# Natural Image Stitching with the Global Similarity Prior
### [[Project page]](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/) [[Paper]](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/ECCV-2016-NISwGSP.pdf) [[Supplementary]](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/ECCV-2016-NISwGSP-supplementary-material.pdf)
<img src="./images/teaser.jpg" width="100%"/>
This repository is our C++ implementation of the **ECCV 2016** paper, **Natural Image Stitching with the Global Similarity Prior**. If you use any code or data from our work, please cite our paper.
### Download
1. [Poster](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/Poster.pdf), [Short Presentation](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/Short-Presentation.pdf) and [Thesis Presentation](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/Thesis-Presentation.pdf)
2. [Paper](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/ECCV-2016-NISwGSP.pdf)
3. [Supplementary](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/ECCV-2016-NISwGSP-supplementary-material.pdf)
* We tested four state-of-the-art methods and ours on 42 sets of images in same setting (grid size, feature points and parameters).
4. [Input-42-data](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/input-42-data.zip)
5. [All our results](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/0_results.zip)
### Ubuntu Version
If you want to build this project under **Ubuntu**, please refer to https://github.com/Yannnnnnnnnnnn/NISwGSP
Thanks a lot to @Yannnnnnnnnnnn!
### Windows Version (Visual Studio)
If you want to build this project under **Windows**, please refer to https://github.com/firdauslubis88/NISwGSP
Thanks a lot to @firdauslubis88!
### Usage
1. Download code and compile.
* You need **Eigen**, **VLFeat**, **OpenCV 3.0.0** and [**OpenMP**](https://github.com/nothinglo/NISwGSP/issues/8) (if you don't need to use omp.h, you can ignore it.)
* My GCC_VRSION is Apple LLVM 6.0
```
GCC_C_LANGUAGE_STANDARD = GNU99 [-std=gnu99]
CLANG_CXX_LANGUAGE_STANDARD = GNU++14 [-std=gnu++14]
CLANG_CXX_LIBRARY = libc++ (LLVM C++ standard library with C++11 support)
```
* My Eigen version is 3.2.7 (development branch). You need to make sure you can use "LeastSquaresConjugateGradient" class.
2. Download [input-42-data](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/input-42-data.zip).
* 42 sets of images: 6 from [1], 3 from [2], 3 from [3], 7 from [4], 4 from [5] and 19 collected by ourselves.
3. Move **[input-42-data]** folder to your working directory.
4. Run the command:
```
./exe folder_name_in_[input-42-data]_folder
```
The results can be found in **[0_results]** folder under **[input-42-data]** folder.
5. Optional:
* You can control the parameters in **Configure.h** or **xxx-STITCH-GRAPH.txt**
### Results
#### More natural:
<table>
<tr>
<th>AutoStitch</th>
<th>Ours</th>
<th>Ours(border)</th>
</tr>
<tr>
<td><img src="./images/SPHP-building-result/SPHP-building-[AutoStitch][BLEND_LINEAR].png"/></td>
<td><img src="./images/SPHP-building-result/SPHP-building-[NISwGSP][3D][BLEND_LINEAR].png"/></td>
<td><img src="./images/SPHP-building-result/SPHP-building-[NISwGSP][3D][BLEND_LINEAR][Border].png"/></td>
</tr>
<tr>
<td><img src="./images/CAVE-church-result/CAVE-church-[AutoStitch][2D][BLEND_LINEAR].png"/></td>
<td><img src="./images/CAVE-church-result/CAVE-church-[NISwGSP][2D][BLEND_LINEAR].png"/></td>
<td><img src="./images/CAVE-church-result/CAVE-church-[NISwGSP][2D][BLEND_LINEAR][Border].png"/></td>
</tr>
</table>
<table>
<tr>
<th>AutoStitch</th>
