摘要
针对传统蚁群算法的图像边缘检测定位不准、鲁棒性较差和易陷入局部最优解等问题.根据传统蚁群算法的边缘检测,同时融入了模拟退火算法,从而提出了改进的图像边缘检测算法.在蚂蚁运动过程中,找出每次迭代的最优解作为本次模拟退火过程的初始解,在初始解的邻域中随机产生一个新解.根据Metropolis抽样准则,判断是否接受此路径的改进,从而解决了传统蚁群算法随机性和正反馈的协调问题.通过与Canny算子和传统蚁群算法进行对比实验表明,提出的改进算法具有较好的边缘检测精度和较高的运行效率.
The traditional ant colony algorithm for image edge detection has the problems of inaccurate positioning,less robust and easily falling into local optimal solution. On the base of the conventional hybrid ant colony algorithm,an improved image edge detection algorithm is proposed with the simulated annealing algorithm. In the process of the ants movement,the optimal solution to each iteration is defined as the initial solution to this simulated annealing process,and a new solution generates the neighborhood of the initial solution randomly. According to the Metropolis sampling criterion,the path improvement schemes can be determined to solve the random and positive feedback coordination problem of the traditional ant colony algorithm. The comparative experiments with the Canny operator and the traditional ant colony algorithm show that the proposed algorithm has better edge detection performance and high operating efficiency.
出处
《兰州工业学院学报》
2016年第3期48-51,57,共5页
Journal of Lanzhou Institute of Technology
关键词
边缘检测
蚁群算法
模拟退火
最优路径
收敛
edge detection
hybrid ant colony algorithm
simulated annealing
optimal path
convergence
