基于方向可控滤波器的图像边缘检测算法研究

Research on edge detection algorithm of image based on directional controllable filter

传统图像处理中的边缘检测方法只在一个边缘像素周围的8邻域中寻找8个方向的边缘像素,而实际的边缘方向理论上可以是360°中的任意方向。本文利用方向可控算子原理,使用一个包含最少基本滤波器的二阶拟合x-y分离滤波器,进而将Canny准则与矩恒定的阈值选择原则相结合,对图像进行边缘检测。实验结果表明,所提出的检测策略在效果上明显优于其它方法。

Traditional edge detection methods in image processing only look for edge pixels in eight directions in eight neighborhoods around one edge pixel,while the actual edge direction can be any direction in 360 degrees in theory.Based on the principle of steerable operator,this paper uses a second-order fitting X-Y separation filter which contains the least basic filter,and then combines canny criterion with the principle of constant moment threshold selection to detect the edge of the image.Experimental results show that the proposed detection strategy is superior to other methods in effect.