<th>AANAP</th>
<th>Ours</th>
</tr>
<tr>
<td width="33%"><img src="./images/NISwGSP-03_SantaMaria-all-result/NISwGSP-03_SantaMaria-all-[AutoStitch][BLEND_AVERAGE].png"/></td>
<td width="33%"><img src="./images/NISwGSP-03_SantaMaria-all-result/NISwGSP-03_SantaMaria-all-[AANAP][BLEND_AVERAGE].png"/></td>
<td width="33%"><img src="./images/NISwGSP-03_SantaMaria-all-result/NISwGSP-03_SantaMaria-all-[NISwGSP][2D][BLEND_AVERAGE].png"/></td>
</tr>
</table>
<table>
<tr>
<th>AutoStitch</th>
<th>AANAP</th>
</tr>
<tr>
<td width="50%"><img src="./images/NISwGSP-08_SienaCathedralLibrary3-result/NISwGSP-08_SienaCathedralLibrary3-[AutoStitch][BLEND_LINEAR].png"/></td>
<td width="50%"><img src="./images/NISwGSP-08_SienaCathedralLibrary3-result/NISwGSP-08_SienaCathedralLibrary3-[AANAP][BLEND_LINEAR].png"/></td>
</tr>
<tr>
<th>Ours(2D)</th>
<th>Ours(3D)</th>
</tr>
<tr>
<td width="50%"><img src="./images/NISwGSP-08_SienaCathedralLibrary3-result/NISwGSP-08_SienaCathedralLibrary3-[NISwGSP][2D][BLEND_LINEAR].png"/></td>
<td width="50%"><img src="./images/NISwGSP-08_SienaCathedralLibrary3-result/NISwGSP-08_SienaCathedralLibrary3-[NISwGSP][3D][BLEND_LINEAR].png"/></td>
</tr>
</table>
#### Stitching of 20 images:
<table>
<tr>
<th>AANAP</th>
<th>Ours</th>
</tr>
<tr>
<td width="50%"><img src="./images/CAVE-times_square-result/CAVE-times_square-[AANAP][BLEND_AVERAGE].png"/></td>
<td width="50%"><img src="./images/CAVE-times_square-result/CAVE-times_square-[NISwGSP][2D][BLEND_AVERAGE].png"/></td>
</tr>
</table>
#### Stitching of 35 images:
<table>
<tr>
<th>AANAP</th>
<th>Ours(2D)</th>
<th>Ours(3D)</th>
</tr>
<tr>
<td width="33%"><img src="./images/CAVE-atrium-result/CAVE-atrium-[AANAP][BLEND_AVERAGE].png"/></td>
<td width="33%"><img src="./images/CAVE-atrium-result/CAVE-atrium-[NISwGSP][2D][BLEND_AVERAGE].png"/></td>
<td width="33%"><img src="./images/CAVE-atrium-result/CAVE-atrium-[NISwGSP][3D][BLEND_AVERAGE].png"/></td>
</tr>
</table>
#### Our method can be used to empower other methods with APAP’s alignment capability:
<table>
<tr>
<th>AutoStitch</th>
<th>AutoStitch + Ours</th>
<th>Ours</th>
</tr>
<tr>
<td width="33%"><img src="./images/DHW-temple-result/DHW-temple-[AutoStitch][BLEND_AVERAGE].png"/></td>
<td width="33%"><img src="./images/DHW-temple-result/DHW-temple-[UG_AutoStitch][BLEND_AVERAGE].png"/></td>
<td width="33%"><img src="./images/DHW-temple-result/DHW-temple-[NISwGSP][3D][BLEND_AVERAGE].png"/></td>
</tr>
</table>
### Debug mode
You can disable debug mode by adding **NDEBUG** macro. Otherwise you will see the intermediate which is located in the **[1_debugs]** folder under **[input-42-data]**. You can download [all intermediate data](https://www.cmlab.csie.ntu.edu.tw/project/stitching-wGSP/1_debugs.zip). The intermediate example:
<table>
<tr>
<th>Border</th>
<th>Mesh</th>
</tr>
<tr>
<td><img src="./images/DHW-temple-result/DHW-temple-[NISwGSP][3D][BLEND_LINEAR][Border].png"/></td>
<td><img src="./images/DHW-temple-result/DHW-temple-[NISwGSP][3D][BLEND_LINEAR][Mesh].png"/></td>
</tr>
</table>
<table>
<tr>
<th>Initial Features</th>
<th>After sRANSAC</th>
</tr>
<tr>
<td><img src="./images/DHW-temple-result/feature_pairs-init-4-5-345.jpg"/></td>
<td><img src="./images/DHW-temple-result/feature_pairs-sRANSAC-4-5-273.jpg"/></td>
</tr>
</table>
<table>
<tr>
<th>Line Data 1</th>
<th>Line Data 2</th>
</tr>
<tr>
<td><img src="./images/DHW-temple-result/line-result-4.jpg"/></td>
<td><img src="./images/DHW-temple-result/line-result-5.jpg"/></td>
</tr>
</table>
### Speed
If you want to speed up, **MATLAB** solver is significantly faster than **Eigen**.
