Compression And Encoding Of Surveillance Video Based On Region Of Interest

Today,with the social security is more and more be valued,video surveillance has been increased exponentially.How to transfer and store vast video surveillance becomes an urgent problem.First,the technology of Region-of-Interest encoding and the characteristic of video surveillance is introduced.And then several kinds of encoding technology which are suitable to video surveillance is analysed.Based on this,this work proposed a new method of initial quantization parameter computation and different region encoding method,and make improvements on region-of-interest detection where there exists a short-stop object.This work proposed a detection method about three background differences based on time domain to solve the problem of video which exists a short-stop object.By giving the definition of static background frame,gradual background frame and current background frame,and computing the differences of the three background frames,we can get the region where the short-stop object is.According to the modern encoding standard,with the same encoding bitrate,different sequences have the same initial quantization parameter.This work proposed a method of initial quantization parameter based on content complexity.Given the same encoding bitrate and the content complexity of the first frame in a sequence,the quantization parameter which makes the least variation of Peak-to-Signal-Ratio,quantization parameter and encoding bits is chosen.This work also proposed a method of encoding prediction mode set division and quantization parameter adaption.For the region-of-interest,transition region and background region,each choose the best mode from corresponding set instead of the whole set,and increase or decrease the quantization parameter to shorten the encoding time and improve the encoding quality of region-of-interest.