous) category. In the example, (left / top), (left / top / top-right), (left / top / bottom-left), (left / top / top-left / bottom-right / top-left), etc. can be considered. When the prediction modes of adjacent blocks are considered as candidates for belonging to a category, at least one prediction mode can be selected from a block in each direction. For example, for the left block, one can be selected in L0 to L3 and for the top block, one can be selected in U0 to U3. A sub-block at a given position (e.g., the left block is L3, the top block is U3) for selecting the prediction mode can be determined by considering a case where an adjacent block in a given direction (in this example, the left, top block) is partitioned into a number of sub-blocks. When the prediction mode for the block at the corresponding position is not available (e.g., when the corresponding block is not encoded, when it is at a position that cannot be referenced, when the encoding mode <intra inter>different, etc.), the prediction mode can be deselected in blocks in the corresponding direction. Alternatively, priority for selecting prediction modes (sub-blocks) can be supported (e.g., L3 -> L0 -> L1 -> L2, etc.) and accordingly, the prediction mode of an available sub-block with a previous priority can be selected as the prediction mode in the corresponding direction. When the prediction mode of a sub-block at all positions is not available, the prediction mode can be deselected in the block in the corresponding direction. And, priority to include prediction modes obtained from blocks in each direction in a category can be supported. For example, arrangements such as left -> top -> top-right -> bottom-left -> top-left, left -> top -> bottom-left -> top-right -> top-left, top -> left -> bottom-left -> top-right -> top-left, etc. can be possible and are not limited to it. Many prediction modes fall into the candidate prediction mode group, but there may be many commonly generated prediction modes among them. For example, edges in the vertical and horizontal directions can be common image features, and many flat areas such as backgrounds can be found. Therefore, a predetermined prediction mode that is predicted to be significantly generated can be selected as a candidate belonging to the first or second category. A non-directional prediction mode Planar (No. 0), DC mode (No. 1) and horizontal (No. 18), vertical (No. 50) and diagonal modes (No. 2, No. 34, No. 66, etc.) can be targets for the candidate prediction modes that have been set that are predicted to be generated significantly. Priority for including prediction modes in categories can be supported. For example, arrangements such as Planar -> DC -> Ver -> Hor, etc. are possible and not limited to it. When all or part of the prediction modes of an adjacent block are unavailable, they can be replaced and filled with more than one of the predefined prediction modes (e.g., Planar, DC, etc.) In the above description, the prediction modes of the adjacent blocks of the target block and the pre-set prediction modes are referred to as the prediction modes considered for the category configuration. However, when there are many candidate groups of prediction modes, it can be difficult to effectively predict the prediction mode of the target block with the above configuration. In the example, when the difference is generated by 1, 2 in the direction mode (based on the prediction mode in Figure 5), the prediction may fail with the above configuration. Therefore, a derived prediction mode based on the prediction mode of the adjacent block and a pre-set prediction mode can be considered to configure the category candidate group. In the example, a k-distance prediction mode based on the prediction mode (in this example, the direction mode) can be considered as a derived prediction mode and can be selected as a candidate belonging to the first or second category. In this case, k can be an integer such as 1, 2, 3, 4, or more. In this case, when it is assumed that the prediction mode based on the prediction mode of the adjacent block or the preset prediction mode is Z, there can be priorities for the derived prediction mode such as Z - 1, Z + 1, Z - 2. In this example, it is assumed that the priorities are determined in the order of absolute values ​​such as 1, 2 and (one absolute value) code <-> followed by code <+>, but not limited to it. In other words, it means that the distance information for filling the derived prediction mode should not start from 1 (for example, integers such as 4, 8, 12, etc.) And, the distance k of the derived prediction mode based on the prediction mode of the adjacent block and the preset prediction mode can be identical or non-identical. In addition, when it is assumed that the prediction mode is based on the prediction mode of the adjacent block or the prediction mode that has been set is X, Y, there can be priorities for the derived prediction modes such as X - 1, X + 1, Y - 1, Y + 1 or X - 1, Y - 1, X + 1, Y + 1. In this example, it is assumed that X has priority over Y, and the <-> code and the <+> code alternate with each other or another code (positive number in this example) is derived after one code (negative number in this example) is completely derived (based on one absolute value), but not limited to it. In addition, a number of modes (X, Y in this example) can be used as the base mode for the derived mode or a predetermined mode from it (i.e., any one of X, Y) can be used as the base mode. These can be classified according to predetermined standards and can be determined based on the position of the obtained block (e.g., the mode of a certain block range between the left block and the top block is set as the base mode), the shape of the target block (e.g., the top block mode between the left block and the top block if the horizontally long rectangular shape and the left block mode if the vertically long rectangular shape are set as the base mode. Or, vice versa, etc.) and the tilt information of the predicted mode (e.g., the mode that is more tilted in a certain direction<kiri atau kanan, atas atau bawah> based on vertical or horizontal mode set as base mode). In short, the prediction mode derived based on the preset prediction mode, the prediction mode of the adjacent block and the prediction mode of the precedent can be included in the category (precedent). Each category may include at least one prediction mode from the preset prediction modes, prediction modes from adjacent blocks and derived prediction modes and are referred to as Group_A, Group_B and Group_C, respectively, for ease of explanation. For example, Group_A can be configured in the first category, Group_A, Group_B and Group_C can be configured in the second category and another mode (can be Group_C in this case) can be configured in the third category. Alternatively, Group_B can be configured in the first category, Group_A, Group_B and Group_C can be configured in the second category and other modes can be configured in the third category. Alternatively, Group_A and Group_B can be configured in the