Pipelined Technique for Image Retrieval Using Texture and Color

Content Based Image Retrieval (CBIR) is an active research field for the retrieval of relevant image from the dataset. Various features like color, texture, shape etc. and their combinations are being used for this purpose. This paper proposes the combination of texture and color features in a pipelined manner for the fast and efficient retrieval as compared to the state-of-art approaches. The pipelined approach has 2 phases. In phase one, texture feature is used for comparing the query image with the dataset to rank the images according to similarity as output. Top 25% images of the output are given as input to the second phase where the query image is again compared according to the color feature to further rank the images according to similarity as final result. Experimental results show that the proposed method is 1.5 times faster while maintaining the precision when compared with other state-of-art approaches.