基于contourlet模极大值的图像边缘检测

现存的基于空间域或小波变换域图像边缘检测算法只能有效检测出图像有限方向的边缘。本文提出在contourlet变换子带上搜索模极大值的图像边缘检测算法。通过适当地调节变换各尺度上的分解级数得到具有抛物线型尺度的基,另使其满足方向消失矩条件,则contourlet变换获得对图像中分片光滑边缘的最佳逼近。对原图像进行非下采样contourlet变换,然后对各子带进行阈值处理。由于contourlet变换子带不同于小波子带的全新特点,根据各方向子带代表的方向及其梯度方向,子带上各样本点与其梯度方向上相邻点比较模值即得变换系数模极大值。该算法不需经过复杂的连接模极大值点操作,亦可得到较高的边缘检测性能,较大地降低了计算复杂度。实验显示本文算法所提取的边缘更加逼近图像真实边缘,且具有较低的计算复杂度。

Image Edge Detection Based on Contourlet Modulus Maxima

Available image edge detection schemes based on the spacial domain or wavelet transform domain can capture only limited directional edges in images.A new edge detection scheme is proposed via searching modulus maxima in the contourlet subbands.The contourlet expansion satisfies the parabolic scaling by adjusting the number of directions at each scale appropriately.With sufficient directional vanishing moments,Contourlets achieve the optimal approximation rate for smooth contours and geometric structures in images.The input image is expanded by the nonsubsampled contourlet transform,and every coefficient in the subbands is thresholded by comparing against the adaptive thresholds.Based on the directions of each directional subband and its gradient,modulus maxima of the transform coefficients are obtained by comparing the modulus amplitudes of the samples and the neighborhood coefficients on the direction of the gradient.The scheme requires lower complexity without complicated link operation,while edge detecting performance is improved.The results show the proposed algorithm is superior to the image edge detection method based on wavelet modulus maxima by approximating better the edges of images and has lower computational complexity.