Deep stereo confidence prediction for depth estimation

We present a novel method that predicts a confidence to improve the accuracy of an estimated depth map in stereo matching. In contrast to existing learning based approaches relying on hand-crafted confidence features, we cast this problem into a convolutional neural network, learned using both a matching cost volume and its associated disparity map. As the size of the matching cost volume varies depending on a search range of stereo image pairs, we propose to use a top-K matching probability volume layer so that an input size for convolutional layers remains unchanged. Experimental results demonstrate that the proposed method outperforms the state-of-the-art confidence estimation approaches on various benchmarks.

1. Introduction

As one of the most important topics in computer vision, stereo correspondence has been actively studied over the last few decades. The disparity map (or depth map) obtained using stereo correspondence is widely used in many applications such as 3D reconstruction [1], object detection [2], and driver assistance system [3].