MSDC-Net: Multi-Scale Dense and Contextual Networks for Stereo Matching

Zhibo Rao, Mingyi He, Yuchao Dai, Zhidong Zhu, Bo Li, Renjie He

Abstract: Disparity prediction from stereo images is essential to computer vision applications such as autonomous driving, 3D model reconstruction, and object detection. To more accurately predict disparity map, a novel deep learning architecture (called MSDC-Net) for detecting the disparity map from a rectified pair of stereo images is proposed. Our MSDC-Net contains two modules: the multi-scale fusion 2D convolution module and the multi-scale residual 3D convolution module. The multi-scale fusion 2D convolution module exploits the potential multi-scale features, which extracts and fuses the different scale features by Dense-Net. The multi-scale residual 3D convolution module learns the different scale geometry context from the cost volume which aggregated by the multi-scale fusion 2D convolution module. Experimental results on Scene Flow and KITTI datasets demonstrate that our MSDC-Net significantly outperforms other approaches in the non-occluded region.