Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/cheind/image-stitch

Image stitching of planar targets based on analytical homographies

https://github.com/cheind/image-stitch

Science Score: 26.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Committers with academic emails
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (10.1%) to scientific vocabulary

Keywords

analytical computer-vision homographies image-stitching math python
Last synced: 6 months ago · JSON representation

Repository

Image stitching of planar targets based on analytical homographies

Basic Info
  • Host: GitHub
  • Owner: cheind
  • License: mit
  • Language: Python
  • Default Branch: main
  • Homepage:
  • Size: 43.4 MB
Statistics
  • Stars: 5
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Topics
analytical computer-vision homographies image-stitching math python
Created almost 2 years ago · Last pushed over 1 year ago
Metadata Files
Readme License

README.md

Planar Image Stitching

This repository demonstrates stitching multiple images of a planar target. We analytically derive homographies by assuming the camera poses with respect to the target are known. Hence neihter feature matching nor Linear Transforms (DLTs) are required to estimate the homographies. The following image shows four views stitched in the green reference camera frame.

In the following image the same scene is stitched in a virtual camera whose image plane aligns with the ground plane, having a pixel resolution of 500px per meter, which is suited well for taking direct metric measurements.

Both images exhibit ghosting artefacts causved by blending moving objects and warping objects that violate the in-target-plane assumption.

Theory

See PlanarImageStitching.md for background information on the stitching process.

Applications

The ghosting effect has intriguing applications in search-and-rescue scenarios. In particular, OutOfFocusAnalysis.md explores techniques for reconstructing objects that are significantly obscured by out-of-focus elements, providing valuable insights for improving visibility in challenging environments. This allows us to go images of mostly foliage (left) to reconstructed hidden duck (right).

Usage

The code provided is for demonstration purposes only. It is limited to a scenario in which a moving camera observes a ground floor. The extrinsics are computed from knowing the fisheye intrinsics/distortions and the calibration pattern configuration.

Stitching

```shell

Stitch in camera 3 view (index starting at zero)

python stitch.py basepath=data/ plane.idx=2 ```

```shell

Stitch in plane pi using px/m of 500

python stitch.py basepath=data/ plane.idx=-1 plane.pxperm=500 plane.extent="[-3, 5, -1, 2]" ```

```shell

Stitch in plane pi using px/m of 10

python stitch.py basepath=data/ plane.idx=-1 plane.pxperm=10 plane.extent="[-3, 5, -1, 2]" ```

Out-of-Focus (Ducky Rescue)

First, stitch in a plane parallel to ground but lifted by 0.03m. We turn on saving of raw values which provides us the necessary information for performing out-of-focus analysis.

```shell

Stitch and save intermediate results

python stitch.py basepath=data/oof plane.idx=-1 plane.extent="[-0.5,1,0,1.5]" plane.z=0.03 save_raw=true ```

```shell

Perform out-of-focus analysis

python oof.py rawpath=tmp/stitch-20241013-044050.npz python oof.py rawpath=tmp/stitch-20241013-044050.npz weightfilter=baseline python oof.py rawpath=tmp/stitch-20241013-044050.npz weightfilter=color
python oof.py rawpath=tmp/stitch-20241013-044050.npz weightfilter=outlier python oof.py rawpath=tmp/stitch-20241013-044050.npz weightfilter=color color.T=10 integrate.min_weight=0.2 ```

Owner

  • Name: Christoph Heindl
  • Login: cheind
  • Kind: user
  • Location: Austrian area

I am a computer scientist working at the interface of perception, robotics and deep learning.

GitHub Events

Total
  • Watch event: 1
Last Year
  • Watch event: 1

Committers

Last synced: 8 months ago

All Time
  • Total Commits: 76
  • Total Committers: 2
  • Avg Commits per committer: 38.0
  • Development Distribution Score (DDS): 0.461
Past Year
  • Commits: 47
  • Committers: 2
  • Avg Commits per committer: 23.5
  • Development Distribution Score (DDS): 0.34
Top Committers
Name Email Commits
Christoph Heindl c****d@p****t 41
Christoph Heindl c****l@g****m 35
Committer Domains (Top 20 + Academic)
profactor.at: 1

Issues and Pull Requests

Last synced: 8 months ago

All Time
  • Total issues: 0
  • Total pull requests: 0
  • Average time to close issues: N/A
  • Average time to close pull requests: N/A
  • Total issue authors: 0
  • Total pull request authors: 0
  • Average comments per issue: 0
  • Average comments per pull request: 0
  • Merged pull requests: 0
  • Bot issues: 0
  • Bot pull requests: 0
Past Year
  • Issues: 0
  • Pull requests: 0
  • Average time to close issues: N/A
  • Average time to close pull requests: N/A
  • Issue authors: 0
  • Pull request authors: 0
  • Average comments per issue: 0
  • Average comments per pull request: 0
  • Merged pull requests: 0
  • Bot issues: 0
  • Bot pull requests: 0
View more stats
Top Authors
Issue Authors
Pull Request Authors
Top Labels
Issue Labels
Pull Request Labels

Dependencies

requirements.in pypi
  • matplotlib >=3.8.3
  • omegaconf *
  • opencv-python-headless >=4.9.0