77 results about "Residual coding" patented technology

A multi-view video coding method and apparatus and a multi-view video decoding method and apparatus for providing a multi-view video service are provided. The multi-view video coding method includes: coding a base layer picture using an arbitrary video codec; generating a prediction picture using at least one of a reconstructed base layer picture and a reconstructed layer picture having a different view from that of the base layer picture; and residual-coding a layer picture having the different view using the prediction picture.

A method for coding video is disclosed. The method generally includes the steps of (A) receiving a video signal having a series of pictures, each of the pictures having a plurality of blocks, (B) analyzing the blocks to forecast if coding the blocks in a zero-residual coding mode would generate a plurality of artifacts, (C) disabling the zero-residual coding mode for the blocks forecasted to generate at least one of the artifacts and (D) enabling the zero-residual coding mode for the blocks forecasted to generate none of the artifacts.

A transform method, according to the present invention, comprises the steps of: obtaining transform coefficients for a target block; determining a non-separable secondary transform (NSST) set for the target block; selecting one of a plurality of NSST kernels included in the NSST set on the basis of an NSST index; and generating modified transform coefficients by non-separable secondary-transform of the transform coefficients on the basis of the NSST kernel that has been selected, wherein the NSST set for the target block is determined on the basis of an intra-prediction mode and/or the size of the target block. According to the present invention, the amount of data transmitted, which is required for residual processing, can be reduced and residual coding efficiency can be increased.

The invention discloses a JPEG-LS run length coding hardware implementation method which improves and optimizes the standard procedure of JPEG-LS run length coding, introduces coding mapping operation, implements coding by comparing a table look-up operation pipeline, and solves cycle coding in multiple clock cycles in an original algorithm; in addition, since only residual coding adopts index value RUN index single-cycle update operation and tour length coding and residual coding adopt index value RUNindex two-cycle update operation, coding speed is improved. The method is completely implemented by FPGA, has the advantages of full pipeline and good real time property, and can be applied to hardware algorithm of JPEG-LS lossless and near lossless compression.

The invention discloses a medical image noise reduction method based on a generative adversarial network and 3D residual coding and decoding, and the method comprises the steps: carrying out the classification collection of training data, carrying out the preprocessing of the training data, wherein the training data comprises a low-quality image and a high-quality image; constructing a convolutional neural network based on a generative adversarial network and 3D residual encoding and decoding; taking a low-quality image with scanning time of 75s and size of N * 9 * 64 * 64 * 1 as training input, taking a high-quality image with scanning time of 150s and size of N * 9 * 64 * 64 * 1 as a training label, and training the network; and carrying out noise reduction on the high-noise image by using the trained convolutional neural network to obtain a high-quality image. By applying the technical scheme of the invention, any high-noise positron emission computer tomography image can be accurately and rapidly denoised after the model is trained by using a small amount of data.

The invention relates to the image selection encryption technical field, and provides an adaptive selection encryption method based on salient areas; the method comprises the following steps: carrying out saliency detection for an image and tagging salient areas; forming a chaotic sequence P; converting a plaintext image color space to YCbCr, and carrying out partitioning diffusion scrambling on the image; executing a DCT conversion and quantification process; forming a DC coefficient matrix MDC and an AC coefficient matrix MAC; executing adaptive encryption for the MDC and MAC; respectively carrying out component encryption for Y, Cb, and Cr; finishing residual coding steps. The method uses saliency detection to determine image salient areas, and selects salient areas for emphasized encryption; compared with edge detection, the novel encryption mode can better ensure image important area security, thus improving image anti-attaching property.

An image decoding method according to the present invention comprises the steps of: deriving, from a bitstream, quantized transform coefficients for a block to be decoded; deriving transform coefficients through dequantization on the basis of the quantized transform coefficients for the block to be decoded; determining a transform configuration group for the block to be decoded; deriving a horizontal transform kernel and a vertical transform kernel on the basis of the transform configuration group and transform index information acquired from the bitstream; deriving residual samples by performing a separable transform on the basis of the horizontal transform kernel and the vertical transform kernel; and generating a reconstructed picture on the basis of the residual samples and prediction samples of the block to be decoded, wherein the transform configuration group includes an identity transform kernel as a candidate transform kernel for the horizontal transform kernel or the vertical transform kernel. According to the present invention, the efficiency of residual coding can be improved.

In transform coefficient coding, predetermined sets of transforms may be used. Embodiments of the present invention provide an encoder and a decoder for encoding a picture into/from a data stream using block-based prediction and block-based residual coding, the encoder and decoder supporting a set of transforms for the block-based residual coding. The encoder and decoder are configured to determine a transform candidate list of transforms for a current residual block out of the set of transforms based on a height and a width of the current residual block, and to encode/decode a prediction residual within the residual block using one selected transform out of the selectable list of transforms and selectively signal/derive the selected transform in/from the data stream.

