Global Feature Properties by Image Modifications and Variations

In this paper, we examine the properties of some typical instance of global image features to provide clues to image retrieval by efficient selection of them as content. It is performed by calculating the feature difference between images including variously transformed images. Image data might be caused image variations by typical modification or deterioration originated from image collection process. We intend to classify the transformed images as identical group and investigate the appropriate kind of feature which is robust against the image variation and difference measuring method distinguishable from other group. The result can be applied for image identification against environmental modifications or identical photographic image grouping working with local feature.