P06641 Algorithm Comparison Results
Below are some of the results found using three test images and several different values for B and C0. For each image, the iterations, computed B, computed C0, and root mean square error(RMSE) was recorded.
(Note: 501 and 100 are the maximum iterations allowed for Algorithm 2 and 1 respectively.)
Results from algorithm comparison, save this document as a excel sheet(.xls) and then open from your desktop.
The Mean Square Error of the resulting images were found to determine which was better since visual inspection was nearly impossible. It was found that Algorithm Two not only resulted in a smaller Mean Square Error but also defogged the image in a shorter amount of iterations then Algorithm One. Algorithm Two however did run into a few problems when the fog was varied in certain images such as lenna and the baboon but broke out after one hundred iterations with a successfully defogged image whose Mean Square Error was better then the image formed by Algorithm One. This same affect was seen while testing Algorithm One as well except its Mean Square Error was not a good as the one found using Algorithm Two. In B testing on Algorithm Two, it was found that a resulting image after around 5 iterations is enough to successfully defog the image.
The Mean Square Error for Algorithm Two was always much smaller then Algorithm Ones, where it was usually around five or six. From these findings, it is safe to say that Algorithm Two is the better of the two algorithms when it comes to overall picture quality as well as runtime.