Monochrome image representation and segmentation based on the pseudo-color and PCT transformations

Monochrome image representation and segmentation based on the pseudo-color transformation and principal components transform (PCT) are presented in this paper. The HLS family of color models is employed to map a monochrome image into a new multidimensional color space where image features are enhanced by color representation. An optimal decomposition is then applied using the PCT transformation of the color space, in which image features are better defined and the automatic image segmentation is easily performed using the PCT-guided median splitting. Attempts are also made to compare the proposed segmentation with the fuzzy c-means (FCM) clustering in terms of the quality and computational complexity involved in segmentation. Results from mammograph and MRI image representation and segmentation are presented.