QR-Code-Scanner

A QR-Code Scanner, written in C++ and using the OpenCV library, which was developed as an optional part of the university course "Computer Vision" in a small team of three.

Features

The program detects QR-Codes from image or video input, and outputs the detected QR-Code as a binary image, where each pixel is associated with a square of the QR-Code, and saves the image into a folder.

Moreover, the program is able to generate a synthetic dataset of QR-Codes from a set of ground truth QR-Codes and background images. This synthetic dataset can then be used to test the QR-Code detection.

Example

The following image presents an example of the QR-Code detection in this project. The image used is from the generated synthetic dataset. As you can see, the QR-Code is rotated and perspectively transformed and the image is very blurry and noisy. The implemented algorithm can handle this though and outputs a QR-Code that isn't too far off from the original one. This is of course an extreme case and an error should be expected.

How to Use

Call the main function with the following parameters to run the respective mode:

• Camera Mode: no input
  Attempt to open a camera feed and continuously search for QR-Codes.

• Folder Scan Mode: [folder-path]
  Scan all images in the input folder and save the detection results into the subfolders ScanPositive and ScanNegative, which will be created at runtime.

• Evaluation Mode: [input-path] [output-path]
  Read a single image stored at input-path and save the detection result to output-path.

• Generate Mode: [-generate] [ground-truth-path] [output-path]
  Read images stored at ground-truth-path and generate synthetic dataset at output-path. The background image folder 99_bg has to be located next to the input folder ground-truth-path.

Examples

• "-generate" "data/00_ground_truth" "data"
  Uses the ground truth QR-Codes given in the project and generates a synthetic dataset in the data folder.

• "data/07_noise"
  After running the first example, this will use the generated images saved in the folder 07_noise and scan every single image in it.

Paper

See our paper for an in-depth explanation of this project. The paper is written in German language.