基于多尺度的Canny边缘检测算法研究

Research on Canny Edge Detection Algorithm Based on Multiscale

为了解决传统边缘检测算法检测精度不高、稳健性不好和抗噪能力差等弱点,文章提出了一种基于多尺度的Canny边缘检测算法,该算法充分利用了小尺度影像和大尺度影像的优点,既能检测出图像边缘的整体轮廓,又能检测出图像细节。并运用MATLAB编程与几种经典边缘检测算法做了详细对比,采用ROC曲线进行了精度定量评价分析。实验结果表明,基于多尺度的Canny边缘检测算法检测结果更清晰,稳健性更好,而且抗高斯噪声能力也相当理想,能够有效地提取图像边缘信息。

In this paper ,a multiscale edge detection algorithm based on Canny i s proposedcomings of the traditionai edge detection algorithms ,such a s low detecti on accuracy ,poor immunity. The algorithm ^akes full advant^age of small scale images and large scale images. Iwhole contour of image edge ,but also detect the detaH of image. And it i s compared with several ction algorithms in detaH through MATLAB.The quantitative analysis of the accuracy was carricurve. Experimental results show that ,the multi-scale Canny edge detection algorithm is more accurate and robustcompare with classical edge detection algorithms. And the ability to resist Gauss noise s goodinformation effectively.