Weighted Region Ranking

There are a number of non-linear methods for enhancing the details in images. One approach I have investigated is a variation of ahe called weighted region ranking. Here a circular neighborhood about each point is searched to count the number of points lighter and darker than the central point. This gives us the rank of each pixel which we display as an output image. The size of the search region and the amount of histogram smoothing prior to ranking control the amount of enhancement performed. These parameters are given as command line arguments to the program wrr. (see code)

Command Usage

wrr [options] infile outfile
    [-d]    Print debugging information
    [-s #]  StdDev of weighting Gaussian (8)
    [-r #]  Radius of weighting region (12)
    [-b #]  StdDev of blurring Gaussian (1)

Example

wrr -s 2 -r 3  input.im wrr0.im
disp input.im wrr0.im

wrr -s 4 -r 6  input.im wrr1.im
wrr -s 8 -r 12 input.im wrr2.im
disp wrr1.im wrr2.im

wrr -s 12 -r 18 -s 1  input.im wrr3.im
wrr -s 12 -r 18 -s 20 input.im wrr4.im
disp wrr3.im wrr4.im