In this paper, the hybridized feature descriptor has been utilized for the CBIR engines. The content based image retrieval (CBIR) systems are being popular among the internet users and they provide them the various ways to find the desired image contents from the online image databases. Content based image retrieval (CBIR) systems offer the query inputs in the form of text keywords, specifications, image properties, color patterns, texture patterns or image itself as the input query. The image based query systems are getting the popularity as they lower the complexity to enter the type of colors, color patterns, texture patterns, etc in the customized query inputs. The proposed model has been designed in the layered approach, where the image features are matched one after one to prepare the final image ranking. The proposed model outperforms the existing models because of its hybrid feature descriptor solution proposed by the amalgamation of the color and low rank features computed with feature fitness validation. The proposed model performance has been evaluated in the form of various performance parameters of accuracy, ranking/index building time, library lookup time, precision, recall, etc. The experimental results have proved the efficiency and robustness of the proposed model and proved it as the effective CBIR system.
Prof. Dr. Bilal BİLGİN