Image Processing Basics - Digital Signal Processing: Fundamentals and Applications - page 695

8

7

6

5

4

3

2

1

0

0 1 2 3 4 5 6 7

Scale level

FIGURE 14.12

Histogram for the equalized image in Example 14.5.

As we can see, the pixel levels in the equalized image are stretched to the larger scale levels. This technique
works for underexposed images.

Next, we apply image histogram equalization to enhance a biomedical image of a human neck in
Figure 14.13A , while Figure 14.13B shows the original image histogram. (The purpose of the arrow
in Figure 14.13A will be explained later.) We see that there are many pixel counts residing at the lower
scales in the histogram. Hence, the image looks rather dark, and may be underexposed.
Figure 14.14A and Figure 14.14B show the equalized grayscale image using the histogram method
and its histogram, respectively. As shown in the histogram, the equalized pixels reside on a larger
scale, and hence the equalized image has improved contrast.

14.2.2 24-Bit Color Image Equalization
For equalizing the RGB image, we first transform RGB values to YIQ values since the Y channel
contains most of the signal energy, about 93%. Then Y channel is equalized just like the grayscale
equalization to enhance the luminance. We leave the I and Q channels as they are, since these contain
color information only and we do not equalize them. Next, we can repack the equalized Y channel
back to the YIQ format. Finally, the YIQ values are transformed back to the RGB values for display.
Figure 14.15 shows the procedure.