Abstract
Currently, most image retrieval systems are based on low level features of color, texture and shape, not on the semantic descriptions that are common to humans, such as objects, people, and place. In order to narrow down the gap between the low level and semantic level, object-based content analysis, which segments the semantically meaningful object on images, is an essential step. This paper describes a novel image similarity measure approach for image comparison at object categories. It is not only suitable for images with single objects, but also for images containing multiple and partially occluded objects. In this approach, the contour of objects is extracted, and feature is obtained from contours. A machine learning categorization algorithm is used to predict the category of each of object-contour segments. The image is represented in a k-dimensional space, where k is the number of categories of objects in all the images. Each dimension represents information about one of the category. The similarity measure between two images is computed using Euclidean distance between images in the k-dimensional space. Experimental results show that this approach is effective, and is invariant to rotation, scaling, and translation of objects.
