基于Contourlet变换和Canny算子的图像边缘检测方法

Edge Detection of Image Based on Canny Operator and Contourlet Transform

现存的基于空间域或小波变换域的图像边缘检测算法只能有效检测出红外图像有限方向的边缘。由于这些算法没有充分利用邻域的信息,因此在边缘较为复杂的区域边缘检测结果中会出现较大的偏差。针对这一问题,提出了一种基于Contourlet变换和Canny算子的边缘检测算法,首先对原始图像进行Contourlet变换,然后利用各方向子带的方向信息及其梯度方向信息,对各个尺度进行边缘检测,最后通过逆变换,得到图像的边缘图像。实验结果表明了新算法可以提高边缘检测和保持的能力,有较强鲁棒性。

Available image edge detection schemes based on the special domain or wavelet transform domain can capture only limited directional edges in infrared images. Some mistakes may be made in the complex region for the incomplete information provided by horizontal and vertical adjacent pixels. To deal with the problem, The Canny algorithm based on the Contourlet transform is implemented. The input image is expanded by the Contourlet transform. According to the directions of each directional sub-band and its gradient, the edges of in every scale are detected using the edge detection method. Finally, the edges of image are reconstructed by inverse Contourlet transform. The experimentation proves the new algorithm has advantage in the ability to detect edges and keep edges.