论文部分内容阅读
借鉴平稳小波变换的多尺度分析思想,结合模糊聚类均值法,提出了一种高鲁棒性的图像边缘提取算法.该算法利用平稳小波变换的位移不变性,将小波分解后的分量进行配准构成一像素的特征向量,然后利用模糊c-均值进行无监督分类,分割图像,最后用Canny算子提取图像边缘.用一系列附加不同强度的高斯白噪声图像测试了该算法的有效性.实验证明在图像受到较强噪声(如附加高斯白噪声)污染时,该算法仍可检测到较好的边缘效果,展现出良好的鲁棒性.
Based on the idea of multiscale analysis of stationary wavelet transform and combining with fuzzy clustering mean method, a robust image edge detection algorithm is proposed. By using the invariance of displacement of stationary wavelet transform, the wavelet decomposition component The feature vector of a pixel is quasi-constructed, then the unsupervised classification is performed by fuzzy c-means, the image is segmented, and finally the edge of the image is extracted by Canny operator.The effectiveness of this algorithm is tested by a series of Gaussian white noise images with different intensities. Experiments show that the algorithm can still detect better edge effect and show good robustness when the image is contaminated by strong noises (such as additional Gaussian white noise).