2018 · 12 citations · 8 references
ComputingComputer ScienceBinocular VisionComputer VisionMachine VisionDepth EstimationObject RecognitionScene ReconstructionStereo VisionDisparity ComputationDisparity EstimationDisparity Map EstimationImage MatchingMatchingStereo CorrespondenceStereo Correspondence AlgorithmsStereo Matching
This paper proposes a high-speed stereo matching algorithm (HSSM) for ultra-high resolution binocular images. By distinguishing the distances of objects in the acquired images, the size of the left and right images is determined. For example, for nearby objects, the binocular camera collects the left and right images that have large parallaxes and contain so many pixels, which consumes large resources in the process of matching. Therefore, a smaller size image is used to match close-range objects in the proposed algorithm. On the contrary, large-size images are used to match distant objects to ensure the matching accuracy of distant objects, and finally the result of hierarchical matching. Combine and get a disparity map of the left and right images captured by the binocular camera. This kind of matching method similar to image pyramid has been verified by our large number of comparative experiments. Compared with the previous binocular stereo matching algorithm, the binocular stereo matching ultra-high resolution image can be greatly guaranteed under the premise of ensuring accuracy. It reduces the time consumption in the matching process and is highly efficient.
8
Kaiming He, Jian Sun, Xiaoou Tang · IEEE Transactions on Pattern Analysis and Machine Intelligence · 2012 · 5.4K citations
Cross-Scale Cost Aggregation for Stereo Matching
Kang Zhang, Yuqiang Fang, Dongbo Min et al. · 2014 · 150 citations