Design Of Image Processing System Based On FPGA

In recent years,with the development of machine vision and deep learning,the technology of image processing based on FPGA procedure is becoming increasingly mature.With the increase of image sharpness,the requirement on image processing speed is higher and higher,and the parallelism of FPGA can finish the image processing quickly.Using FPGA as the hardware platform,this paper studies real-time video image processing,including the image edge detection and face detection in video image.First,I use the surveillance camera to collect analog signals.Then the analog signal is transformed into the signals of luminescence and chrominance by using matrix operation conversion.Then the decoding chip named ADV7180 decodes these video data and removes the blanking zone according to the ITU-R 656 protocol standard.After the YUV sampling format conversion,the system changes the YUV color space to RGB color space.Then the Verilog language is used to realize image filtering,gray image,edge detection,face detection module.Finally,display the real-time video image that is controlled by the VGA controller.This paper uses the fast algorithm of median filtering.Marginal detection is realized through the use of the Sobel edge detection operator process after using the skin color detection algorithm.Facial detection mainly involves the detection of skin colors,fast-connected component labeling,the extraction of features of lips,and in the end,uses NIOS II soft-core processors to calculate and calibrate the area of the face.In the course of design,the Verilog language is used to describe hardware,and the C language to optimize the performance,FPGA is used to perform parallel processing so as to design and construct the whole system and program and debug,and in the end,the real-time processing system of videos based on FPGA is successfully accomplished.