METHOD FOR COMPRESSING VIDEO BASED ON OBJECT USING QUARDRATIC PROGRAMMING

摘要

PURPOSE: A method for compressing video based on an object using a quardratic programming is provided to realize video compression with an easy algorithm by changing only an encoder. CONSTITUTION: An object part is extracted from one block of a video and all other parts of the block except the inside of the object are assumed as variables(S11-S12). The video is transformed through a discrete cosine transform(S13). Values of the variables are decided by using a quardratic programming to make background pixels of radio frequency components zero(S14). The decided values of the variables are substituted for the pixels of the video(S15). The padded video is compressed by a JPEG(Joint Photographic Expert Group) compressing method and a peak signal-to-signal noise ratio of the padded video is compared with a peak signal-to-signal noise ratio of an original video(S16).