首页 > 论文 > 光学学报 > 38卷 > 12期(pp:1215006--1)

自适应权值的跨尺度立体匹配算法

Weight-Adaptive Cross-Scale Algorithm for Stereo Matching

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

现有的多尺度立体匹配算法对各尺度的代价函数采用相同权值,而忽略了各尺度层对整个匹配代价的不同影响,增加了误匹配点。针对此问题,提出了自适应权值的跨尺度立体匹配算法框架。采用统一的代价聚合函数框架在不同尺度上进行代价匹配,并提出利用各像素窗口的信息熵作为不同尺度下匹配代价对整个匹配代价的影响因子;同时为了保证不同尺度下同一像素的代价一致性,在代价函数里加入正则化因子。本文算法框架可以应用在利用多尺度进行代价匹配的算法上,并使原有算法的准确率和稳健性得到提高。基于本文算法框架,分别采用不同代价聚合函数在Middlebury数据集上进行测试。为保证测试的公平性,各算法均未进行后续的视差求精步骤,实验表明,本文算法有效地提高了多尺度立体匹配的准确率和稳健性。

Abstract

In the existing weight-adaptive cross-scale algorithms, the same weight for the cost function of each scale is adopted, the different influence of each scale layer on the whole matching cost is missing, and thus the number of mismatching points increases. As for this problem, a weight-adaptive cross-scale algorithm framework for stereo matching is proposed. The cost matching is performed on different scales in the framework of unified cost aggregation function and the information entropy of each pixel window is used as the influence factor of the matching cost at each scale on the whole matching cost. At the same time, a regularization factor is added to the cost function to ensure the cost consistency at different scales for the same pixel. The proposed algorithm framework can be applied to the multi-scale algorithm of cost matching and improve the accuracy and robustness of the existing algorithms. Based on the proposed algorithm framework, the different cost aggregate functions are tested on the Middlebury dataset. To ensure the fairness of tests, as for each algorithm, there is no a subsequent parallax refinement step. The experimental results show that the proposed algorithm effectively improves the accuracy and robustness of multi-scale stereo matching.

Newport宣传-MKS新实验室计划
补充资料

中图分类号:TN911.73

DOI:10.3788/aos201838.1215006

所属栏目:机器视觉

基金项目:装备预研领域基金(61400010102)

收稿日期:2018-06-12

修改稿日期:2018-07-23

网络出版日期:2018-07-27

作者单位    点击查看

李培玄:中国科学院沈阳自动化研究所, 辽宁 沈阳 110016中国科学院大学, 北京 100049中国科学院光电信息处理重点实验室, 辽宁 沈阳 110016辽宁省图像理解与视觉计算重点实验室, 辽宁 沈阳 110016
刘鹏飞:中国科学院沈阳自动化研究所, 辽宁 沈阳 110016中国科学院大学, 北京 100049中国科学院光电信息处理重点实验室, 辽宁 沈阳 110016辽宁省图像理解与视觉计算重点实验室, 辽宁 沈阳 110016
曹飞道:中国科学院沈阳自动化研究所, 辽宁 沈阳 110016中国科学院大学, 北京 100049中国科学院光电信息处理重点实验室, 辽宁 沈阳 110016辽宁省图像理解与视觉计算重点实验室, 辽宁 沈阳 110016
赵怀慈:中国科学院沈阳自动化研究所, 辽宁 沈阳 110016中国科学院光电信息处理重点实验室, 辽宁 沈阳 110016辽宁省图像理解与视觉计算重点实验室, 辽宁 沈阳 110016

联系人作者:赵怀慈(hczhao@sia.cn)

【1】Liu C, Yuen J, Torralba A. SIFT flow: dense correspondence across scenes and its applications[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(5): 978-994.

【2】Fan H R, Yang F, Pan X R, et al. Stereo matching algorithm for improved Census transform and gradient fusion[J]. Acta Optica Sinica, 2018, 38(2): 0215006.
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, 38(2): 0215006.

【3】Lin S, Yin X Y, Tang Y D. Research status and prospect of binocular stereo matching technology[J]. Science Technology and Engineering, 2017, 17(30): 135-147.
林森, 殷新勇, 唐延东. 双目视觉立体匹配技术研究现状和展望[J]. 科学技术与工程, 2017, 17(30): 135-147.

【4】Scharstein D, Szeliski R, Zabih R. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms[C]∥Proceedings of IEEE Workshop on Stereo and Multi-Baseline Vision, 2001: 131-140.

【5】Mei X, Sun X, Dong W M, et al. Segment-tree based cost aggregation for stereo matching[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2013: 313-320.

【6】Rhemann C, Hosni A, Bleyer M, et al. Fast cost-volume filtering for visual correspondence and beyond[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(2): 504-511.

【7】Yang Q X. A non-local cost aggregation method for stereo matching[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2012: 1402-1409.

【8】Yoon K J, Kweon I S. Adaptive support-weight approach for correspondence search[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(4): 650-656.

【9】Liu J, Zhang J X, Dai Y, et al. Dense stereo matching based on cross-scale guided image filtering[J]. Acta Optica Sinica, 2018, 38(1): 0115004.
刘杰, 张建勋, 代煜, 等. 基于跨尺度引导图像滤波的稠密立体匹配[J]. 光学学报, 2018, 38(1): 0115004.

【10】Mallot H A, Gillner S, Arndt P A. Is correspondence search in human stereo vision a coarse-to-fine process?[J]. Biological Cybernetics, 1996, 74(2): 95-106.

【11】Marr D, Poggio T. A computational theory of human stereo vision[J]. Proceedings of the Royal Society of London. Series B, Biological Sciences, 1979, 204(1156): 301-328.

【12】Zhang K, Fang Y Q, Min D B, et al. Cross-scale cost aggregation for stereo matching[C]∥Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2014: 1590-1597.

【13】Scharsteinand D, Szeliski R, Hirschmüller H. The middlebury stereo vision page[EB/OL]. [2018-06-12]. http: ∥vision.middlebury.edu/stereo/.

引用该论文

Li Peixuan,Liu Pengfei,Cao Feidao,Zhao Huaici. Weight-Adaptive Cross-Scale Algorithm for Stereo Matching[J]. Acta Optica Sinica, 2018, 38(12): 1215006

李培玄,刘鹏飞,曹飞道,赵怀慈. 自适应权值的跨尺度立体匹配算法[J]. 光学学报, 2018, 38(12): 1215006

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF