一种改进Canny算子的图像边缘检测算法

Image Edge Detection Algorithm Based on Improved Canny Operator

在移动机器人自主作业时,环境中往往存在障碍物,路径规划避障时要进行动态目标检测.Canny边缘检测算法可以与众多动态目标检测算法相结合,提高目标检测的效果.但是传统Canny边缘检测存在着自适应性不强,边缘检测可能不连续,或者检测虚假边缘的现象.本文提出了一种优化Canny边缘检测算法,通过改进的自适应中值滤波来预处理图像,对算法效率及对噪声点的处理做出了优化,紧接着增加梯度计算方向,最后结合改进的大津阈值分割法,提出了三阈值分割法代替原始的阈值分割法使图像边缘信息更加完整准确.仿真结果表明,该算法在边缘检测准确率上对比传统Canny边缘检测,Sobel算子与较新改进算法均有20%左右的提升,该算法优化了传统算法检测的连续性和准确率.

When mobile robots operate autonomously,obstacles often exist in the environment.Dynamic target detection should be carried out in path planning to avoid obstacles.Canny edge detection algorithm can be combined with many dynamic target detection algorithms to improve the effect of target detection.However,traditional Canny edge detection has the phenomenon of weak adaptability,edge detection may be discontinuous,or false edge detection.This paper proposes a optimal Canny edge detection algorithm,through improved adaptive median filtering to preprocess image,the algorithm is efficiency and handling of the noise made the optimization,and then increase the gradient direction,finally combining with the improved otsu threshold segmentation method,puts forward the three threshold segmentation method instead of the original threshold segmentation method to make the image edge information more complete and accurate.Simulation results show that the edge detection accuracy of the proposed algorithm is improved by about 20%compared with the traditional Canny edge detection,Sobel operator and relatively new improved algorithm,and the proposed algorithm optimizes the continuity and accuracy of the traditional algorithm.