first category, Group_A, Group_B and Group_C can be configured in the second category and other modes can be configured in the third category. The example above is an example of mode configuration according to category classification and for Group_A and Group_B included in a number of categories, it is understood that other modes not configured in the precedent category are configured in the next category. In addition, mode configuration with various category classifications can be possible without the limitations of the example above. Next, it is assumed that a number of prediction modes for category configuration are supported and when a mode is configured as many unsubordinated categories (e.g., the first category) according to category priority (e.g., 1-2-3 category order), the remaining modes are configured in subordinated categories (e.g., the second category, the third category, etc.). In the examples mentioned after, it is assumed that the priorities are like Planar (assuming that index No. 0 is set)-> L -> A -> DC -> <L 1> -><L + 1> -> 1> -> <a + 1>-> <L - 2> -> <L + 2> -> -> <a + 2>-> Ver -> Hor -><Ver - 4> -><Ver + 4> -><Hor - 4> -><Hor + 4> , etc. are supported. For example, when the number of prediction modes in the first category is 1, the prediction mode belonging to Group_A can be Planar (index No. 0) and as the number of prediction modes increases, Group_A can be configured with subordinated prediction modes. In addition, when the number of prediction modes in the second category is 1, the prediction mode belonging to Group_B can be from the subsequent prediction mode (index No. m+1) of the prediction mode finally configured in the first category (index No. m) and as the number of prediction modes increases, Group_B can be configured with subordinated prediction modes. Additionally, for the third category, Group_C can be configured with the residual prediction mode of the next prediction mode (index No. n+1) of the prediction mode (index No. n) finally configured in the second category. The number of prediction modes in the first to third categories (in this example, when the candidate group of prediction modes is 67) can be variously set as (1, 5, 61), (2, 4, 61), (4, 8, 55), etc. and can be determined based on the number of candidate groups of prediction modes. In the above example, a preset prediction mode can be a permanently supported mode regardless of the coding environment and a derived prediction mode can be obtained by taking an existing prediction mode in the candidate group as a base mode. On the other hand, a case may occur that it is difficult to obtain the prediction mode of adjacent blocks according to the state of adjacent blocks (e.g., partition unit boundary, whether encoded or not, encoding mode, etc.) in this case, the state of adjacent blocks can be classified into (1) a case where all adjacent blocks are not available, (2) a case where some adjacent blocks are not available and (3) a case where all adjacent blocks are available and the example in priority can be explained by assuming a case where all adjacent blocks are available. In addition, the prediction modes of adjacent blocks can be obtained from two or more blocks and the example in priority can be explained by assuming a case where the prediction modes of each block do not overlap. Additionally, an example in priority can be explained by assuming a case where the prediction modes of the adjacent blocks do not overlap with the set prediction mode. The priority for category configurations can be adaptively set by considering the above elements. For example, an index corresponding to the prediction mode of an adjacent block can be removed from the priority, and an index in a mode derived by taking the prediction mode of the adjacent block as the base mode can also be removed. This can be generally applied to both the case where the prediction mode of the adjacent block is directional and the case where the prediction mode of the adjacent block is non-directional. Additionally, when the prediction modes of adjacent blocks overlap, the index corresponding to the other overlapping prediction mode and the index of the derived mode can be removed. Alternatively, any number of priorities that take such elements into account can be supported and a category can be configured accordingly. In the intra-prediction of the present invention, it is described that multiple supported prediction modes can be configured as a candidate group of prediction modes. This can mean that each separate prediction mode (e.g., direction mode) can be supported according to the target block state information (e.g., block size, shape, etc.) For prediction mode coding, combining them into a common prediction mode candidate group configuration may be preceded by coding. For example, prediction mode candidate group configurations within a predetermined range for prediction mode coding may be combined into a prediction mode candidate group configuration and based on that, prediction mode coding may be performed. In this case, the predetermined range may be configured with prediction modes corresponding to No. 0 to No. 66 in Figure 5, but not limited thereto. In unification, mapping within a predetermined range can be performed and can be performed based on the direction of the prediction mode. In the example, for prediction modes outside the predetermined range, mapping (e.g., if the edge direction is the same, etc. In other words, a case includes where the starting point for prediction and the direction of prediction are different, but the edge direction is the same) can be performed with a prediction mode with similar characteristics to the corresponding prediction mode and various other methods can be supported. The information related to the predictions generated in the process can be transmitted to the encoding unit and stored in a bit stream. In the above explanation, a unit that intra prediction is performed is referred to as a target block. In this case, the target block can be arranged as various unit blocks according to the block settings. In an example, when a coding block is the same unit as a prediction block and a transformation block (i.e., when a coding block is immediately set up as a prediction block or a transformation block), the target block of intra prediction for reference pixel area setting, reference pixel configuration (reference pixel filtering / reference pixel interpolation), prediction mode candidate group configuration, prediction performance, prediction mode coding, etc. can be selected as the coding block. In the example, when a coding block is a unit that can be the same or not the same as a prediction block and a transformation block (i.e., a coding block can be partitioned into two or more sub-blocks<blok prediksi atau blok transformasi> ), the target block of intra prediction can be set as one of encoding block, prediction block and transformation block. For example, the encoding block can be set as the target block for (prediction mode encoding, prediction mode candidate group configuration / reference pixel