基于余弦角距离的蚁群边缘检测算法

Ant Colony Algorithm for Image Edge Detection Based on the Cosine Angle Distance

提出一种基于余弦角距离的蚁群边缘检测算法,该算法利用改进的Sobel算子来计算梯度值,综合像素的灰度、梯度、领域特征进行特征提取,以余弦角距离为半径进行聚类,同时通过设置初始聚类中心、启发式引导函数和信息激素提高聚类速度.实验表明该算法优于Sobel、Canny算子和基于欧氏距离的基本蚁群分割算法,是一种有效的方法.

This paper presents a method of edge detection using the ant colony algorithm based on the cosine of the angle distance. The method adopts an improved Sobel operator to compute the gradient and extract features, inclu- ding pixel grayscale, gradient and domain feature, and replaces the Euclidean distance by the cosine angle for the clustering radius distance. At the same time, the setting of the initial clustering center, the heuristic guide function and the pheromone can enhance the clustering speed. The experiment shows that the algorithm is an effective meth- od. It is better than which using the Sobel, Canny operator and the basic ant colony segmentation algorithm.