Average Common Submatrix: A New Image Distance Measure
A new information-theoretic distance measure for images is proposed. The measure is based on the concept of average common sub-matrix by considering the pixel matrices associated with the images. An algorithm to compute such a value is described, and its computational complexity analyzed. Experimental results show that the measure is able to discriminate images by correctly reflecting human perception. Furthermore, comparison with state-of-the-art information-theoretic measures, points out that the new measure outperforms these measures in terms of retrieval precision.
Keywordsimage retrieval similarity measure pattern matching