Research On Intra Lossless Coding In HEVC

As we all know, video plays an important role in people’s daily life and entertainment life. Video is a combination of multi-frame image is made, if the video images are encoded directly in the transmission of video, there will be a great waste of network bandwidth, therefore, the need for effective video image compression and storage, network transmission after coding representation. However, there is temporal redundancy and spatial redundancy between video images, so how to effectively use spatial temporal redundancy has been a hot research for effective team video in image coding and compression.The latest international video coding standard HEVC(High Efficiency Video Coding) includes lossy compression coding and lossless compression coding. Lossy and lossless compression applications in different areas. In a television broadcast, video teleconferencing, lossy compression coding can achieve better efficiency. Lossless compression is often used in the area which image fidelity requires good art,such as medical images, satellite remote sensing, fingerprint identification and the like. Firstly, we study the HEVC intra lossless coding in this paper, since in the lossless intra coding, video coding residuals are directly transmitted to the entropy coding, there will be a waste of bits to be encoded. Therefore, in this paper, based on a conversion type HADMARD butterfly transformation, we propose the residuals to be transformed, then transmitted, thereby reducing the number of bits. It can effectively reduce the bitrate in HEVC intra lossless encoding.Since HEVC lossless intra coding of each coding unit is the composition of the luminance and chrominance encoding unit. For residual luminance block coding, we propose a HADMARD butterfly transform, as the residual is the difference between the residual block after the intra prediction according to the luminance blocks obtained prediction mode and the reference pixels. In lossless coding, the residual coefficients bypass the transformation, quantization code and the in-loop filter, directly be transmitted to be encoded. In lossy coding, transform process is commonly adopted the DCT transform, after which the energy will be concentrated in the high frequency region, and then quantized, the energy of low-frequency region is set to 0, so as to effectively compress the number of bits to be coded in residual block. However, lossless coding requirements the decoded video image and the original image is identical, so the encoder cannot use transform coding, quantization step. In HEVC the residual block is encoded directly encoded, and a great waste of bits coded. Therefore, after the butterfly transform proposed in this paper, it can effectively save the encoding residual number of bits to be encoded, thereby reducing the bitrate of intra lossless encoding, improves the performance.In this paper, using the transform based on the HADMARD row butterfly transform, or directly encode the residual coefficients in HEVC, based on adaptive decision between the two modes, so as to effectively improve the coding performance. Experimental results show that after using this transformation, for five standard sequence HEVC tests, compared with the reference software HM15.0, average bitrate is saved 0.92%.Using the transform based on the column HADMARD butterfly transform, or directly encode the residual coefficients in HEVC, based on adaptive decision between the two modes, so as to effectively improve the coding performance. Experimental results show that after using this transformation, for five standard sequence HEVC tests, compared with the reference software HM15.0, average bitrate is saved 0.98%.HADMARD two-dimensional butterfly transform, and encode the residual block directly in HEVC lossless coding, using any adaptive manner, so as to effectively improve the coding performance. Experimental results show that after using this transformation, for five standard sequence HEVC test, compared with the reference software HM15.0, average bit rate is saved about 2.38%, in this paper, we propose to use the column HADMARD butterfly transformation first, and the row HADMARD butterfly transform second.Finally with the mode in HEVC to directly encode their residual data, there will be four mode to process the residual block, so we set two flags, the encoder decide which mode to be selected by the bit rate, and the flag will be passed to the decoder to decode in the same way. For five standard sequence HEVC tests and compared with the reference software HM15.0, the bit rate is reduced average 2.69%. According to the statistical properties of the residuals after the butterfly transform proposed in this paper and we change the appropriate scan mode for the block before the entropy encoding for 4*4 size transform block, and the result shows that compared with HM15.0 reference software, the average bit rate of five standard test sequences is saved 3.71 %.