Image Retrieval based on Color Histogram and DCT

This paper proposes a content-based image retrieval method based on color histograms and the discrete cosine transform (DCT). Color histogram characterizes an image by its color distribution, but the drawback of a global histogram representation is that spatial information (such as object location, shape, and texture) is discarded. Thus the images retrieved by using the color histogram only may not be desired even though they share similar color distribution. In our approach, each image is first transformed into the YUV space for the purpose of extracting the features based on color tones easily. Then, the histogram for each component (i.e., luminance Y, blue chrominance U, and red chrominance V) of the image is obtained, which can be served as the color feature of the image; furthermore, to compensate this inherent shortcoming of the color histograms, the DCT is applied to extract the texture and shape features from the Y component of the images. We have performed experiments on a database with 1000 images and the results show the effectiveness of our approach.