configuration), etc. and the prediction block or transformation block can be set as the target block for (reference pixel area setting / prediction performance), etc. Alternatively, a coding block can be set as the target block for (prediction mode coding / prediction mode candidate group configuration), etc. and a prediction block or transformation block can be set as the target block for (reference pixel area setting / reference pixel configuration / prediction performance), etc. In short, the target block can be set as one of the encoding block, prediction block and transformation block and the target block unit of detailed configuration can be determined according to the encoding setting. In an image decoding method according to an embodiment of the present invention, intra-prediction can be configured as follows. The intra-prediction in the prediction unit can include decoding a prediction mode, configuring a reference pixel and generating a prediction block. In addition, an image decoding device can be configured to include a prediction mode decoding unit, a reference pixel construction unit and a prediction block generation unit which includes decoding a prediction mode, configuring a reference pixel and generating a prediction block. Some of the above-mentioned processes can be omitted or other processes can be added, and can be changed in another order, not in the order included above. Since the reference pixel construction unit and the prediction block generation unit of the image decoding device play the same role as the corresponding configuration of the image encoding device, the detailed description is omitted and the decoding unit of the prediction mode can be performed by reversibly using the method used in the prediction mode of the encoding unit. Figure 7 is a flowchart showing an intra-prediction modification method according to an embodiment of the present invention. Referring to Figure 7, the intra-prediction mode of the target block can be determined (S700) and the prediction block can be generated (S710) based on the intra-prediction mode. The prediction block modification settings can be determined (S720) and the prediction block modification can be performed (S730). The intra prediction of the target block can be selected in the candidate group of prediction modes obtained based on the target block state information and one prediction mode from it can be determined. A luma component may refer to a prediction mode candidate group configured with directional modes and non-directional modes and a chroma component may refer to a prediction mode candidate group that color modes or color copy modes are supported for directional modes and non-directional modes. In this case, it can be classified into directional mode / non-directional mode / color copy mode according to the prediction method and each method can use extrapolation, interpolation or average, copy, but not limited to it and other modifications can be possible. In the example, according to the reference pixel area setting, interpolation (bidirectional prediction) as well as extrapolation can be supported in directional mode. In summary, according to each prediction method, the reference region for intra prediction can be set differently and for extrapolation, at least one of the blocks in the left, top, top-left, top-right and bottom-left directions of the target block can be set as the reference region and for interpolation, at least one of the blocks in the right, bottom and bottom-right directions of the target block can be set as the reference region in addition to extrapolation. In this case, when the adjacent region is not encoded, it can be derived from one or more pixels in the encoded region and filled. In addition, for copying, the block corresponding to the target block in another color space can be set as the reference region. Based on the data in the reference area and the intra prediction mode, a prediction mode can be generated. Modification of prediction blocks can be done to reduce the discontinuous boundary features with adjacent blocks in the prediction block generated according to the predetermined prediction mode. However, modification according to image features can cause side effects, so it is necessary to make modifications by considering various coding elements. For whether to support modifications for prediction blocks, related information can be explicitly generated or implicitly specified. And, even if it is specified to support modifications in higher units such as sequences, images, etc., the target block modification settings can be set based on the image type, color components, target block state information, encoding mode, intra-prediction mode, whether intra-prediction in sub-block units is applied, reference pixel line selection information, etc. In addition, modification-related setting information (e.g., flags about whether modifications in block units are performed, weight information applied for modifications, etc.) can be explicitly generated in lower units (e.g., for previous information, blocks, sub-blocks, etc.) or higher units (e.g., for last information, sequences, images, slices, tiles, etc.) The modification settings can be configured by including whether the modification is performed, the number, position, weight, etc. of the reference pixels used for the modification, etc. Even though the modification is performed, the modification can be performed for all or some of the pixels in the target block. For example, when the target block is mxn, the modification can be performed for 1 to (mxn) pixels. In this case, whether modification (in pixel units) is applied, (when modification is performed) the number (1~5), position, weight, etc. of the reference pixels used for modification, etc. can be set based on the pixel position in the block as well as the modification settings. In this case, a modification-supported intra-prediction mode can be determined according to whether it belongs to a predetermined prediction mode group. For example, such a prediction mode group can be configured among Planar, DC, horizontal, vertical and diagonal modes (e.g., top-right direction mode, bottom-right direction mode, bottom-left direction mode, etc. modes No. 2, 34, 66 in Figure 5) and color copy mode. Alternatively, such a mode can be configured by additionally including prediction modes derived from the prediction mode group (e.g., a k-spacing mode based on a direction mode. In this case, k is an integer whose absolute value is 1, 2, 3, 4 or more.) A target block that supports modification can be smaller than or equal to a predetermined first threshold size. In this case, the first threshold size can be the maximum size the target block can have. Additionally, a block supported by modifications can be larger than or equal to a second predetermined threshold size. In this case, the second threshold size can be the minimum size the target block can have. In this case, the first threshold size and the second threshold size can be expressed as width (W), height (H), W x H, W*H, etc. and W and H can be integers such as 4, 