International Journal of Engineering Research and Development

Color is the most important properties for object discrimination. It is an object component that adds a new dimension to the machine vision. The color images on low devices are displayed having limited number of colors called as color palette reduction to cut down the storage requirements. In this paper we investigate the problem of averaging aggregation functions to select the appropriate function using neural network to reduce the color of an image. The neural network represents the large set of input features with a smaller set of rules to predict the exact penalty of the aggregation function to have good approximation. Then train the neural network with the image samples so that it creates the set of rules to predict the color values using aggregation functions as the features and it comes to an end with some experimental results and conclusions.