Mathematical Foundations - Control of Color Imaging Systems: Analysis and Design - page 153

FIGURE 3.22 Error image.

functions and the weights of the basis function, the PSNR will decrease and the
compressionlosswillbemorevisible.

Example 3.43

To compress an RGB image, we transform the image fromRGB color space to YC b C r
color space [5]. The three channels are independently partitioned into 8 8 blocks

and the 64 64 sample covariance matrix is estimated for each channel. There are

64 basis functions per channel. Each basis function is an 8 8 image. To compress

the image, we use 8 of these principal component images corresponding to the 8
largest singular values. The original image, the compressed image, and the error
image are shown in Figures 3.23 through 3.25, respectively. The PSNR of the

FIGURE 3.23 (See color insert following page 428.) Original RGB image.