基于图像配准的混合遗传FCM算法研究

针对目前图像变化检测的相关研究,提出一种新的算法:基于SAR图像配准的混合遗传FCM算法.算法主要分为4个步骤.第一步,利用Harris算法和SIFT算法对两幅图像进行匹配,证明它们是同源不同时相的图像.第二步,利用两种不同变化检测方法提取初步差异图像.第三步,利用PCA方法对差异图像进行降维处理.第四步,利用混合遗传FCM算法对特征矢量空间进行分类,并将分类结果与参考差异图像进行比较,获得变换信息.采用渥太华地区的部分图像作为检测算法的性能的数据库.获得的结果与FCM算法相比较,结果表明,提出的算法具有最高的全局正确率98.10％,算法效果更佳.

A HYBRID GENETIC FCM ALGORITHM RESEARCH BASED ON IMAGE REGISTRATION

A hybrid genetic FCM algorithm based on SAR images registration is proposed in this paper in view of the present researches of the image change detection. This proposed method is divided into three steps. In the first step, Harris algorithm and Sift algorithm are used to match different images, proved that they are the homologous images from same region achieved at different time. In the second step, with the using of two change detection methods, the primarily difference image is obtained. In the third step, PCA method is used for feature extraction and dimension reduction. In the fourth step, the feature vector space information is divided into two classes based on hybrid genetic FCM algorithm. The change information is achieved by comparing the classification results and reference difference image. This method takes the parts of image of Ottawa area as data set for the performance evaluation. Compared with other FCM method, the results show that the change detection accuracy of the proposed algorithm reaches 98.10%, so it is better than other algorithms.