Content-based HEVC (High Efficiency Video Coding) code stream quality prediction model

Abstract

The invention discloses a content-based HEVC (High Efficiency Video Coding) code stream quality prediction model. The content-based HEVC code stream quality prediction model comprises a video feature extraction module, a content type classification matrix calculation module and a quality prediction model module, wherein the video feature extraction module is used for building a content correlation matrix in the content type classification matrix calculation module by using video feature information extracted from code streams, and then building a video quality prediction model through the quality prediction model module by using the content correlation matrix in order to predict the video quality; and the content type classification matrix calculation module comprises a time domain complexity sub-module, an airspace complexity sub-module, a code rate standard difference sub-module and a non-zero transformation coefficient percentage sub-module. Through adoption of the content-based HEVC code stream quality prediction model, the quality of a coded video can be predicted by the information in the code streams under the situation of ensuring low complexity, and the prediction accuracy is higher than other algorithms.

Description

technical field [0001] The invention relates to a video quality measurement system, in particular to a video quality prediction model, which is applied to the technical field of video codec for predicting coded video quality by using code stream information. Background technique [0002] The perceived quality of the end user is a key factor in forming the quality of service (QoE) of the user. Under the same encoding parameters, such as frame rate, sampling format, resolution, etc., different video contents can obtain different sensory qualities. Due to the content dependence of video quality, it is especially necessary to consider different content characteristics when performing video quality prediction. [0003] Common objective measures for perceptual video quality estimation can be divided into two categories: based on human visual properties and based on video parameters. In order to ensure low complexity of video quality measurement, it is necessary to make full use ...

AI Technical Summary

Problems solved by technology

In the video quality prediction model in [4], the types of user quality of experience content considered include: bit rate at the sending end, block distortion rate and average distortion length, but motion features and other characteristics are not considered

Literature [5] uses space-time domain content features to establish a prediction algorithm, but does not consider the type of frame

Literature [6] fused HEVC encoding parameters, quantization parameters and content types into the quality prediction model, but did not consider motion features and sequence complexity information

Images