8, 16, 32 or more. In this case, the first threshold size can be greater than or equal to the second threshold size. Figure 8 is a diagram of the arrangement at pixels of the target block and adjacent blocks according to an embodiment of the present invention. Referring to Figure 8, it consists of pixels (a ~ p) in the target block, pixels (A ~ Q) in the encoded adjacent block and pixels (aa ~ ii) in the unencoded adjacent block. In this case, modifications can be made to pixels in the target block and pixels in the adjacent block can be referenced for modification. In this case, the reference pixel in the adjacent block can be one of the values ​​obtained after the reference pixel preprocessing process (e.g., reference pixel filtering) is not performed or is performed. This can be specified in block units or pixel units. The following represents the equation that the modified pixels are obtained after the modification is applied to the prediction pixels. Z = (z * w0) + (a * w1) + (b * w2) + (c * w3) In this case, z and Z mean the predicted pixels and modified pixels, respectively, a to c mean the reference pixels used for modification and w0 to w3 mean the weight applied for modification. According to the number of reference pixels used for modification, the reference pixels and weight can be determined. In this case, a value that considers normalization with including 0 can be assigned to the weight. According to the target block prediction mode, the position and number of reference pixels can be determined. For example, when the target block prediction mode is Planar, DC mode, vertical, horizontal, a, b, c can be set as the top-left reference pixel outside the target block (e.g., <-1,1>), the reference pixel above the target block that responds to or corresponds to the x component of the target pixel (e.g.,<x, -1> ) and the reference pixel to the left of the target block that responds to or corresponds to the y component of the target pixel (e.g., <-1, y>). In this case, for Planar, DC mode, weights can be assigned to w2 and w3 to reflect the gradient of the reference pixels above and to the left of the target block. This can be an example (e.g., w1 is 0) of setting the weights based on how much they change from the reference pixel corresponding to the x and y components of the target pixel. In this case, for the vertical mode, the negatively coded weight can be set to w0 and the positively coded weight (i.e., the reverse code) can be set to w3 to reflect the gradient of the reference pixel corresponding to the prediction direction. It can be an example (e.g., w2 is 0) to set the weight based on how much the reference pixel corresponding to the y component of the target pixel is changed from the top-left reference pixel. For the horizontal mode, the corresponding explanation can be derived in the vertical mode, so the detailed explanation is omitted. Alternatively, when the prediction mode of the target block is diagonal mode, a, b, c can be set as the top-left reference pixels outside the target block, the reference pixels (e.g.,<x+y+1, -1> ) that responds to or corresponds to the starting point in the target block prediction direction and a reference pixel (e.g., <-1, x+y+1>) that responds to or corresponds to the side opposite to the starting point in the target block prediction direction. In this case, when such reference pixels are not obtained in integer units (i.e., when interpolation is required in decimal units), they can be set by either a method that obtains them by being replaced by one adjacent reference pixel or a method that obtains them by interpolating through adjacent reference pixels on both sides. In this case, for the diagonal mode in the top-right direction, a weight can be assigned to w3 to reflect the gradient of the reference pixel positioned at the starting point of the prediction direction and positioned on the opposite side. This can be an example (e.g., w1 and w2 are 0) in setting the weight based on how much it is changed from the reference pixel in the prediction direction of the target pixel. For the diagonal mode in the bottom-left direction, the last explanation can be lowered in the mode, so the detailed explanation is omitted. Figures 9A and 9B are example diagrams of a modified method based on a compound reference pixel line according to an embodiment of the present invention. The examples mentioned after can be configurations that can be supported separately from the configuration that the reference pixel lines to be used for intra-prediction are selected in the multiple reference pixel lines mentioned above and will be explained based on them, but can also be applied to inter-combinations. Referring to Figure 9A represents an example of modification in horizontal or vertical mode. In detail, it can be an example of the case where reference pixels in two or more reference pixel lines are used for modification. When the target block prediction mode is vertical, the modification accuracy can be improved by adding a* as a reference pixel. Alternatively, a** can be obtained by applying a weighted average to a and a* to apply them to the modification equation mentioned above. Since block partitioning may not be performed to have boundaries that accurately divide image features, it can be for the same reason as using reference pixels adjacent to the target block as prediction pixels. When the target block prediction mode is horizontal mode, the modification accuracy can be improved by adding b* setting as the reference pixel for b and b** can be obtained by applying the average weight on b and b* to apply them to the modification equation mentioned above. Referring to Figure 9B represents an example of modification in diagonal mode. In detail, it can be an example of the case where a reference pixel in two or more reference pixel lines is used for modification. When the target block prediction mode is the bottom-left direction, the modification accuracy can be improved by adding a* setting as a reference pixel for a at the prediction starting point or a** can be obtained by applying an average weight on a and a* to apply it to the modification equation mentioned above. In addition, the modification accuracy can be improved by adding b* setting as a reference pixel for b at the opposite position to the prediction starting point or b** can be obtained by applying an average weight to b and b* to apply them to the modification equation mentioned above. For diagonal mode, a reference pixel is obtained based on one or more reference pixels at the prediction starting point or at a position opposite the starting point. In this case, the reference pixel can be selected for the pixel in the prediction mode direction. In the above example, when the reference pixel on the compound reference pixel line is used for modification, whether to support can be determined based on the image type, color component, target block state information, intra prediction mode, whether the intra prediction mode in the sub-block unit is applied, reference pixel line selection information, etc. Figure 10 is a