A method for video processing includes applying a sub-block residual coding scheme to a current video block, wherein the residual of the current video block is split into a plurality of sub-blocks each having a size of M*N, determining at least one sub-block with non-zero coefficients based on characteristic associated with each of the plurality of sub-blocks; coding the residue of the current video block based on the derived at least one sub-block with non-zero coefficients.

The invention discloses a code rate control method in video coding. The method comprises the following steps: dividing an input video signal into serial video frame images; updating the number of remaining encoding bits of the current encoding frame; calculating an MAD value of a current basic unit according to a time domain linear prediction model; if MADcb is greater than TH, performing airspacecorrection; if MADcb is less than TH, not performing the airspace correction; calculating the number of coding bits used by a current macro block; calculating and correcting a quantization parameterQP value of the current macro block; executing rate distortion optimization to obtain an actual MAD value of the current macro block; performing circular execution until the macro blocks in the current basic unit are completely processed; and calculating the next basic unit, and performing circular execution until the basic units in the video frames are completely processed. The movement conditions of the current processing unit on time and space can be well predicted to accurately predict and process the video coding quantization parameter.

The invention discloses an image lossless compression method implemented on basis of an FPGA, and belongs to the technical field of image processing of machine vision. The image lossless compression method comprises the steps of: initializing coding parameters into zero, wherein the coding parameters comprise data of a predicted row, an accumulated predicted residual and context combination accumulated occurrence times; selecting two reference pixel points to calculate a residual of a current pixel point according to pixel values of the two reference pixel points; according to a relationship between a pixel value P of the current pixel point and the pixel values of the two reference pixel points, selecting a corresponding residual coding method to carry out coding on the residual of the current pixel point; and after completing residual coding of all the pixel points of a to-be-compressed image, updating the coding parameters. According to the invention, improvement and optimization are carried out on the basis of an FELICS (Fast and Efficient Lossless Image Compression System) algorithm, a decompression speed of an upper computer is improved when an image compression ratio is ensured, and an algorithm running speed can be matched with a gigabit Ethernet transmission speed.

The invention discloses a JPEG image compression artifact elimination algorithm based on a cascade residual error coding and decoding network CRED-Net, which comprises the following steps of: firstly,providing QF value estimation of a compressed image by using a quality factor (QF) prediction network, and then removing an appropriate cascade of artifacts of the compressed image by selecting an appropriate QF value for the estimated QF value. According to the method provided by the invention, multi-scale feature learning is carried out in a pixel domain, a discrete cosine transform (DCT) domain and a discrete wavelet transform (DWT) domain without depending on compressed encoding information of the JPEG image, and when the QF value of the image is not within the consideration range of a trained CREDNet model, a correction network is introduced to further correct the restored image. Experimental results prove that compared with other methods. The method provided by the invention can obtain better performance in a wider compression level range.

The invention discloses a TU tree structure decision algorithm based on HEVC video coding, comprising the following steps of (a) transforming TU residuals to obtain a coefficient C0 before quantization; (b) quantifying C0 obtained in step (a) to obtain a coefficient C1; (c) using a fixed binary Bin and a bits compression ratio to obtain an estimate value of the number of bits; (d) dequantizing C1 obtained in step (b) to obtain a coefficient C2; (e) calculating a distortion value of the TU; (f) calculating the rate distortion cost after the TU residual coding; (g) calculating the rate distortion cost of the TU residual without any coding; and (h) comparing the rate distortion cost obtained in the step (g) with the rate distortion cost obtained in the step (f), and taking the rate distortion cost of a smaller value as the best price of this TU. The invention adopts the above algorithm, which can reduce the complexity of the calculation, reduce the hardware overhead, and improve the hardware performance.

Provided are an intraframe coding method, device and encoder for depth images. The method is applied to 3D video coding or multi-view video coding. The method comprises: in an intraframe coding mode for the depth images, generating a non-all-zero residual and an all-zero residual when a prediction unit is coded; and selecting rate-distortion optimization between the non-all-zero residual and the all-zero residual so as to be used as the residual coding scheme of the prediction unit. The device is applied to 3D video coding or multi-view video coding. The device comprises: a residual generation unit used in an intraframe coding mode for depth images, for generating a non-all-zero residual and an all-zero residual when a prediction unit is coded; and a rate-distortion optimization selecting unit used for selecting rate-distortion optimization between the non-all-zero residual and the all-zero residual so as to be used as the residual coding scheme of the prediction unit. By adopting the method, device and encoder provided by the invention, the code rate required for coding depth images is reduced, and the coding performance is increased.

An image decoding method performed by a decoding device according to the present document is characterized by including the steps of: receiving image information including a residual coding flag indicating whether a transform skip residual coding syntax structure is available for the current slice; determining whether the transform skip residual coding syntax structure is available for the current block in the current slice on the basis of the residual coding flag; parsing residual information for the current block on the basis of the results of the determination to derive residual samples of the current block; and generating a reconstructed picture on the basis of the residual samples.