Abstract
In deep learning-based local stereo matching methods, larger image patches usually bring better stereo matching accuracy. However, it is unrealistic to increase the size of the image patch size without restriction. Arbitrarily extending the patch size will change the local stereo matching method into the global stereo matching method, and the matching accuracy will be saturated. We simplified the existing Siamese convolutional network by reducing the number of network parameters and propose an efficient CNN based structure, namely adaptive deconvolution-based disparity matching net (ADSM net) by adding deconvolution layers to learn how to enlarge the size of input feature map for the following convolution layers. Experimental results on the KITTI2012 and 2015 datasets demonstrate that the proposed method can achieve a good trade-off between accuracy and complexity.
Highlights
Three-dimensional scene reconstruction is important in autonomous driving navigation, virtual/augmented reality, etc
Binocular vision camera converts vision into a coo Stereo matching [3] between the left and the right images can be used to calculate theStereo disparity
We propose a typeat ofthe neural network with deconvolution layers patch by deconvolution layers instead of using naive up-sampling methods and obtively expand the size of the input patches by referring to the GA-Net feature extraction
Summary
Three-dimensional scene reconstruction is important in autonomous driving navigation, virtual/augmented reality, etc. 1. Binocular vision camera converts vision into a coo Stereo matching [3] between the left and the right images can be used to calculate theStereo disparity. Thematching regional minimization-based matching assumes that the disparities pixels a texture/color can be modeled algorithms by a linear function. Typical are By cross-based local st disparities that are obtained from local minimization-based matching, the disparities of orthogonal [4], stereo matching with improved the pixels in eachintegral region can images be modified by local regression [6]. Theofboundary e parameters [14,15], andthe the other is to use CNNs to extract features blocks in the left and the right images for further matching, a kind of network that is usually because the global energy function cannot deal with the local texture a simpler than the former.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have