The Color Layout Descriptor (CLD), which captures the spatial distribution of colors in images or video frames, is created by partitioning the image into 8×8 blocks, deriving the average color of each block, and preforming discrete cosine transform (DCT) to the luminance (Y), the blue chrominance (Cb), and the red chrominance (Cr). This yields to three sets of 64 signal amplitudes, expressed by the DCT coefficients DCTY, DCTCb, and DCTCr. The DCT coefficients include DC coefficients and AC coefficients, i.e., coefficients that have zero frequency in both dimensions (the mean value of the waveform), and coefficients that have non-zero frequencies. These DCT coefficients are then quantized and zig-zag scanned.
<beach.jpg> mpeg7:YACCoeff5 "16 12 15 12 17" ; mpeg7:CbACCoeff2 "22 17" ; mpeg7:CrACCoeff2 "16 14" ; mpeg7:YDCCoeff "50"^^mpeg7:unsigned6 ; mpeg7:CbDCCoeff "34"^^mpeg7:unsigned6 ; mpeg7:CrDCCoeff "30"^^mpeg7:unsigned6 .
The first three coefficient properties are declared by multiple 5-bit property values of the type mpeg7:unsigned5, i.e., nonnegative integers in the range [0, 31]. Since the length of the coefficient values is indicated by a trailing number in MPEG-7, YACCoeff5 is declared by five mpeg7:unsigned5 values, while CbACCoeff2 and CrACCoeff2 are declared by two mpeg7:unsigned5 values. The 6-bit values of the last three coefficients, YDCCoeff, CbDCCoeff, and CrDCCoeff, are of the mpeg7:unsigned6 type, i.e., nonnegative integers in the range [0, 63].