Local pixel analysis in color constancy to improve feature extraction process

Abstract

The feature extraction process on a digital image uses a digital image matrix value. Digital image matrix has position information in the index of columns and rows, and the value of the degree of color within a given range of values. The 8-bit digital image has a 256 degrees color rating ranging from the lowest value 0, and the highest value is 255. Based on the combination of position information and the degree of color, more information can developed, such as information on the shape of the object. Gray images have a single dimension color. This paper used the process of feature extraction area used as the basis of information to find the resemblance of the object on two images. Feature will extracted from two digital images to compared, the image converted into gray image to speed up the computation process. The color space changes based on the base color of the RGB color space. A digital camera has an auto white balance feature that can affect the color consistency of an object in a digital image. This may affect the feature extraction process based on the color degree value. This paper proposed a method to get a consistent color configuration. The color difference value is acquire from the analysis of similar feature to the specified size. Color consistency in the image expected to improve accuracy in the process of recognizing objects based on features