45results about How to "Improve subjective image quality" patented technology

An image sensor for electronic cameras has an image field having at least one image field block which includes a plurality of light sensitive pixels arranged in rows and columns for generating exposure dependent pixel signals, wherein the pixel signals of the pixels of the respective column of the respective image field block can be read out via one or more respective column lines extending parallel to one another. The image sensor has at least one first row of column amplifiers and at least one second row of column amplifiers for the respective image field block. The image sensor is adapted to read out the pixel signals of the pixels of the respective column of the respective image field block of an image or of two mutually following images partly via the at least one first row and partly via the at least one second row of column amplifiers (25).

In an image encoder, a first quantization section which is selected when normal image quality is required performs quantization by dividing a wavelet transformation coefficient by a quantization step size and by thereafter rounding down a fraction thereof. On the other side, a second quantization section which is selected when high image quality is required performs quantization by dividing a wavelet transformation coeffient by a quantization step size, by adding 0.5 to the dividing result, and by thereafter rounding down a fraction thereof. Therefore, the width of a dead zone where coefficients are quantized to a value of 0 is narrower than that in the first quantization section, and higher image quality is obtained accordingly.

Disclosed is a method and apparatus for generating a quantization matrix for encoding an image or image sequence, wherein the effect of greatly reducing the data rate in video encoding is achieved by quantizing the transformed frequency coefficients or components of a pixel block, so that only Fewer amplitude levels are encoded, and a portion of quantized amplitudes becomes 0 and does not need to be encoded as quantized amplitudes. The quantization matrix generation of the present invention starts with a default quantization matrix and thus results in the perceptually optimal quantization matrix for a given sequence of images. In the first pass, candidate quantization matrices are iteratively constructed for a given sequence of images by simultaneously increasing the scaling values ​​for some coefficient positions and decreasing the scaling values ​​for some other coefficient positions. In the second pass, the resulting quantization matrix is ​​applied to the image sequence for re-encoding.

There is provided an information processing apparatus and method that makes it possible to suppress a reduction in a subjective image quality. Point cloud data before update of position information is searched for a point to be compared, the point to be compared being a point to be compared with a current point, the current point being a processing target point of the point cloud data with the updated position information, and the retrieved point to be compared is compared with the current point, and attribute information of the current point is set. The present disclosure is applicable to, for example, an information processing apparatus, an image processing apparatus, electronic equipment, an information processing method, a program, etc.

A loop filtering method includes calculating block effect value and pixel difference, judging, and performing average filtering, difference filtering, on not filtering according to judged result. A loop filter is composed of the filter judging region consisting of image storage unit, threshold determining unit, pixel difference unit, filter judging unit, and block effect value calculating unit, and the filter executing region consisting of trigger unit, filter determining unit, average filter and difference filter. It can increase the S / N ratio by 0.20 db.

An image encoding device includes: a first encoding unit for calculating a generated code amount by encoding image data for each GOP (Group of Picture); a code amount control unit for setting quantization information for realizing a target generated code amount based on the generated code amount; a quantization information distinguishing unit for calculating a DCT (Discrete Cosine Transform) coefficient, and distinguishing quantization information that minimizes the summation for each picture of remainders when performing division of the DCT coefficient, as quantization information used for performing the last encoding; a picture-type setting unit for setting a picture type to the image data for each GOP, and when this set picture type differs from the picture type of the distinguished quantization information, matching the picture types by controlling the settings of the subsequent picture types; and a second encoding unit for encoding the image data based on the set picture type.

The present invention relates to a method which can selectively skip a transform to alleviate problems of deterioration in compression rate and image quality which are mainly generated when applying the motion compensation-transform encoding image compression technique which is mainly used currently, to an image having a drastic spatial change in a pixel value such as a computer image, and can skip the transform by comparing a transform size of a current block and the size of a maximum transform skip block with each other, and provides an efficient signaling technique therefor. The method includes the steps of: comparing the size of a current transform unit (TU) with the size of a maximum transform skip block when determining whether to transform a residual signal corresponding to a difference value between the current block and a prediction block; and determining whether to perform the transform on the basis of the comparison information on the size of the TU and the size of the maximum transform skip block.

