The Logo Recognizer is simple program that by using image processing algorithms localizes and recognizes sample logo.
In this project I use logo of some big supermarket chain that has its shops in many of European countries.
This implementation is project for my Digital Image Processing class and it is for better understanding image procesing and recognition.
To use this program to recognize other logos you should change two things:
- Change colors range in tresholding function
- Change recognition function (all parameters NM1, NM2, etc.) to regonize shapes from your logo
You can check sample of parameters in Kaufland logo parameters file.
- You should add operation of morphological closing after tresholding to fill small holes in shapes that are some artifacts.
- Add a checking the position of shapes relative to each other to better logo regognition.
This parser has been used only on Linux, but it probably can works on any other platform as well.
git clone https://github.com/BartekkPL/logo-recognizer.git
cd logo-recognizer/
mkdir build
cd build/
cmake ..
make
And after these steps you have builded executable ready to use.
You can find it at logo-recognizer/build/logo-recognizer.
logo-recognizer <image_path>LOGO-RECOGNIZER is released under Apache License.