The Research And Implementation Of Image Compression Technology Based On Compressed Sensing

With the rapid development of the mobile Internet and multimedia services，mobile phones and other mobile intelligent terminal which are small, portable andinstant in communication are widely used in our daily life. The application oftraditional video coding method, due to its highly complex processes in the videocoding side and its requirement for stable channel, is limited in the channel which isless stable and in the mobile intelligent terminals whose computing resources andenergy consumption are constrained.Compressed sensing has been used on manyoccasions because it samples speed signal with a low-speed sample rate,andcompletes the data compression at the same time. The distribute vedio compresssensing theory which is derived from the theory of compressed sensing overcomestwo shortcomings of the application of the traditional video compression coding.This paper introduces compressed sensing in detail and disusses the three keyissues in the theory including the sparse representataion of signals, the design of theobservation matrix and the signal reconstruction algorithm. The effect of thecompressed image restoration with different orthogonal basises, the different ratio ofthe observation and the different signal reconstruction algorithms is analyzedthrough the simulation results.According to the characteristics of the video signal and distributed compressedsensing theory,an application framework of the distributed vedio compressedsensing is designed.A series of simulations of the two image sequences with slightmovement and intense movement are performed. The properity of applicationframework is improved according to the simulation results. At last an applicationframwork of the self-adaptive threshold-judged distributed vedio compressedsensing is put forward.The image acquisition system based on FPGA and DSP hardware platform isdesigned. The detailed modular design is put forward as well. It includes thecombined activation of DSP and FPGA, the initial configurations of the camera andthe network chip, the pretreatment of the image signal in FPGA through theconnection of a FIFO and a dual-port RAM, the communication between DSP andFPGA through a dual-port RAM, the transmission of etherent data through W5100.At last the software of Ethernet communication on PC taking advantage of theWINSOCKET functions in VC++is designed.At last, the acquisiton of image based on compressed sensing is realized by theimage acquisition system which is designed in this paper.The randomized observation in the compressed sensing is runned in the DSP. The compressedsensing signal reconstruction algorithm is runned in the PC to reconstruct the imagesignal.