Abstract
Various domains such as medical science, forensics science and education etc. are generating lot of images on daily bases. As a result of these content generation large image databases are available. These databases are considered as very helpful such as suspects can be searched from forensics database, similarly medical image database can be utilized for the diagnosis purposes. However, proposer management of these databases like, storing and retrieving of images is the demand of the day. Relevant content searching from these databases is a difficult task, however content based image retrieval (CBIR) playing a very important role for searching the relevant contents from these large databases. But this approach is facing some issues. One of the famous issues of CBIR is to describe the image in terms of as feature. This research work aimed is to present a new scheme of image representation by combing the texture and color signature to increase the accuracy of CBIR. Color signatures are generated through Scalable Color Descriptor (SCD) while texture feature are extracted by Edge Histogram Descriptor (EHD). The proposed technique is assessed by testing on the coral image data set and validated by comparing the results with other CBIR approaches. (C) 2016 The Authors. Published by IASE. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).