The traditional Canny operator uses the global threshold method, but when the gray of the input image's background and foreground change largely, the global threshold method will lose some weak edge. Concerning this issue, an improved adaptive Canny operator was put forward. Firstly, the image was divided into blocks according to the gradient variance. Secondly, the threshold of the sub blocks was obtained by Otsu method. Then, the threshold matrix was got by interpolation. Finally, an improved edge connection algorithm was proposed to extract the edge with the threshold matrix. The experimental results show that the improved Canny algorithm has not only a good anti-noise function, but also a very good precision.%传统的Canny边缘检测算子采用全局阈值选取方法,当图像的背景和目标灰度变化比较大时,采用全局阈值法将丢失一些局部边缘信息.针对该问题,提出了一种自适应求取动态阈值的改进Canny边缘检测方法.该方法利用图像的梯度方差作为判据对图像进行分块,继而对各子块运用最大类间方差法求得相应的阈值,并通过插值得到阈值矩阵,最后结合改进的边缘连接法对图像进行边缘检测.实验结果表明,该算法在具有良好的抗噪性能同时,具有很好的检测精度.
展开▼
