[WIP] Python Image Processing module to stitch a series of similar images to create a mosaic.
- OpenCV 2.4
- Python 2.7
To run an interactive demo. Make sure you have OpenCV 2.4 with python bindings
loaded in PYTHONPATH
. There is a script that'll walk you through the
stitching process in a GUI.
cd to/mosa/path
python scripts/demo.py
If you have the dependencies installed, this should pop up the hummingbird image
with the features found and linked. To go to the next frame, type q
.
You can benchmark different OpenCV feature detection and matching algorithms to see how they perform against each other given two images.
The benchmarking script will output a CSV file and mix and match the algorithms
specified in the benchmark.yml
file in
here.
The result are the of benchmarking on two images are:
- Number of keyponints
- Time to find keypoints
- Number of features
- Time to find features
- Number of matches
- Time to find matches
- Number of outliers
- Number of KNN matches
- Time to find KNN matches
- Number of KNN outliers
Right now, only the following algorithms are supported.
To run the benchmark, do the following:
cd to/mosa/path
python scripts/benchmark/benchmark.py path/to/image1 path/to/image2
The above script will generate a out.csv
file in most/scripts/benchmark
.