The spectral distribution of the different points of a color image contains information that often is relevant for recognition purposes. For example, objects that cannot be distinguished in black-and-white images are easily discriminated when their color distribution is considered. Said another way, the problem of recognition is greatly improved when one considers the spectral distributions of the different points of the image. In addition, the color distribution of images sometimes contains information on the chemical composition of the objects present in the image.
Accordingly, in recent years a variety of proposals have been made to introduce the color information of images in pattern recognition tasks using optical correlation. Additionally, some works on filter design for color pattern recognition have been proposed for color image-processing techniques. Some are the generalization of well-known filters for monochromatic images, but others are specifically designed for color pattern recognition.
Online access to SPIE eBooks is limited to subscribing institutions.