Colors Reduction In Images Using Neural Networks (Kohonen)

This research uses a method based on Self-Organizing Feature Maps (SOFM) neuralnetwork. The (SOFM) is competitively trained according to the kohonen learningalgorithm.After training, the output neurons of the (SOFM) define proper classes of colors thatbest define the image by using Manhattan distance instead of the traditional Euclideandistance. Next each pixel is compared with resulted colors. Then each pixel is redrawnaccording the new colorsIn order to reduce the size of the image a transformation was made from (True Type) to (8Bit).Also an enhancement has been made on the resulted image to hide some side affectsresulted from the color reduction operation.The research was applied on colored and Gray level images.