针对机器视觉领域对图像成像质量的要求,设计了基于FPGA的非线性双边滤波系统,通过双边滤波系统对噪声图像进行去噪预处理。从双边滤波原理、双边滤波“保边性”原因、双边滤波参数取值影响3方面对双边滤波系统进行理论分析,通过查找表使用、并行流水线处理等方式在FPGA上实现双边滤波系统,通过模块化设计和参数化编程的设计思想提升了双边滤波系统的可移植性。对设计的双边滤波系统进行测试验证,经双边滤波系统降噪后图像PSNR值提高了3.284 2 dB,SSIM值提高了21.82％,所设计的双边滤波系统起到了一定的噪声抑制作用。

In order to meet the requirement of image quality in machine vision field,a nonlinear bilateral filter system based on FPGA is designed,and the noise image is denoised through the bilateral filter system.This paper theoretically analyzes the bilateral filter system from three aspects:the principle of bilateral filter,the reason of "edge preservation" of bilateral filtering,and the influence of bilateral filter parameter values.The bilateral filter system is realized on FPGA through look-up tables,parallel pipeline processing,etc.The portability of bilateral filter system is improved through modular design and parametric programming.The designed bilateral filter system is tested and verified.The PSNR value of the image after noise reduction by the bilateral filter system is 3.284 2 dB higher than that of the noisy image,and the SSIM is 21.82% higher.The designed bilateral filter system has achieved certain noise suppression effect.