50results about How to "Reduce the theoretical limit entropy" patented technology

The invention relates to a compression method and equipment based on image prediction. The method comprises the following steps: receiving a to-be-compressed image frame; determining multiple encodingunits of the to-be-compressed image frame; predicting multiple encoding units in turn by adopting multiple prediction modes; determining an optimal prediction mode of each encoding unit by utilizinga preset formula according to prediction results of multiple encoding units so as to accomplish the compression of the to-be-compressed image frame. Compared with the existing method, the compressionmethod provided by the invention can improve the image encoding compression rate for the to-be-compressed image frames at different scenes, thereby further reducing the theory limit entropy of the compression.

The invention relates to a lossless compression method and a lossless compression device for video data. The lossless compression method comprises the following steps: receiving video data comprisingat least one selected from an image, an image sequence and a video; dividing the video data into a plurality of coding blocks; and selecting one compression mode from a plurality of predetermined compression modes to encode the encoding block to form a video data code stream. According to the method and device, the coding block is coded through two compression modes to obtain compressed data, a final compression mode and the compressed data are selected according to the absolute value sum of the compressed data to form the video data code stream, the image coding compression rate can be improved for a complex texture area, and the theoretical limit entropy of compression is further reduced.

The invention relates to compression coding methods. A method comprises the following steps of (a), obtaining a first macro block; (b), obtaining sampling points and non-sampling points of the first macro block; (c), obtaining first predicted residual errors according to the sampling points and the non-sampling points; (d), obtaining residual error distribution types according to the first predicted residual errors; (e), obtaining a quantization matrix according to the residual error distribution types; and (f), obtaining quantized residual errors according to the quantization matrix and the first predicted residual errors. According to the method, for an image with complex textures, prediction precision for the textures at boundaries can be greatly improved; theoretical limit entropy is reduced; a bandwidth compression ratio is improved; moreover, quantization parameters are adaptively set according to different texture complexity; the transmission bit number is further reduced; and the bandwidth compression ratio is improved.

The invention relates to a method for selecting a bandwidth compression prediction mode. The method comprises the following steps of: selecting a macro block in an image; calculating a first prediction residual error of the macro block by adopting a first prediction mode; calculating a second prediction residual error of the macro block by adopting a second prediction mode; calculating a first residual subjective sum according to the first prediction residual error, and calculating a second residual subjective sum according to the second prediction residual error; and selecting the minimum value in the subjective sum of the first residual error and the subjective sum of the second residual error, and performing bandwidth compression by adopting a prediction mode corresponding to the minimum value. According to the embodiment of the invention, the most suitable prediction mode is selected from multiple prediction modes by adopting a selector, so that the most suitable reference pixel can be found, the minimum prediction residual error is obtained, and the theoretical limit entropy is reduced.

The invention relates to an adaptive template prediction method used for bandwidth compression. The method comprises the following steps: step 1, building an adaptive template list; step 2, performinginitial filling for the adaptive template list; step 3, selecting a reference MB, and updating the adaptive template list according to the reference MB; step 4, selecting one or more candidate adaptive templates from the updated adaptive template list; step 5, weighting the candidate adaptive templates to obtain a predicted value of the current MB; step 6, acquiring a predicted residual error ofthe current MB according to the predicted value. In the adaptive template prediction method provided by the invention, through customizing the adaptive template, selecting reference pixels for the current pixel according to the template and then calculating the predicted residual error, a problem of how to improve quality of a prediction module when texture of a to-be-compressed picture is complicated and changeable is solved, and theoretic limit entropy can be reduced further.

The invention relates to a dual-mode selection prediction method for a complex texture in bandwidth compression, which comprises the steps of: partitioning a to-be-encoded video image into a pluralityof macro blocks, and determining a to-be-encoded pixel component; by adopting an adaptive template prediction method, selecting a first reference pixel of each currently encoded pixel in a currentlyencoded macro block in an adaptive template, and calculating to obtain a group of first predicted residuals; by adopting an adaptive rectangular window prediction method, selecting a second referencepixel of each currently encoded pixel in the currently encoded macro block in a rectangular prediction search window, and calculating to obtain one group of second predicted residuals; calculating a first subjective difference according to one group of first predicted residuals, and calculating a second subjective difference according to one group of second predicted residuals; and comparing the first subjective difference with the second subjective difference, and determining an optimal prediction method of the currently encoded macro block to obtain one group of optimal predicted residuals.The dual-mode selection prediction method uses the macro blocks as prediction units, and according to different texture characteristics in different regions of the image, the optimal prediction methods are adaptively selected, so that the dual-mode selection prediction method is better in prediction effect.

The invention relates to an adaptive template prediction method for bandwidth compression, which comprises the steps of 1, defining the epitope quantity and epitope serial numbers of an adaptive template; 2, updating the adaptive template list according to MB reconstruction values of adjacent reference directions of the current MB (Macro block); 3, carrying out adaptive texture prediction on the current MB and the MB reconstruction values in the adaptive template list to obtain prediction residuals of the current MB; and 4, judging whether all MB prediction residuals are obtained or not, if so, ending the prediction, otherwise, jumping to the step 2. The predicted pixel value of the current micro block is calculated through defining the adaptive template. Compared with the existing method,the adaptive template prediction method can improve the accuracy of solving prediction residuals for a complex texture region, thereby further reducing the theoretical limit entropy, and improving the bandwidth compression rate.

The invention relates to an adaptive texture gradual change prediction method in bandwidth compression, which comprises the steps of selecting N sampling modes to sample the current MB (Macro block),wherein N is a natural number greater than 1; predicting the current MB to obtain prediction residuals of the current MB; solving the sum of absolute difference (SAD) of the current MB; and determining the sampling mode of the current MB according to the sum of difference. According to the invention, various sampling modes are adopted for the current MB, and the prediction residuals and the SAD ofthe current MB are obtained. When a compressed image with the texture being complex is processed, for the current MB located on the texture boundary of the current to-be-compressed image, the correlation between the current MB and the surrounding MBs is poor because the current MB and the surrounding MBs are not in the same texture region. According to the gradual change theory of the texture, the prediction residuals are obtained according to texture characteristics of the current MB rather than depend on the surrounding MBs of the current MB, and the accuracy of solving the prediction residuals for a complex texture region can be improved.

The invention relates to a prediction method of video compression. The method comprises that an image is divided into multiple MBs in the same size, and one MB is selected as an MB to be predicted; the MB to be predicted is predicted in a first prediction mode, and a first residual error subjective sum is obtained; the MB to be predicted is predicted in a second prediction mode, and a second firstresidual error subjective sum is obtained; and the first residual error subjective sum is compared with the second residual error subjective sum to select a final prediction mode of the MB to be predicted. A better prediction method can be selected from the first and second prediction modes via a prediction selection algorithm, and the prediction effect of a complex texture image is further optimized.