Abstract
This paper proposes a novel stereo matching method with a matching cost function learned from training data. Because the cost function includes a considerably large number of parameters required to select their values, it is nearly impossible to manually select the values. We employ an evolutionary algorithm to automatically optimize the parameter values for each dataset. Using Middlebury, KITTI 2012, and KITTI 2015 dataset, we compare the proposed stereo matching method with state-of-the-art stereo matching methods that can achieve real-time computation. Experimental results show that the proposed method outperforms the real-time state-of-the-art stereo matching methods.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have