flowchart showing an intra-prediction modification method according to an embodiment of the present invention. Referring to Figure 10, an arbitrary pixel for target block modification can be obtained (S900), modification settings can be determined based on the target block and intra prediction mode (S910) and prediction block modification can be performed (S920). For prediction block modification, blocks adjacent to the target block can be configured to be tilted in a certain direction (e.g., in a top-left direction), which is generated by a coding sequence according to a raster scan or Z-scan method. In addition to predicting and modifying reference pixels tilted in a specific direction, prediction accuracy can be improved by using data from regions where encoding has not yet been completed. To do this, a process can be used to obtain arbitrary pixels. In this case, the number of arbitrary pixels can be an integer such as 1, 2, or more. For example, an arbitrary pixel used for modification of the target block can be specified between pixels aa to ii in Figure 8. In detail, a pixel that belongs to an area where coding has not been completed between adjacent blocks and the target block can be set as an arbitrary pixel. Alternatively, it can be defined between pixels a to p in Figure 8. In detail, as the target block also belongs to the unencoded pixels, it can be included in the target selected as an arbitrary pixel. Figures 11A, 11B, 11C, 11D and 11E are examples of diagrams at arbitrary pixels used to modify the prediction pixels according to embodiments of the present invention. Referring to Figures 11A, 11B, 11C, 11D and 11E, example diagrams are shown where for Figure 11A, the bottom-right pixel is in the target block, for Figure 11B, the bottom-right pixel is outside the target block, for Figure 11C, the bottom and left pixels are outside the target block, for Figure 11D, the bottom pixel is outside the target block and for Figure 11E, the right pixel outside the target block is defined as an arbitrary pixel (X, X1, X2). For whether modifications are performed using arbitrary pixels, related information can be explicitly generated or implicitly specified. And, even if it is specified to use arbitrary pixels in higher units such as sequences, images, etc., settings for arbitrary pixels can be set based on the image type, color components, target block state information, encoding mode, intra-prediction mode, whether intra-prediction in sub-block units is applied, reference pixel line selection information, etc. In addition, setting information related to arbitrary pixels (e.g., flags about whether modifications are performed using arbitrary pixels in unit blocks, etc.) can be explicitly generated. Settings on arbitrary pixels can be configured including whether arbitrary pixels are used (i.e., whether arbitrary pixels are used for modification), the number, position, etc. of arbitrary pixels, etc. In this case, the intra-prediction mode that performs modifications using arbitrary pixels can be determined according to whether it belongs to a predetermined prediction mode group. For example, between Planar, DC, horizontal, vertical, diagonal modes (e.g., top-right direction mode, bottom-right direction mode, bottom-left direction mode, etc. modes No. 2, 34, 66 in Figure 5) and color copy mode, the prediction group can be configured. Alternatively, it can be configured by additionally including prediction modes (e.g., k-spacing mode based on the direction mode. In this case, k is an integer whose absolute value is 1, 2, 3, 4, or more) derived in the prediction mode group. A target block that modifications are made to using arbitrary pixels can be smaller than or equal to a predetermined first threshold size and the first threshold size can be the maximum size that the target block can have. Alternatively, it can be greater than or equal to a predetermined second threshold size and the second threshold size can be the minimum size that the target block can have. In this case, the threshold size can be indicated as width (W), height (H), W x H, W*H, etc., and W and H can be integers such as 4, 8, 16, 32 or more and the first threshold size can be greater than or equal to the second threshold size. In the above explanation, it is understood that the modification arrangement mentioned through the previous embodiment includes or is combined with the lower configuration, so that the overlapping explanation is omitted. In the examples mentioned afterward, it is assumed that the arbitrary pixels are arranged as Figure 11A. Since an arbitrary pixel position is an unencoded region, the pixel value at the corresponding position can be obtained by various methods. For example, data at arbitrary pixel positions can be explicitly encoded. The pixel value at the corresponding position can be encoded as is (e.g., based on bit depth) or a value obtained by a predetermined division value (e.g., quantization. The dequantized value is used for modification) can be encoded. Alternatively, two or more candidate data values ​​at arbitrary pixel positions can be obtained in an encoded region adjacent to the target block and can be encoded by configuring their index information. In the example, when a total of 2 candidates are supported as pixel values ​​at arbitrary pixel positions is 67, the first candidate value obtained in the region adjacent to the target block (e.g., the top region) is 70 and the second candidate value obtained in the region adjacent to the target block (e.g., the left region) is 85, the candidate value selection information (e.g., the first candidate value is selected. 1-bit flag) can be encoded. In this case, obtaining candidate values ​​can be mentioned in the examples mentioned afterward. Such candidate values ​​can be supported as many as an integer number such as 2, 3, 4 or more. For the above example, it could be a case where the information about the data at an arbitrary pixel position is explicitly encoded and the data at an arbitrary pixel position can be implicitly obtained. For example, it can be obtained from reference pixels at predetermined positions or reference pixels that respond or correspond to arbitrary pixel positions. For example, for predetermined positions, it can be obtained from reference pixels (A, E, M in Figure 8) at the top-left, top and left positions outside the target block and reference pixels at positions (top-right, bottom-left) based on the width or height of the target block (I, Q in Figure 8) can be obtained at those positions. Additionally, for example for a reference pixel corresponding to an arbitrary pixel position, it can be obtained from the reference pixels corresponding to the x or y components of the arbitrary pixel (e.g.,<x, -1> , <-1, y>,<x+y+1, -1> , <-1, x+y+1>, etc.) The data value of one reference pixel among the reference pixels at a position can be obtained simply as the data value at an arbitrary pixel position or the value derived based on two or more reference pixels can be obtained as the data value at an arbitrary pixel position. For example, a reference pixel