摘要
提供了对图像边缘的两步检测法。先利用样条小波变换模极大值法对图像进行处理,接着利用Otsu的阀值法把模值分为边缘点,非边缘点和待选边缘点。对待选边缘点进行Gabor小波分解,再对其系数进行单阀值处理,就可得到不同尺度下较精确的图像边缘。用小波变换对图像进行边缘检测可以很好地抑制噪声,但其缺点是边缘存在不连续的现象。提供的方法保留了小波变换的优点又进一步提高了检测的精确度和图像边缘的连续性。
A new 2-step method for solving the edge detection problem is presented. The magnitude is extracted by the B-spline wavelet-based edge detector. The image pixels as edge point, candidate edge points or edge points are classified by the double thresholding using extended Otsu＇ thresholding method. Using this method the continuity of the wavelet-based edge detection is improved.
出处
《成都信息工程学院学报》
2007年第2期195-197,共3页
Journal of Chengdu University of Information Technology
