Video compression artifact adaptive removing method based on depth learning

Abstract

The invention discloses a video compression artifact adaptive removing method based on depth learning, which adopts a depth-dense connected convolution network to automatically extract the compressioncharacteristics of a video frame, and can effectively avoid the shortcomings caused by the manual design of a filter in the traditional method. The invention acts on the post-processing stage of thevideo, and does not affect the processing flow and the real-time performance of the existing video coding and decoding algorithm. A new image quality prediction model is proposed to realize automaticselection of compression artifacts with different intensities, which has strong adaptive ability. Deep-connected convolution network is used to remove video compression artifacts, which can effectively alleviate the problem of gradient disappearance, deepen the network structure and enhance the network non-linear expression ability. At the same time, the network can also make full use of the characteristics of the middle layer, which not only enhances the propagation and multiplexing of features, but also greatly reduces the network parameters.

Description

technical field [0001] The invention relates to the field of video processing and deep learning technology, in particular to a deep learning-based adaptive removal method for video compression artifacts. Background technique [0002] Video compression artifact removal is a technique used to improve video quality. Among them, the compression artifacts of the video are produced by the encoding method of the video. [0003] With the rapid growth of Internet video data, in order to control the cost of video storage and transmission, a higher compression rate is usually used in video encoding. Generally speaking, lossy compression algorithms, such as the common MPEG and H.26X series, are used in the video encoding process. Among them, H.264 is the most widely used video coding method. While reducing the video volume, these encoding methods introduce video artifacts such as blockiness, ringing, flickering, and mosquito noise due to compression. These artifacts can severely deg...

AI Technical Summary

Problems solved by technology

While alleviating video compression artifacts, these two built-in filters also increase the complexity of encoding and affect the real-time performance of the encoding algorithm.

[0005] In general, these traditional video compression artifact removal methods have the following problems: First, filters need to be manually designed, and such filters are usually only for a certain type of artifact, and the versatility is poor

Second, the threshold of the filter needs to be set according to experience. The setting of the threshold usually has a greater impact on the filtering results, and the robustness is weak.

Third, the use of embedded filters to alleviate video compression artifacts increases coding complexity and affects the real-time performance of coding algorithms

Fourth, traditional algorithms usually seldom use the effective information generated by the encoding process, it is difficult to automatically adjust the filter strength, and the adaptive ability is weak

However, the current out-of-loop filtering method based on deep learning technology still lacks the ability of self-adaptation

In other words, a single convolutional neural network model cannot handle video artifacts of various intensities well

The transcoding of Internet video usually adopts a constant bit rate (CBR) method, which will lead to various compression artifacts of different strengths in the same video sequence

Images