Git Product home page Git Product logo

crossscalestereo's Introduction

Cross-Scale Cost Aggregation for Stereo Matching (CVPR 2014)

Compilation

Windows

The code is a Visual Studio 2010 project on Windows x64 platform. To build the project, you need to configure OpenCV on your own PC. (version 2.4.6, however, other versions are acceptable by modifying CommFunc.h).

Other Platforms

The code requires no platform-dependent libraries. Thus, it is easy to compile it on other platforms with OpenCV.

Usage

Run the program with the following paramters: Usage: [CC_METHOD] [CA_METHOD] [PP_METHOD] [C_ALPHA] [lImg] [rImg] [lDis] [maxDis] [disSc]

  • [CC_METHOD] -- cost computation methods, currently support:
  • [CA_METHOD] -- cost aggregation methods, currently support:
  • [PP_METHOD] -- post processing methods, currently support:
  • [C_ALPHA] -- regularization paramter, i.e. $\lambda$ in the paper.
  • [lImg] -- input left color image file name. (all formats supported by OpenCV)
  • [rImg] -- input right color image file name.
  • [lDis] -- output left disparity map file name.
  • [maxDis] -- maximum disparity range, e.g. 60 for Middlebury and 256 for KITTI dataets.
  • [disSc] -- scale disparity, e.g. 4 for Middlebury and 1 for KITTI datasets.

Hint: to enable post-processing, you must uncomment // #define COMPUTE_RIGHT in CommFunc.h to allow computing right disparity map.

Citation

Citation is very important for researchers. If you find this code useful, please cite:

@inproceedings{CrossScaleStereo,
        author    = {Kang Zhang and Yuqiang Fang  and Dongbo Min and Lifeng Sun and Shiqiang Yang  and Shuicheng Yan and Qi Tian},
        title     = {Cross-Scale Cost Aggregation for Stereo Matching},
        booktitle = {CVPR},
        year     = {2014}
}

Since some cost aggregation methods (GF, NL, ST) are built uppon other papers' code, you also need to cite corresponding papers as listed below.

Reference

[CT]: R. Zabih and J. Woodfill. Non-parametric local transforms for computing visual correspondence. In ECCV, 1994

[GF]: C. Rhemann, A. Hosni, M. Bleyer, C. Rother, and M. Gelautz. Fast cost-volume filtering for visual correspondence and beyond. In CVPR, 2011

[ST]: X. Mei, X. Sun, W. Dong, H. Wang, and X. Zhang. Segment-tree based cost aggregation for stereo matching. In CVPR, 2013

[BF]: K.-J. Yoon and I. S. Kweon. Adaptive support-weight approach for correspondence search. TPAMI, 2006

[NL]: Q. Yang. A non-local cost aggregation method for stereo matching. In CVPR, 2012

crossscalestereo's People

Contributors

rookiepig avatar sunshinnnn avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

crossscalestereo's Issues

the error rate?

why I test on the kitti training set,my error is lower than your result ,futhermore,the difference is very big,because the dataset have been updated?thank you ,very much

can not open opencv_calib3d246d.lib issue

when you change opencv version yourself, and change CommFunc.h to your opencv version, but it still procuces this issue. That because you also need to change opencv version in /CAST/StereoDisparity.h .

memory leak

hi,while run the program,I found memory leak in the program,but I dont know where the bug?

请问关于论文中实验的相关部分

您好,请问您的论文里测试的M27的数据集使用的是原始分辨率的图像还是3倍下采样后的图像呢?看到您的论文里只有说新出的Middlebury 2014使用的是四分一分辨率的图像(这个可能是您的那篇TCSVT)。
还想请教您一个问题,就是我在利用左右视差真值计算被遮挡区域时,利用一致性计算(数据集给的SDK程序),但是通常得到的被遮挡图的左侧原本应该是被遮挡区域,而我计算得到的有的不是被遮挡,这个问题请问您当时是如何解决的?也就是左图的被遮挡区域应该包括左图像的最左侧几列的区域,而利用一致性计算时有时候那部分并不是遮挡区域。(我思考可能是因为那部分都是和相机平行的平面,视差与未被遮挡的部分相同,导致阈值选择为1时无法筛掉)主要针对Middlebury 2005及2006没有给出遮挡区域的图,故而想问问您!谢谢

the identifier “m_mst_value_sum_aggregated_from_child_to_parent” could not be found?

When I tried to integrate the SSCA algorithm into my project, I encounterd this problem. The Chinese characters of the error information on this picture represents that the compiler could not find the identifier about "m_mst_value_sum_aggregated_from_child_to_parent", "m_mst_value_sum_aggregated_from_parent_to_child" and "m_mst_value_to_be_filtered". My running environment is as follows:
windows10,
Visual Studio 2019(community),
opencv 4.3.0.
Is there anyone who can help me with this problem? Thanks!

——
error

Installing on Linux

Is there a readily available Linux installation package for CSSM (with cmake files etc), or do we need to create one on our own?

Question about time

Did you try your method using parallel computing? I know in your paper you calculate the time complexity although 1/7 seems not to much.(KITTI show your method need 140s ?)

程序时间问题

你好,请问你所达到的计算时间是在debug模式下的吗,我用在debug下,时间都是30多秒,而且看了下都是在s.v阶段花的时间长。

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.