at a predetermined position among a plurality of reference pixels used for arbitrary pixel positions can be obtained as a data value at the arbitrary pixel position. Alternatively, a value obtained in a predetermined process (e.g., maximum value, minimum value, middle value, etc.) among a plurality of reference pixels can be obtained as a data value at the arbitrary pixel position. Alternatively, a value obtained in a predetermined process (e.g., weighted average, etc.) among a plurality of reference pixels can be obtained as a data value at the arbitrary pixel position. The following describes a method of obtaining data values ​​at arbitrary pixel positions based on a number of reference pixels. It is assumed that A, E, M in Figure 8 are used as reference pixels and in the examples mentioned afterward, they are referred to as x, y, z. [Table 2] Candidate 0 1 2 3 4 5 6 Pixel value xyz y+zx z+(yx) / 2 y+(zx) / 2 (y+z) / 2 Candidates No. 0 to 2 in the table mean a case where a reference pixel at a predetermined position is obtained as a data value at an arbitrary pixel position. Candidates No. 3 to 6 mean a case of obtaining data at an arbitrary pixel position by reflecting the gradient of the target block. Specifically, it can correspond to a case where the gradient of the target block is obtained based on the reference pixel at a predetermined position. It should be understood that the example can be an example for an equation derived from a number of pixels assuming that the target block has a square shape and can be transformed and applied according to the block's aspect ratio. In addition, data values ​​at arbitrary pixel positions can be obtained in various ways without being limited to the example. Figures 12A, 12B, 12C, 12D, 12E and 12F are examples of diagrams where modifications are made based on arbitrary pixels according to embodiments of the present invention. Referring to Figure 12A represents an interpolation process with arbitrary pixels and reference pixels corresponding to the upper block and represents a process where modifications are made using reference pixels in the left, right, upper, upper-left and upper-right directions of the target block obtained therefrom. This process can be applied when the modification is performed by reflecting the gradient of the reference pixels in the left and right directions when the target block prediction mode is vertical. This may differ from the case when the modification is performed based on the gradient of the reference pixel in the left direction. Referring to Figure 12C represents the interpolation process with arbitrary pixels and reference pixels corresponding to the left block and represents a process where modifications are made using the reference pixels in the top, bottom, left, top-left and bottom-left directions of the target block obtained therefrom. This process can be applied when the modification is performed by reflecting the gradient of the reference pixel in the up and down directions when the target block prediction mode is horizontal. This may differ from the case when the modification is performed based on the gradient of the reference pixel in the up direction. Referring to Figure 12E represents an interpolation process with arbitrary pixels and reference pixels corresponding to the left and top blocks and represents a process where modifications are made using reference pixels in the top, bottom, left, right, top-left, top-right, bottom-left and bottom-right directions of the target block obtained therefrom. The following represents the equation that the modified pixel is obtained by applying modifications to the predicted pixel. In this case, the reference pixel can be configured differently from the previous equation that an arbitrary pixel or a reference pixel obtained based on an arbitrary pixel is included. Z = (z * w0) + (a * w1) + (b * w2) + (c * w3) + (d * w4) + (e * w5) In this case, Z and Z denote the predicted pixels and modified pixels, respectively, a to e denote the reference pixels used for modification and w0 to w5 denote the weight applied for modification. According to the number of reference pixels used for modification, the reference pixels and weight can be determined. In this case, a value that considers normalization with including 0 can be assigned to the weight. According to the target block prediction mode, the position and number of reference pixels can be determined and various cases such as Figures 12A, 12B, 12C, 12D, 12E and 12F can be possible. Since other explanations about the modification settings can be derived in the section regarding the previous equations, detailed explanations are omitted. A case where modifications are made in the intra-prediction process is described through various embodiments. In this case, it is described only if the modifications are made after obtaining the prediction block, but it may be possible to make modifications after reflection on the intra-prediction process (i.e., the process for generating the prediction block). In addition, the flag information generated in the modification process can be encoded / decoded in a configuration separate from the intra-prediction mode, but a configuration combined with or dependent on the intra-prediction mode is possible. In other words, it should be understood that it is not limited to additional or subsequent sections after the intra-prediction process. The method according to the present invention can be recorded in a computer-readable medium after being embodied in the form of program instructions that can be executed by various computer means. A computer-readable medium can include program instructions, data files, data structures, etc. alone or in combination. A program instruction recorded in a computer-readable medium can be especially designed and configured for the present invention or can be made available after being notified to a person skilled in computer software. An example of a computer-readable medium may include hardware devices such as ROM, RAM, Flash memory, etc., particularly those configured to store and execute program instructions. Examples of program instructions may include high-level language code that can be executed by a computer with an interpreter, etc., in addition to machine language code that can be generated by a compiler. The aforementioned hardware devices may be configured to operate as at least one software module to perform the operations of the present invention, and vice versa. In addition, the above-mentioned methods or tools can be realized after all or part of the configurations or functions are combined or separated. While the above is described with reference to preferred embodiments of the present invention, one skilled in the relevant field of technology will understand that the present invention may be variously modified and altered in a range not extending beyond the ideas and areas of the present invention included in the following claims. Industrial Application This invention can be used to encode / decode video signals. ​< / intra> < / qt> < / tinggi> < / lebar> < / tengah>

