This post is a continuation from Biomedical Image Processing – IV.
Contrast stretching (a.k.a. normalization) is very similar to histogram equalization, but it maps only a certain portion of the histogram to the full range. An example of this image enhancement technique is illustrated in Figure 17. If a and b are the lowest and highest pixel values available (e.g., 0 and 255, respectively, in an 8-bit gray scale image), and c and d are the lowest and highest current pixel values, then the mapping function works in the following manner, where Pout is the output pixel and Pin is the input pixel:
8-bit gray scale images
In an 8-bit gray scale image, any values of Pout below 0 or above 255 are set to 0 and 255, respectively.
Figure 17. Example of contrast stretching [http://www.dai.ed.ac.uk/HIPR2/stretch.htm].
Logarithmic and Exponential Operators
Logarithmic/exponential (a.k.a. “raise to power”) operators are used to enhance low/high pixel values, respectively, where a relative loss of information occurs in the high/low pixel values.