摘要
针对边缘检测得到的边缘通常存在不连续现象,基于改进的蚁群算法理论,在Canny算子得到边缘图的基础上,提出了一种有效的边缘连接方法。该方法利用能见度矩阵设置像素点的初始信息激素并调节信息激素的更新,根据像素点梯度幅值和梯度方向来确定引导函数,从而减小了在算法运行初期定位错误的概率,提高了迭代过程中信息更新的准确性。仿真实验结果表明,该方法能够有效地改善边缘检测后的边缘不连续现象,满足边缘检测的轮廓封闭性要求。
The edges obtained by the traditional edge detection are usually not continuous. An effective edge linking method was proposed by using the edge graph obtained through the Canny operator based on the improved ant colony algorithm. The method utilized visibility matrix to set up initial pheromone values and controlled pheromone update. Then, guiding function was determined by gradient amplitude and gradient direction, thus could reduce the early positioning error probability and improve accuracy of the information renew during the iterative edges and meet the needs process. of edge Simulation results show that the method can efficiently link gaps between linking demand in some applicaion cases.
出处
《电光与控制》
北大核心
2012年第10期42-45,69,共5页
Electronics Optics & Control
基金
国家自然科学基金资助项目(60805015)
关键词
图像处理
边缘检测
蚁群算法
能见度矩阵
image processing
edge detection
ant colony algorithm
visibility matrix