Claims

1. A method for decoding a video signal with a decoding apparatus, including: determining an intra-prediction mode of a current block in the video signal; constructing a reference sample of the current block; and generating a prediction block of the current block based on the intra-prediction mode and the constructed reference sample, wherein the constructed reference sample includes a left reference sample and a right reference sample, wherein the left reference sample belongs to a reconstructed block adjacent to a left portion of the current block and adjacent to a left boundary of the current block, and the right reference sample belongs to a reconstructed block adjacent to a right portion of the current block and adjacent to a right boundary of the current block,wherein the generation of a prediction block of the current block includes generating a prediction sample of the current sample in the current block using a specific number of reference samples among the reference samples constructed when the intra prediction mode is a non-directional mode, wherein the specific number is determined as 5, wherein the specific number of reference samples includes a top reference sample, two left reference samples, and one right reference sample, wherein the two left reference samples include a bottom reference sample among the left reference samples, and wherein the position of the bottom reference sample is (-1,H-1) relative to the top left sample of the current block, H refers to the height of the current block., 2. The method of claim 1, wherein constructing a reference sample of the current block includes determining the availability of the reference sample.

3. The method of claim 2, wherein the at least one reference sample is determined to be unavailable when the at least one reference sample is located outside the image boundary, the at least one reference sample is not contained in the same tile as the current block, or the decoding of the at least one reference sample is not complete.

4. The method of claim 3, wherein when the limited intra prediction is used for the current block, the availability of the left reference sample is determined based on the encoding mode of the reference sample candidate block adjacent to the left portion of the current block, and the availability of the right reference sample is determined based on the encoding mode of the reference sample candidate block adjacent to the right portion of the current block.

5. The method of claim 1, wherein the specific number of reference samples includes a first reference sample determined based on the position of the current sample within the current block and a second reference sample determined based on the width or height of the current block.

6. The method of claim 5, wherein the predicted sample of the current sample is generated using the arbitrary sample and the first reference sample is determined based on the position of the current sample among a specific number of reference samples, and wherein the arbitrary sample is obtained by a weighted sum of the second reference sample among the specific number of reference samples.

7. The method of claim 6, wherein the predicted sample of the current sample is generated by applying weights to each of the arbitrary samples and the first reference sample.

8. The method of claim 7, the weight being determined based on the position of the current sample within the current block.

9. A method of encoding a video signal with a coding apparatus, including: determining an intra-prediction mode of a current block in a video signal; constructing a reference sample of the current block; and generating a residual block of the current block using the prediction block, the prediction block being generated based on the intra-prediction mode and the reference sample; wherein the constructed reference sample includes a left reference sample and a right reference sample, wherein the left reference sample belongs to a reconstructed block adjacent to a left portion of the current block and adjacent to a left boundary of the current block, and the right reference sample belongs to a reconstructed block adjacent to a right portion of the current block and adjacent to a right boundary of the current block,wherein the generation of a prediction block of the current block includes generating a prediction sample of the current sample in the current block using a specific number of reference samples among the reference samples constructed when the intra prediction mode is a non-directional mode, wherein the specific number is determined as 5, wherein the specific number of reference samples includes a top reference sample, two left reference samples, and one right reference sample, wherein the two left reference samples include a bottom reference sample among the left reference samples, and wherein the position of the bottom reference sample is (-1,H-1) relative to the top left sample of the current block, H refers to the height of the current block., 10. A non-transient computer-readable storage medium that stores a bit stream generated by a method of encoding a video signal with an encoding apparatus, the method comprising: determining an intra-prediction mode of a current block in the video signal; constructing a reference sample from the current block; and generating a residual block from the current block using the prediction block, the prediction block being generated based on the intra-prediction mode and the reference sample; wherein the constructed reference sample includes a left reference sample and a right reference sample, wherein the left reference sample belongs to a reconstructed block adjacent to a left portion of the current block and adjacent to a left boundary of the current block, and the right reference sample belongs to a reconstructed block adjacent to a right portion of the current block and adjacent to a right boundary of the current block,wherein the generation of a prediction block of the current block includes generating a prediction sample of the current sample in the current block using a specific number of reference samples among the reference samples constructed when the intra prediction mode is a non-directional mode, wherein the specific number is determined as 5, wherein the specific number of reference samples includes a top reference sample, two left reference samples, and one right reference sample, wherein the two left reference samples include a bottom reference sample among the left reference samples, and wherein the position of the bottom reference sample is (-1,H-1) relative to the top left sample of the current block, H refers to the height of the current block., 11. A method for decoding a video signal with a decoding apparatus, comprising: determining an intra-prediction mode applied to the intra-prediction for the current block; and generating a prediction block from the current block based on the intra-prediction mode; wherein generating the prediction block from the current block includes: configuring reference samples for the intra-prediction for the current block; obtaining a value from a first prediction sample in the prediction block from the current block based on at least one of the configured reference samples; and obtaining a value from a second prediction sample in the prediction block based on the first prediction sample in the prediction block and at least one of the configured reference samples.

