Article Preview
Top1. Introduction
Obtain voice, graphics, images and videos through multimedia, and use multimedia technology to quickly and accurately obtain information, text and voice in video from images (Jiang & Li, 2021). Multimedia technology combines communication means, signal and image processing means and transmission means, but it not only brings benefits to people, but also brings some troubles to people. The amount of information of voice and text is much smaller than that of image and video. If video and image are not compressed, it will cause large resource space, small space utilization and difficult communication and transmission, Therefore, it is urgent to solve the problems of image, video and other media storage, transmission and communication (Pfaff et al, 2019). There are two kinds of video compression, one is lossy compression and the other is lossless compression. After the video is compressed and stored, the video is reconstructed and decoded to see whether the video is consistent with the state before compression. If the state before compression can be restored, it indicates lossless compression; If the state before compression cannot be restored and there is a certain loss in the decoding process, it indicates that it is lossy compression. If the requirements for video quality are not high, such as video conference, the video is allowed to have loss during decoding, lossy compression can be selected (Irannejad & Mahdavi-Nasab, 2018). For areas with high video image quality requirements, such as medical images, satellite remote sensing images, image recognition, image distortion is not allowed. In order to avoid distortion in the video decoding process, lossless compression video will be selected, so as to reduce the compression ratio and ensure that the video has no loss after compression. With the progress of science and technology, people's demand for video is increasing, and image and video coding and decoding technology plays an increasingly important role in people's daily life. Due to the large amount of video data, compression coded video lossless decoding is particularly important (Bidgoli et al., 2020).
Dai et al. (2018) proposed a compression video codec method based on super-resolution reconstruction, which collects compressed video, performs denoising processing on the collected compressed video, performs HEVC (High Efficiency Video Coding) coding on the compressed video according to the denoising results, and reconstructs the encoded compressed video to its original size through super-resolution reconstruction method. LSM-MNLR is used to decode reconstructed compressed video lossless. However, the SNR of the video can only be improved by 1dB. Wang et al. (2018) Compressed-sensing deep video codec method based on adaptive bit rate allocation. Firstly, the compressed sensing method is used to compress the video to obtain the compressed sensing depth video, and the video is segmented to obtain video blocks of the same size. Then, the Canny operator is used to extract the boundary of the video blocks. According to the extraction results, the encoding and decoding model is constructed by the adaptive rate allocation method to complete the encoding and decoding of compressed sensing depth video. This method can improve the signal-to-noise ratio (SNR) by about 20dB for a specific set of data. Sun (2018) put forward by using the method of multidimensional vector matrix to scan motion adaptive video compression codec, multidimensional vector matrix orthogonal transform compression technique was used to construct multidimensional moderate video matrix orthogonal transform coefficient, energy distribution function of scanning motion adaptive video compression processing, according to the results of the compression codec model was constructed, the motion adaptive scanning video is codec. Experimental results show that this method can improve the 4-dimensional scanning effect of images, and the scanning accuracy can be increased by about 10%. Liu et al. (2019) through adaptive interpolation filter for nondestructive video codec, via minimization interframe prediction error to build adaptive interpolation filter, the adaptive interpolation filter will be divided into sub blocks of non-destructive video pixels, pairs in the preset period piece of similarity calculation, according to the calculation results, The video is compressed and encoded, and the encoding stream is obtained and decoded. Experimental results show that this method can save up to 2.2% of the code rate.