An adaptive loop filter classifies a decoded image from a deblocking filter into a smooth region class and an edge / texture region class, in accordance with results of edge detection from an edge detecting unit. The adaptive loop filter performs filter coefficient calculation for each of classified classes such that the residual difference between an original image from a screen rearranging buffer and an image from the deblocking filter is smallest. The adaptive loop filter performs filter processing using filter coefficients calculated for each classified class, and outputs an image after filter processing to frame memory. The filter may be applied to an image encoding device for performing encoding with the H.264 / AVC format as a basis, for example.

An image sensor for electronic cameras includes a plurality of light sensitive pixels arranged in rows and columns, wherein the pixels of a respective column can be read out via a respective column line and includes a plurality of data outputs, wherein a plurality of column lines are associated with the respective data output via at least one multiplexer device. The column lines are divided into a plurality of column line groups, wherein the respective column line group includes a plurality of column lines arranged next to one another; and wherein the number of column lines of the respective column line group corresponds to the number of the column lines associated with the respective data output.

The invention discloses a control method for constant code rate, which comprises the following steps: the current image is divided into a plurality of image groups; each image group performs image coding by adopting frame-layer code rate control and micro-block layer code rate control. The invention has the advantages of high control precision, small code stream fluctuation, good visual image quality of coding and low calculation complexity, therefore the invention is applicable for coding various software and hardware.

A foreground object part is distinguished from a substantially static background part of each videoframe within a video sequence. Each video frame is divided into a number of video blocks each of which comprises one or more pixels. A mask frame is generated in respect of each video frame, each mask frame having a mask block corresponding to each video block in each respective video frame. Each mask block is then set to either an object value, indicating that the corresponding video block in the corresponding video frame includes one or more pixels depicting a foreground object part, or to another value.

This invention provides a code rate control method, which applies code rate controls of a frame level and a macroblock level to encode images according to the image kind of the current frame.

Provided is a moving picture processing method which prevents degradation of subjective image quality caused by a motion of a moving picture. The moving picture processing method includes the following steps: a motion compensation step (S116) of shifting a noise indicated by a stored noise signal (FMN) according to a motion of a moving picture indicated by an input image signal (VIN), by performing motion compensation on the stored noise signal (FMN) that is used for superimposition on a picture indicated by an input image signal (VIN) and is stored in a noise frame memory (102); and a superimposition step (S120) of superimposing the shifted noise, as a superimposition noise signal SN, onto a current picture to be processed which is indicated by an input image signal (VIN) and is a moving picture that appears after the motion has occurred.

The present invention relates to an encoding apparatus, an encoding method, an encoding program, a decoding apparatus, a decoding method and a decoding program for adaptively controlling an encoding bit rate. Coefficient data obtained using wavelet transform is quantized, and segmented into bit planes. The coefficient data is entropy encoded from the bit plane of the MSB in a direction from an upper bit position to a lower bit position. Entropy encoding is stopped at a bit position where an amount of code has reached a target amount. On the bit plane, the coefficient data is entropy encoded from a lower frequency region to a higher frequency region. Data greatly affecting subjective image quality of a decoded image is selectively output while the amount of code is easily controlled.

An image processing method includes determining a first bandwidth based on original image data; computing a filter coefficient array for implementing frequency characteristics corresponding to a band limitation using the first bandwidth; generating first filtered image data by subjecting the original image data to a filtering process using the first filter coefficient array; deriving an objective image-quality estimation value of the first filtered image data, and computing an allocation coefficient used for determining an optimum bandwidth, based on the objective image-quality estimation value; determining the optimum bandwidth corresponding to the computed allocation coefficient by referring to a table in which a correspondence relationship between the allocation coefficient and the optimum bandwidth is defined; computing an optimum filter coefficient array for implementing the frequency characteristics corresponding to a band limitation using the optimum bandwidth; and generating optimum filtered image data by subjecting the original image data to a filtering process using the optimum filter coefficient array.

Provided is a moving picture processing method which prevents degradation of subjective image quality caused by a motion of a moving picture. The moving picture processing method includes the following steps: a motion compensation step (S116) of shifting a noise indicated by a stored noise signal (FMN) according to a motion of a moving picture indicated by an input image signal (VIN), by performing motion compensation on the stored noise signal (FMN) that is used for superimposition on a picture indicated by an input image signal (VIN) and is stored in a noise frame memory (102); and a superimposition step (S120) of superimposing the shifted noise, as a superimposition noise signal SN, onto a current picture to be processed which is indicated by an input image signal (VIN) and is a moving picture that appears after the motion has occurred.

