Hybrid optimization of SA-GA depth map extraction base on second-order prior model

Abstract

This work deals with the depth information estimation of 3-D surface in a series of images from different shot angle.The proposed method used a new second-order prior model to model the smooth 3D surface information which cause intractable(non-submodular) optimization problems;improved optimization efficiency by using the hybrid optimization strategy of Simulated Annealing(SA) algorithm and Genetic Algorithm(GA).Experimental results demonstrate that the second-order prior model is a better model of typical scenes than the first-order prior one and the performance of the hybrid algorithm outperforms SA and GA alone.