蚁群算法实现的图像边缘连接

Implementation of edge linking by ant colony algorithm

针对金相晶界重建问题,提出了一种基于蚁群算法的图像边缘断点连接方法。以原图像和传统边缘获取方法得到的边缘为基础,进行边缘的端点分析,并在端点处建立信息素的高斯分布模型,从而促使蚂蚁向端点方向移动;综合考虑了像素间相似性、邻域差和边缘走向三个因素来确定启发式引导函数,使蚂蚁沿着真正的边缘行走。实验表明该算法速度较快,补偿的边缘能够反映原图像边缘信息,是一种有效的边缘断点连接方法。

The study presented an ant colony algorithm to implement edge linking.The proposed approach was based on original image and the edge image gained by traditional approaches to analyze the endpoints,and then set up pheromone values with Gaussian distribution around the endpoints in order to make the ants move to the endpoints faster.The visibility of the paths was determined by pixel similarity,neighboring difference and the direction of the edges synthetically,which made the ants move along the real edge pixels.The experimental results indicate that the proposed edge linking approach is efficient and good at compensating the broken edges.