“…In this paper, we adopt the bag-of-features MPEG-7 [1,2,3,4,13,14,22,23,24] defined by the MPEG organization, which consists of color description (i.e., two color descriptors), texture description (i.e., two texture descriptors), and shape description (i.e., one shape descriptor). Among them, the descriptors relevant to image characteristics are as follows and the numbers of their features are shown in Table 1 To extract CLD features, an image on the RGB color space is partitioned into 8*8=64 blocks first; then, the average color of each block is calculated and converted into YCbCr color; next, each Y, Cb, Cr color space is transformed by 8x8 DCT, so that three sets of 64 DCT coefficients are obtained; finally, a zigzag scanning is performed with these three sets of 64 DCT coefficients, and CLD features are extracted with the specified length 12 (i.e., 6 coefficients from Y, 3 coefficients from Cb, and 3 coefficients from Cr).…”