Research On ROI-based Video Compression And Evaluation Algorithm

With the continuous improvement of video pixels in recent years,the industry has put forward higher requirements for video compression algorithms.HEVC is a relatively advanced lossy compression algorithm,and has been further modified on the H264 video coding framework.The performance of the game saves nearly half of the compression code rate.Although HEVC has a better compression ratio under the same performance,too long coding time makes it unsuitable for real-time video processing.By analyzing the ROI characteristics of the human visual system,the mathematical model is used to quantify the ROI features of the human visual system,and the quantized attention coefficients are used to modify the HEVC quadtree partitioning model.The algorithm finally encodes the NROI region quickly.The speedup of the HEVC compression algorithm is achieved.After experimental testing,the algorithm can increase the encoding speed by about 30% without affecting the user's feelings.In general,the research work of this paper mainly includes the following aspects:1.Based on the HEVC video compression algorithm,through the analysis of the HEVC hybrid coding framework,the possibility of using the attention characteristics of the human visual system to improve the HEVC coding rate is given.Furthermore,the characteristics of the human visual system are analyzed from the aspects of the structural features of the eyeball and the test data of the eye tracker.The tendency of the human visual system in the three aspects of the position,texture and motion characteristics of the video is obtained.It provides a theoretical basis for the subsequent modeling of the characteristics of the human visual system.2.Aiming at the defect that the industry generally adopts a single ROI factor to model the characteristics of human visual system attention,this paper verifies the attention of these features to the human visual system by analyzing the positional features,texture features and motion characteristics of the video.The characteristics affect the impact,and based on this,the mathematical models are used to quantify these three features,and the quantized attention coefficients are used to characterize the attention characteristics of the human visual system.Compared with other three similar ROI algorithms,the proposed algorithm is more robust.3.Considering the defect of HEVC video compression algorithm in coding rate,this paper uses the constructed human visual system attention model to modify the HEVC quadtree partitioning method.For the NROI region,the algorithm allocates smaller CU depth to reduce the encoding time.For ROI regions,the algorithm allocates a larger CU depth to preserve video detail.Through these two aspects of work,the algorithm of this paper can save about 30% of coding time without affecting the user's subjective feelings.4.Aiming at the defect that the standard video quality evaluation algorithm can not evaluate the ROI characteristics,this paper proposes a video quality evaluation algorithm based on position feature.The algorithm can fit the human visual system well by combining the test results of the eye tracker.The characteristics of attention.Then the proposed algorithm is used to test the four algorithms including the algorithm.The test results further verify that the proposed algorithm has greater advantages in the establishment of ROI model of human visual system.