Research On Deep Sea Video Compression Method Based On Region Of Interest

This paper originates from the video coding and decoding problem in the special project of”Full ocean depth video acquisition,transmission,processing technology research and development,system integration and demonstration application” of the national key R & D plan ”Deep-sea key technology and equipment”,focuses on the video compression method based on the ROI in deep-sea environment,completes the development of the compression system,and verifies the performance of the compression algorithm in the actual system.The full ocean depth refers to the depth of 11,000 meters under the sea.When the lander dives to the full depth of the sea,the submarine video needs to be transmitted back to the surface mother ship through a special cable.Due to the limitation of the operating environment,the bandwidth of the communication link is narrow,and the stability of the line is poor.In deep-sea surveillance applications,there are a large number of static areas(sea water background)in the scene.These background pixels have little effect on the monitoring of the submarine scene,and the actual operation often focuses on the area of the scientific research target in the video.If the sea water background and the scientific research target area are coded according to the same coding strategy in video coding,and the same coding quality is given,it undoubtedly wastes precious bandwidth resources.By detecting scientific research objects and setting them as ROI,improving the coding quality of ROI and reducing the coding quality of seawater background,not only play a great role in improving the subjective experience of video,but also can achieve the effective use of bandwidth resources.In this paper,the characteristics of the deep-sea video data are analyzed to study the method of ROI extraction,and based on this,a video coding algorithm based on the ROI is designed.At the same time,we completed the implementation of the algorithm in the hardware system and evaluated the effect of the algorithm.The main work of this article includes:(1)At present,there are few researches on ROI extraction in deep-sea scene,and the speed cannot meet the real-time processing of UHD video.In this paper,an improved ROI extraction algorithm for deep-sea video coding system is proposed.Firstly,according to the characteristics of deep-sea scene,a static target area extraction algorithm for deep-sea scene is proposed.Compared with other algorithms,this algorithm is more suitable for deep-sea scene.However,the efficiency of this algorithm is not ideal.Therefore,this paper improves the foreground judgment method in the algorithm and effectively improves the calculation speed.Secondly,in order to solve the problem that the algorithm cannot extract the dynamic target area,the two-frame difference method is selected to improve,which solves the problem of insufficient operation efficiency of the two-frame difference method and the problem of ghost in the extraction result.Finally,the improved static target region extraction algorithm is combined with the improved two-frame difference method to propose an improved ROI extraction algorithm for deep-sea video coding systems.The experimental results show that the algorithm proposed in this paper performs well in the deep-sea scene,and can extract the static target area and the dynamic target area at the same time.Compared with other algorithms,the operation speed is greatly improved and real-time processing of 4k@30fps UHD video can be achieved.(2)The deep-sea environment limits the volume of the detection equipment.In response to this problem,this paper designs and implements a deep-sea UHD video encoding system based on Xilinx Ultrascale+FPGA,and implements an ROI-based encoding algorithm on the system.This system is an embedded system,so this paper also completes the design of the hardware and software co-design,namely the design and implementation of the underlying hardware architecture and the top software system.The test results show that the system meets the system index requirements in terms of resource consumption,coding time and code rate,and can achieve real-time encoding and transmission of 4k@25fps UHD video with good encoding quality.(3)The subjective quality assessment and objective quality assessment are performed on the video output from the deep-sea UHD video encoding system based on ROI,and the in-depth analysis of the experimental results is carried out to study the impact of bit rate and ROI settings on the quality of encoded video.The experimental results show that the quality of the output video of the deep-sea video coding system proposed in this paper is better than that of the system without ROI,and the ROI maintains better video quality at low bit rates.