结合Powell-RWACO的图像边缘提取算法*

针对蚁群算法在图像边缘提取中经常出现收敛速度慢、检测精度低、停滞等问题，提出一种结合Powell法的排序加权蚁群(Rank Weighted Ant Colony Optimization, RWACO)图像边缘提取算法。该算法将RWACO算法与Powell法相结合，利用RWACO算法进行全局优化，然后将全局最优值作为Powell法的初始点进行局部优化。实验结果表明，该算法兼顾了全局优化和局部优化的优点，与蚁群算法和Canny算法相比，明显提高了图像边缘精度，计算效率比蚁群算法提高了两倍多，并克服了其停滞等缺点，能够高效地检测出图像的边缘。从而验证了该算法的可行性，对今后的图像边缘检测具有参考价值。

Image edge extraction combined with Powell-RWACO algorithms

For using Ant Colony Optimization to extract image edge, there is always slow convergence, inefficiency and stagnation, so presenting a image edge extraction combined with Powell and rank weighted ant colony optimization (RWACO) algorithms. The algorithm was the combination of RWACO and Powell. Global optimization algorithm was completed by RWACO, then the global optimum value as the initial point of Powell began to local optimization. The experimental results show that the algorithm combines global optimization and local optimization advantages, Compared with the ant colony algorithm and Canny algorithm, the precision improves significantly, computational efficiency increases more than twice than the ant colony algorithm, and it also overcomes shortcomings, such as slow convergence, stagnation. So it can effectively detect image edge. The feasibility of the algorithm is verified, and it has important reference value on image edge extraction.