### Publication
[Yu-Sheng Chen](https://www.cmlab.csie.ntu.edu.tw/~nothinglo/) and [Yung-Yu Chuang](https://www.csie.ntu.edu.tw/~cyy/).
[National Taiwan University](https://www.ntu.edu.tw)
Natural Image Stitching with Global Similarity Prior.
Proceedings of European Conference on Computer Vision 2016 (ECCV 2016), Part V, pp. 186-201, October 2016, Amsterdam, Netherland.
### Citation
```
@INPROCEEDINGS{Chen:2016:NIS,
AUTHOR = {Y
NISwGSP (natural image stitching with the global similarity prio...
3星 · 超过75%的资源 需积分: 50 157 浏览量
2018-12-05
16:36:57
上传
评论 2
收藏 85.84MB ZIP 举报 
NISwGSP (natural image stitching with the global similarity prior)算法 （609个子文件） 
sift.1 5KB mser.1 4KB vlfeat.7 3KB vlfeat.bib 7KB covdet.c 110KB kmeans.c 71KB sift.c 70KB svm.c 67KB generic.c 56KB gmm.c 52KB hog.c 37KB kdtree.c 34KB imopv.c 34KB mser.c 30KB scalespace.c 28KB mathop.c 28KB vl_covdet.c 28KB sift.c 26KB dsift.c 24KB vl_svmtrain.c 22KB aib.c 21KB liop.c 21KB homkermap.c 19KB fisher.c 19KB vl_alldist2.c 17KB mser.c 17KB host.c 16KB vl_sift.c 14KB quickshift.c 14KB slic.c 14KB stringop.c 13KB pgm.c 13KB mathop_sse2.c 12KB lbp.c 12KB svmdataset.c 12KB vl_gmm.c 12KB vl_kmeans.c 11KB vlad.c 10KB vl_dsift.c 10KB vl_hog.c 10KB vl_imsmooth.c 10KB vl_mser.c 9KB getopt_long.c 9KB rodrigues.c 9KB hikmeans.c 9KB vl_ubcmatch.c 9KB random.c 8KB test_gmm.c 8KB ikmeans.c 8KB imopv_sse2.c 8KB vl_aib.c 7KB vl_siftdescriptor.c 7KB vl_localmax.c 7KB mathop_avx.c 6KB vl_ihashsum.c 6KB vl_inthist.c 6KB vl_hikmeanspush.c 6KB vl_imwbackwardmx.c 6KB vl_hikmeans.c 6KB vl_kdtreebuild.c 6KB vl_alldist.c 6KB vl_homkermap.c 6KB array.c 6KB vl_erfill.c 6KB vl_fisher.c 5KB vl_liop.c 5KB vl_twister.c 5KB vl_vlad.c 5KB vl_kdtreequery.c 4KB vl_ihashfind.c 4KB vl_ikmeans.c 4KB vl_slic.c 4KB vl_aibhist.c 4KB test_heap-def.c 4KB vl_imdisttf.c 4KB vl_binsum.c 4KB vl_quickshift.c 4KB vl_cummax.c 4KB test_stringop.c 4KB vl_ikmeanspush.c 4KB vl_sampleinthist.c 3KB vl_imintegral.c 3KB test_imopv.c 3KB vl_lbp.c 2KB test_threads.c 2KB test_getopt_long.c 2KB vl_irodr.c 2KB vl_rodr.c 2KB test_svd2.c 2KB vl_tpsumx.c 2KB vl_binsearch.c 2KB test_vec_comp.c 2KB test_kmeans.c 2KB vl_version.c 2KB test_mathop.c 2KB aib.c 1KB test_mathop_abs.c 1KB vl_threads.c 995B vl_simdctrl.c 985B test_sqrti.c 941B共 609 条
- 1
- 2
- 3
- 4
- 5
- 6
- 7
资源评论
robot19922020-10-01数据集有吗
