Research Of High-speed Image Processing Algorithms Based On FPGA

Real-time processing technology is widely used in modern economy and national defence.In the implement of image guidance field ,real-time processing technology is the research focus.The progress of VLSI technology provided the hardware foundation for the digital image real-time processing technology. The characteristic of FPGA (field programmable gate array) lead itextremely fit be used in image processing to carry out some parallel processing tasks.This thesis analyzes the character and the base way of image processing. The max- medianfilter in all directions for target identify is given and it's mathematic Statistic characteristic isanalyzed, availability is proved.This thesis expatiates the principle and strategy for FPGA design, elementarily studiesparallel-structure of algorithm for FPGA image processing. Through improving algorithms andoptimizing structures, with conditions of e?ciently using of hardware resource, the utility of theparallel algorithm is exploited more e?ciently in the design. With pipeline optimization algo-rithm, the speed of module dealing with images is largely improved.With leading purpose of project application, using FPGA on high-speed-parallel-image-processing ?at, some processing algorithm is implemented like fast Histogram statistics, medianfilter , Max-median Filter in All Directions, average filter, 9*9 cross-high-pass filter, auto-fit lin-earity image enhance ,sobel module edge detecting using , binary statistic filter, target labellingand so on. furthermore, several peripheral interface module for image processing like cameralink-interface, SDRAM-controller ,image store module is also implemented . Basing on module design,three image processing ?ows have been implemented for di?erent characters targets. Their physi-cal implement is running at 192MHz. This is mean that, the processing ability is more than 120c/s for 1280*1024 16bit gray images.