An Improved Median Filter Denoising Algorithm Structure Design Based On FPGA

Various noise interferences inevitably exist during image acquisition and transmission,thus affecting image quality,so image denoising is an indispensable part of image processing.With the development of the semiconductor chip industry,the traditional method of image noise removal by software only can no longer meet the demand for real-time performance in some application scenarios.Improving the speed of the denoising process while ensuring the image denoising quality has become an important research topic.This paper proposes an improved median filtering algorithm by combining various algorithms’ advantages to address the severe degradation of the denoising effect of traditional filtering algorithms in a high-density pretzel noise environment.The algorithm sorts the pixel values in the filter window,performs extreme value cropping,and performs weighted average processing on the remaining pixel values according to their distance from the median point.The processed pixel values are compared with the center pixel value of the original filter window by making a difference operation,and the obtained difference value is compared with the set threshold to determine whether the original center pixel point is a noise point.If the point is a noise point,the pixel value of the point is replaced.Otherwise,it is retained.Secondly,to solve the problem of real-time denoising processing,the improved median filtering algorithm is implemented on a field-programmable logic gate array(FPGA)hardware platform in this paper.Each module of the hardware system is designed in Quartus Prime 18.1 software platform using Verilog language.Including image acquisition and input module,grayscale conversion module,video noise addition module,improved median filtering module,edge enhancement module,and video output display module.Finally,the improved median filtering algorithm is simulated using Matlab software.And the three evaluation indexes of peak signal-to-noise ratio(PSNR),mean square error(MSE),and structural similarity(SSIM)are used to compare with other filtering algorithms for noise removal.The main functional modules of the system are simulated using Modelsim software to verify the correctness of the module design and timing logic,and the designed system files are burned to the FPGA to observe the experimental results.The experiments demonstrate that the improved median filtering system can perform realtime denoising of images and achieve better denoising results.