Abstract:-Weight Co-occurrence based Integrated Color and Intensity Matrix for CBIR

Friday, 26 August 2011 09:45 administrator
Print PDF
Abstract—This paper proposed a new technique; weight co-occurrence based integrated color and intensity matrix (WCICIM) for content based image retrieval. The proposed image indexing technique introduced a feature that utilizes color and texture properties of the image. Hue, saturation and intensity (HSV) color space is the good representation to extract color and texture information. In this paper advantages of HSV space are utilized in calculating color and intensity weights for each pixel. We construct the weight matrixes by applying suitable weights function to decide the contribution of color and intensity for each pixel. WCICI matrix is constructed as the combination of 4 correlation matrixes (color - color, color-intensity, intensity-color and intensity-intensity) based on neighboring pixel variations in weight matrixes with the help of quantized HSV space. Integrated color and intensity co-occurrence matrix (ICICM) based image retrieval technique does not consider the weight correlation. Proposed image retrieval feature with the combination of ICICM feature improves the retrieval performance significantly. The proposed technique results are compared with combination of wavelet decomposition and gradient vector features [8], motif co-occurrence matrix [9], ICICM [10], color-texture and color-histogram based image retrieval system (CTCHIS) [13], color correlogram [14] based CBIR techniques
Download this file (Weight Co-occurrence based Integrated.pdf)Weight Co-occurrence based Integrated.pdf[Megha Agarwal, R.P. Maheshwari Indian Institute of Technology Roorkee–247667, Uttarakhand, India]2358 Kb
Last Updated on Thursday, 20 October 2011 10:14