改进的LOG算子与小波变换融合的边缘检测方法

摘要

传统的LOG算子在高斯滤波方差的选择上不具备自适应能力,方差的大小会直接影响到去噪和边缘保持效果.针对此问题给出了一种改进的LOG算子方法,并对原图像分别采用改进的LOG算子和小波变换两种方法进行边缘提取,最后将这两种方法检测出来的边缘图像进行融合.实验表明,融合后的图像结合了两种检测方法的优点,有效地提高了边缘检测准确性,获得了比较理想的边缘检测效果.%The traditional LOG operator does not have the adaptive ability itself in the variance of Gaussian filtering. The variance directly affects the de-noising and the edge preserving effect. Aiming at on the problem, a new kind of LOG operator is proposed. It detects the edge of original image by the means of improved LOG operator and wavelet transform respectively, and then gets a new image by fusing the two edge images. It is shown that the fusion method contain the advantage of two detecting methods and obtain more perfect edge image.