2D Code Extraction - Extract 2D Code from Invoice Images

Introduction

This is a sub problem coming from a real world project: an automatic invoice information logging system for a supermarket which extracts information inhibited by the 2D codes printed at the upper right placeholder of the invoices. Unfortunately some of the 2D codes were so badly posed that they overlapped with the border of the box, which makes locating and extracting these codes more difficult. We dealt this problem with a simple scanning line algorithm and came up with a quick and dirty solution. By the way, this project is just for demonstration and won’t be pushed to the trunk of that project.

This program uses a simple and trival method which only apply to exactly this kind of invoice, we hardcoded the approximate location and size of the 2d code as well as the padding around it to make the result more accurate. Now it can even handle the worst case they have ever met.

Try a Shakedown Run

Build the project with make and go to bin folder, which contains a demo script and some sample invoice images, type ./demo.sh sample_1.ppm to launch the program, you can also play with the other 3 samples in the same way.