文章基本信息

标题：ROTATION AND SCALE INVARIANT FEATURE EXTRACTION FOR MRI BRAIN IMAGES
本地全文：下载
作者：NAVEEN KISHORE GATTIM ; V RAJESH
期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2014
卷号：70
期号：1
出版社：Journal of Theoretical and Applied
摘要：Texture classification of images with varied orientations and scale changes is a challenging and considered to be important in image analysis. Feature extraction can be used to increase the efficiency of texture classification using log polar wavelet energy signatures. For the image to be rotation and scale invariant two major steps are applied which involves applying log polar transform and adaptive row shift invariant wavelet transform. Log polar transform eliminates the rotation and scale effects and causes a row shifted log polar image, which undergoes adaptive row shift invariant wavelet transform to remove the row shift effects. Finally they obtained output wavelet coefficients are rotation and scale invariant. The complexity of O (n log n) is efficient with adaptive row shift invariant wavelet packet transform. From the log polar wavelet energy signatures a feature vector is generated which are extracted from each sub band of wavelet coefficients. In the experiments the features are extracted for images considering different orientation and scale changes and simultaneously experiment is simulated for few wavelet families. The experiment results show the efficiency of few wavelets in extracting the features of a given image. The overall accuracy rate for this approach is 87.05 percent representing that the extracted energy signatures are effective rotation and scale invariant features.
关键词：Log polar transform; Row shift invariant wavelet transform; Rotation and scale invariance; Feature extraction.