Multidimensional histogram equalization and modification
Author
McCollum, Anthony
Clocksin, William
Attention
2299/13799
Abstract
We describe an approach toward multidimensional histogram equalization for the enhancement of color and multispectral images, based on the histogram explosion algorithm. A development from first principles is presented. We show that in, the multidimensional case, one-dimensional histogram equalization becomes a scalar multiplication of a vector quantity. An image noise reduction scheme is devised using a model of multispectral noise behaviour that is analogous to the behaviour of local spatial noise. Practical results are presented indicating that the method reveals subtle image features and respects the tone of the originals more consistently than earlier methods.