Stereo matching based on disparity propagation using cellular evolutionary neural networks

In this paper, we propose a novel stereo matching algorithm based on disparity propagation using cellular evolutionary neural networks (CEN). Most of previous works have drawbacks and advantages in accuracy, running time and scene types of image; however, our advantage is obtaining not exceedingly-high but satisfactory accuracy for various scenes with low computational cost. Our algorithm calculates initial disparities with a simple local method, and then propagates those disparities using CEN. The direction of propagation is controlled by a reliability map, which is created by checking left-right consistency of the initial disparity maps. We test our algorithm with the Middlebury stereo dataset, and experimental results show that our algorithm is able to produce more accurate disparities than common local and global methods for many types of scenes within just two seconds.