Region and Object Based Image Retrieval Technique Using Textural and Color Expectation Maximization Method

Abstract

Users of commercial CBIR systems prefer to pose their queries in terms of key words. To help automate the indexing process, we represent
images as sets of feature vectors of multiple types of abstract regions, which come from various segmentation processes. With this representation,
we have developed an algorithm to recognize classes of objects and concepts in outdoor scenes. We have developed a new method for object
recognition that uses whole images of abstract regions, rather than single regions for classification. A key part of our approach is that we do not
need to know where in each image the objects lie. We only utilize the fact that objects exist in an image, not where they are located. We have
designed an EM-like procedure that learns multivariate Gaussian models for object classes based on the attributes of abstract regions from multiple
segmentations of color images. The objective of this algorithm is to produce a distribution for each of the object classes being learned. It uses the
label information from training images to supervise EM-like iterations.