Comparative study of color histogram based bins approach in RGB, XYZ, Kekre´s LXY and L′X′Y′ color spaces

Content based image retrieval is one of the popular image processing field having huge scope for researchers to work out the novel ideas that will produce the promising results. Core phases of CBIR where the research contribution is desired, are feature extraction based on image contents, Similarity measures used for comparison and the performance evaluation. This paper explores each of these phases by contributing into each of them. Feature extraction phase is based on the 8 bins approach that actually works for dimension reduction. Bins are the feature vector components acquiring the image contents. This approach is dealing mainly with color and texture contents of the image. Texture contents are extracted in the form of statistical moments whereas color contents and their role in the system is evaluated by using four color spaces namely, RGB, XYZ, LXY, and L´X´Y´ color spaces. Three similarity measures Euclidean distance, Cosine correlation distance and Absolute distance are used to work out the comparison of query and database image features. Performance of bins approach in all four color spaces is evaluated by three parameters PRCP (Precision Recall Cross over Point), LS (Longest String), and LSRR (Length of String to Retrieve all Relevant images).