Spatial feature extraction and image retrieval based on entropy

According to the color spatial feature of an image, a new algorithm based on entropy is proposed. It introduces color spatial distribution entropy (SDE) as the spatial descriptor of an image, which describes the distribution of pixels with the same color in the image. SDE is also rotation, translation and scale invariant. Integrated with the human vision and the characters of entropy, improvements are made on the new partition method and the weight function. Based on SDE and color histogram, two retrieval methods are proposed to compute the similarity of two images. Experiments show that better results can be achieved than color histogram.