Research On Lossless Compression Technology Of Infrared Video Based On Heterogeneous Multi-core Processor

In recent years,with the advent of the era of big data and rapid development,images and video are widely used in communications,television broadcasting,multimedia transmission and other fields,and are important media for transmitting data and human access to information.Among them,infrared video plays a role in important areas such as reconnaissance,surveillance and security.This is because infrared radiation has a long distance and strong penetrability.In the environment with low visibility,infrared imaging technology can still pass thermal radiation energy.Get information to make up for the lack of visible light imaging.Therefore,in the rain,snow,night,fog,infrared video can still capture practical information,playing an important role in the field of reconnaissance,monitoring and security.In the process of data development,the amount of data of infrared video increases rapidly,storage consumes a large amount of memory and direct transmission causes a huge waste of bandwidth.Regardless of the speed of device update or the speed of economic development,effective encoding of infrared video is most effective and Convenient solution.In view of the main field of action of infrared video and the characteristics related to daily safety,it is necessary to keep the video data in the process of storage and transmission without distortion,otherwise important safety information may be lost.Therefore,research on lossless compression technology of infrared video has been studied by everyone.hot spot.This paper proposes a low-complexity lossless compression algorithm based on embedded system operation for infrared video.For single-frame image size of 640x512,16-bit infrared video per pixel,we strive to obtain a compression ratio of 2.0 and improve the efficiency of video processing.The frame rate is 50 Hz.We implemented the system using the AM5728 development board,which has a heterogeneous multi-core ARM Cortex-A15 RISC CPU and dual DSP cores to meet the strong processing demands of modern embedded products.ARM processor has high speed,low power consumption,has its own efficient compilation and debugging environment,and a wide range of third-party cooperation support,which can facilitate system porting and development.The DSP processor has its own complete instruction system,which uses digital signals to process information.It has efficient computing power and high flexibility for software programming.It provides an effective way to participate in various complex applications and can be implemented in real time in real time.Various signal processing algorithms.As the parallel computing language,OpenCL is the core of distributed processing to provide a high-speed computing solution.It has good parallel computing advantages for large-scale data with large density and similar data format.The development board AM5728 supports OpenCL for program development,making it easy for users to perform high-level computing tasks with DSP acceleration.Video is the processing of frame data.There are a lot of similar pixels in the interframe frame,which perfectly fits the requirement of OpenCL to use parallel computing with dual DSPs,and accelerates the computational speed of the algorithm.In this paper,the research on infrared video lossless compression technology has certain innovations in the prediction step.Firstly,the infrared video is decomposed into sequence frames and each frame image is divided into blocks,and the correlation is predicted by time and space;then the residuals obtained by the prediction are calculated,and the residuals with correlations are still predicted twice,and the optimal is selected.The predictor;finally,the processed residual is entropy encoded to complete the lossless compression of the infrared video.After comparing the results of the algorithm,the general compression algorithm Gzip and the image compression algorithm LOCO-I for multiple infrared videos,it shows that the compression algorithm can obtain a higher compression ratio and achieve the compression ratio of 2.0.At the same time,using OpenCL to achieve dual DSP parallel computing,the compression efficiency is higher,and the processing speed of about 0.02 s per frame can be obtained,that is,the target of processing frame frequency 50 Hz is realized,and the encoding method does not need additional auxiliary information,which is simple and convenient.It can effectively improve the infrared video storage problem and ultimately achieve efficient information transmission.