Abstract:

A stereo matching algorithm based on the hybrid optimization method is proposed to improve the low reliability of the disparity map obtained by the linear growth algorithm. Considering the calculation efficiency and the image reliability comprehensively, the proposed algorithm converts disparity matching into a multi-objective optimization problem. This optimization problem is solved by the pigeon-inspired optimization algorithm based on simulated annealing for realizing the adaptive adjustment of the disparity threshold and obtaining the optimal disparity value of the corresponding root point. The proposed hybrid optimization method overcomes the shortcomings of the local and the global optimization methods, which are vulnerable to initial values and have slow convergence. In addition, the filtering method is used to remove unreliable disparity for further improving the reliability of the disparity map. The simulation results show that the algorithm can get more depth informations, which enhances the reliability and the robustness of the disparity map calculated by the linear growth algorithm.

Key words: stereo matching, disparity map, self-adaptive adjustment, pigeon-inspired algorithm, simulated annealing algorithm