On Comparative Performance Analysis of Color, Edge and Texture based Histograms for Content Based Color Image Retrieval

Image Retrieval is a system which extracts the relevant set of images and matches with query image from large collection of dataset. It is used in variety of domains including finger print identification, biodiversity information system, digital library, medical imaging etc. An effective and efficient system is required to improve its retrieval performance. In CBIR, the images are indexed according to image contents. Contents of images are color, shape and texture that are derived from images. The minimum distance between query and dataset images implies that the dataset image is similar to query image. In this paper, we study and compare different approaches used in CBIR and similarity measure taken for finding the similarity between two images. Strengths and weaknesses of these methods are observed using performance parameters such as precision, recall and similarity measures.