A Gaussian pyramid and anisotropic filtering-based image depth of field rendering method disclosed by the present invention comprises the following steps of for an inputted depth image, constructing a Gaussian pyramid to pre-process; for the filtered depth image, utilizing an optical depth of field model to calculate the radius of a dispersion circle; for a focusing area in a dispersion circle image, constructing an isotropic Gaussian filtering kernel to filter; rounding and layering the filtered dispersion circle image; for the layers of the dispersion circle, constructing an anisotropic Gaussian filtering kernel and combining the inputted image to filter. According to the present invention, by constructing the Gaussian pyramid to pre-process the input scene depth image, and introducing the fuzzy and hierarchical anisotropic Gaussian filtering of the dispersion circle image, the rendering method is suitable for a scene depth image that is obtained by a mainstream depth image obtaining method, is low in resolution and has the black noisy points, and a rendering result has the better subjective image quality and a more accurate depth of field effect.

The invention provides a control method of constant bit-rate based on a handset television, which divides current images into a plurality of image groups and adopts frame layer rate control and macro-block layer rate control to each image group to code an image. The constant bit-rate control method provided by the invention has high control accuracy, little fluctuation of output bit stream, good subject image quality of coding and very low computation complexity, and is suitable for the application of coding of various software and hardware.

The invention discloses a self-adaptive frame skipping method and a device. The method comprises the following steps: the mean value of the motion vector of a current frame is calculated; the weighted sum is used as a smoothing factor, and the weighted sum is composed of the ratio of the mean value of the motion vector of the current frame and the mean value of the average motion vector of at least one adjacent latest frame which is calculated as well as the ratio of the relative mean value of the motion vector of the current frame and the relative mean value of the average motion vector of at least one adjacent latest frame which is calculated; and if the value of the smoothing factor is within the range of the threshold value, the current frame is processed through frame skipping. The self-adaptive frame skipping device is also provided. The self-adaptive frame skipping method and the device accord with the HVS visual characteristics, ensure that a terminal user obtains optimal visual quality, and satisfy the image quality without affecting the visual quality when a user views the image.

Disclosed herein is an image processing device for coding image data, which is divided by a given pixel count into macroblocks, on a macroblock-by-macroblock basis, the image processing device including: a quantization section configured to quantize image data on a macroblock-by-macroblock basis; a detection section configured to detect a macroblock containing a given color pixel from the image data; and a control section configured to control the quantization section, the control section including a distance calculation section configured to calculate the distance between a given image position in a picture of the image data and an image position in the macroblock detected by the detection section, and a bitrate control section configured to control the quantization section so that a higher bitrate is assigned with decreasing distance calculated by the distance calculation section.

An image sensor for electronic cameras has an image field having at least one image field block which includes a plurality of light sensitive pixels arranged in rows and columns for generating exposure dependent pixel signals, wherein the pixel signals of the pixels of the respective column of the respective image field block can be read out via one or more respective column lines extending parallel to one another. The image sensor has at least one first row of column amplifiers and at least one second row of column amplifiers for the respective image field block. The image sensor is adapted to read out the pixel signals of the pixels of the respective column of the respective image field block of an image or of two mutually following images partly via the at least one first row and partly via the at least one second row of column amplifiers (25).

The invention discloses a hardware-oriented joint denoising and demosaicing method, which comprises the following steps of: 1, calculating accurate weights in left, right, upper and lower directions in a window taking red or blue as a center according to the characteristics of Bayer Raw data, and interpolating all missing green information; 2, calculating a high-frequency coefficient required for interpolating a green channel; 3, calculating a global direction weight and a voting direction weight; 4, interpolating the green channel; 5, performing edge-preserving direction wavelet denoising on the green channel; 6, performing interpolation on the color channel by adopting an interpolation matrix designed for the color channel; and 7, performing direction wavelet denoising on the color channel. According to the method, in a low-cost range which can be accepted by a mobile hardware platform, the noise can be removed while the mosaic is removed, so that the image quality of an existing scheme can be improved, and the operation complexity and the power consumption are improved.

A loop filtering method includes calculating block effect value and pixel difference, judging, and performing average filtering, difference filtering, on not filtering according to judged result. A loop filter is composed of the filter judging region consisting of image storage unit, threshold determining unit, pixel difference unit, filter judging unit, and block effect value calculating unit, and the filter executing region consisting of trigger unit, filter determining unit, average filter and difference filter. It can increase the S / N ratio by 0.20 db.