A New Approach for Color-Based Object Recognition with Fusion of Color Models

In real images, color recognition is a challenging problem because of natural variations in light and shadows. Most existing color models have limitations in distinctive representation of colors in real scenes. Here, a linear combination of multiple color models is assumed. The approach makes color edges in the target as main character for recognition. It combined multiple color models to make the recognition stable under varying imaging conditions. A proper weighting scheme to combine color spaces is obtained with the mathematical selection model of Markowitz, and the corresponding weights are obtained by multi-objective optimization. Effectiveness and reliability of the proposed method is demonstrated through various natural scene images.