12. The method of claim 11, wherein the acquisition of the value of the second prediction sample is based on an interpolation using the first prediction sample and at least one of the configured reference samples.

13. The method of claim 11, wherein the obtained value of the first predicted sample includes the average of two reference samples that are adjacent to each other.

14. A method for encoding a video signal with an encoding apparatus, comprising: determining an intra-prediction mode applied to the intra-prediction for the current block; and generating a prediction block from the current block based on the intra-prediction mode; wherein generating the prediction block from the current block includes: configuring reference samples for the intra-prediction for the current block; obtaining a value from a first prediction sample in the prediction block from the current block based on at least one of the configured reference samples; and obtaining a value from a second prediction sample in the prediction block based on the first prediction sample in the prediction block and at least one of the configured reference samples.

15. A non-transient computer-readable recording medium storing a bit stream generated by a method of encoding a video signal with an encoding apparatus, the method comprising: determining an intra-prediction mode applied to the intra-prediction for the current block; and generating a prediction block from the current block based on the intra-prediction mode; wherein generating the prediction block from the current block includes: configuring reference samples for the intra-prediction for the current block; obtaining a value from a first prediction sample within the prediction block from the current block based on at least one of the configured reference samples; and obtaining a value from a second prediction sample within the prediction block based on the first prediction sample within the prediction block and at least one of the configured reference samples.

16. A method of decoding an image with a decoding apparatus comprising: determining an intra-prediction mode of a target block; generating a prediction block of the target block based on the intra-prediction mode; and modifying the prediction block, wherein the intra-prediction mode of the target block is determined as a mode in a candidate group of prediction modes according to the state information of the target block.

17. The method of claim 16, wherein when the color component of said target block is a luma component, a prediction mode candidate group comprising a directional mode and a non-directional mode is referenced, wherein when the color component of said target block is a chroma component, a prediction mode candidate group in which at least one of a directional mode, a non-directional mode, a color mode or a color copy mode is supported is referenced.

18. The method of claim 17, wherein said group of candidate prediction modes is classified into a plurality of categories by considering a maximum number or priority of prediction modes capable of being included in each category.

19. The method of claim 17, wherein said candidate group of prediction modes is classified into a first category including non-directional modes and directional modes and a second category including color copy modes.

20. The method of claim 18, further comprising: obtaining first information that specifies one of a plurality of categories; and obtaining second information that specifies an intra-prediction mode of a target block in a category according to the first information, wherein the intra-prediction mode of said target block is determined from a category determined based on the first information and the second information.

21. The method of claim 20, wherein said second information is not obtained when only one prediction mode is included in the category according to the first information.

22. The method of claim 16, further comprising, constructing a reference pixel used for intra prediction, wherein said reference pixel includes all or part of a plurality of reference pixel lines supported in the decoding apparatus.

23. The method of claim 22, further comprising, performing at least one of a weighted filter or an interpolation filter for the constructed reference pixel.

24. The method of claim 16, wherein said prediction block modification is selectively performed based on predetermined coding information, wherein said coding information includes at least one of an image type, a color component, state information, a coding mode, an intra-prediction mode, whether intra-prediction in a sub-block unit is applied or a reference pixel line.

25. A method of encoding an image with an encoding apparatus comprising: determining an intra-prediction mode of a target block; generating a prediction block of the target block based on the intra-prediction mode; and modifying the prediction block, wherein the intra-prediction mode of the target block is determined as a mode in a candidate group of prediction modes according to the state information of the target block.

26. A non-transient computer-readable recording medium that stores a bit stream generated from an image encoding method, the method comprising: determining an intra-prediction mode of a target block; generating a prediction block from the target block based on the intra-prediction mode; and modifying the prediction block, wherein the intra-prediction mode of the target block is determined as a mode in a candidate group of prediction modes according to the state information of the target block.

Images