[root]/Modules/IntensityNormalization/Testing
| Author | Changes | Lines of Code | Lines per Change |
|---|---|---|---|
| sylvain | 1 (100.0%) | 0 (-) | 0.0 |
ENH: adding a module to normalize the intensity of images. It is designed as a pre-processing step required for the Expectation-Maximization (EM) algorithm. The algorithm works as follows. First it computes the histogram of the input image, second it locates the mean value of the histogram, and finally it applies a normalization factor on the entire image. The parameters are the percentage of the pixel intensities to take into account and the regularization factor applied to the histogram. The rationale for not selecting the entire spectrum of pixel intensities is to remove the outliers in the image histogram. The regularization of the histogram is performed by an iterative averaging of the histogram bins. The significance of the regularization factor is a number of regularization steps. A regularization factor will smooth out the local minima in the histogram allowing for a better localization of the mean intensity at the cost of a larger processing time.
0 lines of code changed in 1 file: