Novel Multi-objecitve Evolutionary Algorithm for Color Filter Arrays Design

Most digital cameras use a single sensor covered with a Color Filter Array (CFA) in order to reduce the size, complexity and cost. Now, more and more researchers have paid attentions to the representation of CFA since its crucial role in the process of reconstructing the full color image. The representation of CFA in frequency domain records all the frequency information of image mosaicked with the CFA, thus provides a theoretical approach to design the CFA. However, almost all the existing CFA design methods in frequency domain have limitations. In this paper, we propose a new automatic CFA design method in frequency domain. We choose the optimal frequency structures which satisfy all the design principles through a novel coded multi-objective evolutionary algorithm (MOEA) at first. Then we convert the parameters optimization of the given frequency structures into a constraint single optimization problem, and we solve it by converting it into a multi-objective optimization problem without constraints which is hard to handle. So a MOEA with novel selection and crossover-mutate strategies is proposed to solve the model. The experimental results show that the proposed CFA design method has more advantage when compared with the other existing method.