基于八方向Sobel算子的边缘检测系统设计

Design of Edge Detection System Based on Eight-dimensional Sobel Operator

针对经典的Sobel边缘检测算法存在定位不准确并且有因阈值固定而导致边缘粗的问题,提出一种八方向的Sobel算子与八方向的非极大值抑制相结合的边缘检测算法。首先采用八方向的卷积模板提高边缘定位准确度,再采用八方向的非极大值抑制方法细化扩展后的图像边缘,提高算法检测复杂图像边缘时的信息准确度。由于改进后的算法计算量倍增,选用现场可编程逻辑门阵列(FPGA)平台实现,实验结果验证了系统进行实时检测的可行性。

Since the classical Sobel edge detection algorithm has the problem of inaccurate positioning and thick edges due to fixed thresholds,an edge detection algorithm combining eight-direction Sobel operator and eight-direction non-maximum suppression is proposed.First,the eight-direction convolution template is used to improve the edge positioning accuracy,and then the eight-direction non-maximum suppression method is used to refine and expand the edge.It can improve the information accuracy of the algorithm when detecting the edge of complex image.Because the calculation of the improved algorithm is doubled,the field programmable logic gate array(FPGA)platform is used,and the experiment verifies the feasibility of its real-time detection.