基于互信息的立体匹配方法
|
摘要
介绍一种基于灰度互信息的立体匹配方法,用于从双目图中获得视差图。该方法将灰度互信息作为衡量输入图像对相似度的指标,先构建单像素点的代价,再在代价聚合环节加入邻域点的约束,使用多方向的扫描线优化来优化代价聚合函数。反复地进行优化过程,让输出的视差图渐渐地接近于真实情况,使用随机图作为迭代初始状态,迭代固定次数后输出。灰度互信息作为代价标准,使得该方法在代价的计算环节比一般的全局匹配方法更快,并对光照具有一定鲁棒性,领域点的约束让视差图更加平滑稠密。
This paper introduces a stereo matching method based on grayscale mutual information, which is used to obtain disparity map from binocular map. In this method, the gray-scale mutual information is used as an index to measure the similarity of the input image, and the cost of the single-pixel point is constructed. Then the constraint of the neighborhood point is added in the cost aggregation, and finally the multi-directional sweep line optimization is used to optimize the cost aggregate function. Repeating the optimization process is to make the output disparity map gradually close to the real situation. This method uses a random graph as the initial state of the iteration, iterates a fixed number of times and output. Gray-scale mutual information as a cost criterion makes the method faster in the calculation of the cost than the general global matching method, and has certain robustness to illumination. The constraint of the domain point makes the disparity map more smooth and dense.
This paper introduces a stereo matching method based on grayscale mutual information, which is used to obtain disparity map from binocular map. In this method, the gray-scale mutual information is used as an index to measure the similarity of the input image, and the cost of the single-pixel point is constructed. Then the constraint of the neighborhood point is added in the cost aggregation, and finally the multi-directional sweep line optimization is used to optimize the cost aggregate function. Repeating the optimization process is to make the output disparity map gradually close to the real situation. This method uses a random graph as the initial state of the iteration, iterates a fixed number of times and output. Gray-scale mutual information as a cost criterion makes the method faster in the calculation of the cost than the general global matching method, and has certain robustness to illumination. The constraint of the domain point makes the disparity map more smooth and dense.
引文
[1]Kaehler Adrian,Bradski Gary.Learning OpenCV3[M].O Reilly Media,2016.704-710.
[2]Zhang Kang,Fang Yuqiang,Min Dongbo,et al.The IEEE Conference on Computer Vision and Pattern Recognition(CVPR)[C].2014.1590-1597.
[3]Boykov Y,Veksler O,Zabih R.Fast Approximate Energy Minimization via Graph Cuts[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2001,23(11).
[4]Kolmogorov V,Zabih R.Computing Visual Correspondence with Occlusions using Graph Cuts[J].IEEE International Conference on Computer Vision,2001,I:508-515.
[5]Sun J,Zheng N-N,Shum H-Y.Stereo Matching Using Belief Propagation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(7).
[6]Sun J,Li Y,Kang S B,et al.Symmetric Stereo Matching for Occlusion Handling[J].IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2005,II:399-406.
[7]Yang Qingxiong,Yang Ruigang.Stereo Matching with Color-Weighted Correlation[J].Hierarchical Belief Propagationand Occlusion Handling,IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(3).
[8]Yang Qingxiong.A non-local cost aggregation method for stereo matching[C].IEEE Conference on Computer Vision and Pattern Recognition,2012.26.
[9]Yang Qingxiong.Stereo Matching Using Tree Filtering[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(4).
[10]Bishop Christopher M.pattern recognition and machine learning[M].London:Springer,2011.39-45.
[11]Egnal Geoffrey,Wildes Richard P.Detecting Binocular Half-Occlusions:Empirical Comparisons of Five Approaches[J].Pattern Analysis and Machine Intelligence,2002,24(8):1127-1133.
[12]范海瑞,杨帆,潘旭冉,等.一种改进Census变换与梯度融合的立体匹配算法[J].光学学报,2018,(02).
[13]周文晖,林丽莉,顾伟康.一种鲁棒的基于互信息的实时立体匹配算法[J].传感技术学报,2006,(04).
[14]李金凤.立体匹配算法的研究[J].黑龙江科技信息,2015,(27).
[15]罗大思,王进华.基于双目视觉的立体匹配算法研究[J].微型机与应用,2016,(20).
[2]Zhang Kang,Fang Yuqiang,Min Dongbo,et al.The IEEE Conference on Computer Vision and Pattern Recognition(CVPR)[C].2014.1590-1597.
[3]Boykov Y,Veksler O,Zabih R.Fast Approximate Energy Minimization via Graph Cuts[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2001,23(11).
[4]Kolmogorov V,Zabih R.Computing Visual Correspondence with Occlusions using Graph Cuts[J].IEEE International Conference on Computer Vision,2001,I:508-515.
[5]Sun J,Zheng N-N,Shum H-Y.Stereo Matching Using Belief Propagation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(7).
[6]Sun J,Li Y,Kang S B,et al.Symmetric Stereo Matching for Occlusion Handling[J].IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2005,II:399-406.
[7]Yang Qingxiong,Yang Ruigang.Stereo Matching with Color-Weighted Correlation[J].Hierarchical Belief Propagationand Occlusion Handling,IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(3).
[8]Yang Qingxiong.A non-local cost aggregation method for stereo matching[C].IEEE Conference on Computer Vision and Pattern Recognition,2012.26.
[9]Yang Qingxiong.Stereo Matching Using Tree Filtering[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(4).
[10]Bishop Christopher M.pattern recognition and machine learning[M].London:Springer,2011.39-45.
[11]Egnal Geoffrey,Wildes Richard P.Detecting Binocular Half-Occlusions:Empirical Comparisons of Five Approaches[J].Pattern Analysis and Machine Intelligence,2002,24(8):1127-1133.
[12]范海瑞,杨帆,潘旭冉,等.一种改进Census变换与梯度融合的立体匹配算法[J].光学学报,2018,(02).
[13]周文晖,林丽莉,顾伟康.一种鲁棒的基于互信息的实时立体匹配算法[J].传感技术学报,2006,(04).
[14]李金凤.立体匹配算法的研究[J].黑龙江科技信息,2015,(27).
[15]罗大思,王进华.基于双目视觉的立体匹配算法研究[J].微型机与应用,2016,(20).