基于类间方差的图像Canny边缘提取改进

An improved algorithm for adaptive Canny edge detection based on otsu algorithm

到目前为止,在边缘检测方面,Canny算子因为它优良的边缘检测特性,应用在很多方面。然而它也有局限性,传统Canny算子高低阈值是通过人工凭借经验来选取,且低阈值固定是高阈值的一半,这就使得它在一些特殊领域无法应用,且实际中图像易受光照、杂散光以及其他各种噪声的干扰。因此针对传统Canny算子阈值确定的困难性,针对性地提出了利用双重Otsu方法确定高低阈值的方法,此方法利用两次Otsu方法分别确定高阈值和低阈值。

So far,in terms of edge detection,Canny algorithm is widely used in many aspects because of its excellent edge detection feature. However,it also has its limitations. For example,the threshold value of the traditional Canny algorithm is selected by people and the low threshold is constant the half of the high threshold. So this leads to the useless in some special area. In fact,the picture is easily influenced by sun light,stray light and some other noise. To overcome this difficulty,the paper proposes method that selecting threshold by double Otsu method.