Skip to content

Desolatexxx/UnsupervisedDeepImageStitching

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 

Repository files navigation

Unsupervised Deep Image Stitching: Reconstructing Stitched Features to Images

Lang Nie*, Chunyu Lin*, Kang Liao*, Shuaicheng Liu`, Yao Zhao*

* Institute of Information Science, Beijing Jiaotong University

` School of Information and Communication Engineering, University of Electronic Science and Technology of China

image

Dataset (UDIS-D)

The details of the dataset can be found in our paper. (arXiv, IEEE Xplore)

We release our testing results with the proposed dataset together. One can download it in in Google Drive or Baidu Cloud(Extraction code: 1234).

Compared with ours

You can try the testing set of the proposed dataset with your own algorithm. And our results in the testing set are also provided with the testing set.

Codes

Requirement

  • python 3.6
  • numpy 1.18.1
  • tensorflow 1.13.1
  • tensorlayer 1.8.0

How to run it

Due to the memory limitation of our GPU, we implement this unsupervised solution in two stages:

For windows system

For windows OS users, you have to change '/' to '\\' in 'line 71 of ImageAlignment/Codes/utils.py' and 'line 48 of ImageReconstruction/Codes/utils.py'.

Experimental results on robustness

By resizing the input images to different resolutions, we simulation the change of feature quantity to compare ours with other methods in robustness. The results can be available in https://drive.google.com/drive/folders/1URFKTiUxaZ8i6pcHIKhxVTf-LkTNnXpK?usp=sharing.

Note: Since the RANSAC algorithm randomly selects the sample points, and the feature (SIFT) detection is not strictly consistent each time, different tests on the same image may differ. But the overall performance should be close to the results reported in our experiments.

Meta

NIE Lang -- [email protected]

@ARTICLE{9472883,
  author={Nie, Lang and Lin, Chunyu and Liao, Kang and Liu, Shuaicheng and Zhao, Yao},
  journal={IEEE Transactions on Image Processing}, 
  title={Unsupervised Deep Image Stitching: Reconstructing Stitched Features to Images}, 
  year={2021},
  volume={30},
  number={},
  pages={6184-6197},
  doi={10.1109/TIP.2021.3092828}}

References

[1] L. Nie, C. Lin, K. Liao, M. Liu, and Y. Zhao, “A view-free image stitching network based on global homography,” Journal of Visual Communication and Image Representation, p. 102950, 2020.
[2] L. Nie, C. Lin, K. Liao, and Y. Zhao, “Learning edge-preserved image stitching from large-baseline deep homographyn,” arXiv preprint arXiv:2012.06194, 2020.
[3] T. Nguyen, S. W. Chen, S. S. Shivakumar, C. J. Taylor, and V. Kumar. Unsupervised deep homography: A fast and robust homography estimation model. IEEE Robotics and Automation Letters, 3(3):2346–2353, 2018.
[4] J. Zhang, C. Wang, S. Liu, L. Jia, N. Ye, J. Wang, J. Zhou, and J. Sun, “Content-aware unsupervised deep homography estimation,” in European Conference on Computer Vision, pp. 653–669, Springer, 2020.

About

TIP2021 - Unsupervised deep image stitching network

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%