基于快速中值边缘检测系统的FPGA实现  

FPGA implementation based on a fast median edge detection system

作　　者：徐志辉 和晓军 XU Zhihui;HE Xiaojun(Shenyang Ligong University,Shenyang 110159,China)

机构地区：[1]沈阳理工大学自动化与电气工程学院,辽宁沈阳110159

出　　处：《通信与信息技术》2024年第5期29-34,共6页Communication & Information Technology

摘　　要：针对经典的Sobel边缘检测系统存在定位不准确以及消除高斯与椒盐混合噪声差而导致的图像边缘特征提取模糊或无法提取的问题,提出了一种改进的快速中值滤波算法与八方向的Sobel算子相结合的边缘检测算法。首先,采用快速中值滤波算法消除图片噪点;然后,再采用八方向的Sobel算子提高边缘定位的准确性,提高算法在面对噪声时的边缘信息提取的准确性。改进的快速中值边缘检测系统能在密集的混合噪声中准确定位到图像边缘。改进后的算法计算量倍增,选用FPGA硬件实现,实验结果验证了算法的可行性。Aiming at the problem that the classic Sobel edge detection system has inaccurate positioning and eliminates the blurred or impossible extraction of image edge feature extraction caused by the noise difference between Gaussian and pepper and salt,an improved fast median filtering algorithm combined with the eight-direction Sobel operator is proposed.Firstly,the fast median filter-ing algorithm is used to eliminate the image noise,and then the eight-direction Sobel operator is used to improve the accuracy of edge localization and improve the accuracy of edge information extraction in the face of noise.An improved fast median edge detection sys-tem accurately locates image edges in dense mixed noise.The improved algorithm doubles the amount of computation and uses FPGA hardware to implement it,and the experimental results verify the feasibility of the algorithm.

关 键 词：混合噪声 快速中值滤波 八方向的Sobel算子 边缘检测 FPGA 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]