WordOCR
How hard can it be to make a computer to recognize text from pictures? As a part of Data Mining course, we implemented a small program that extracted letters from pictures and then recognised them. This project was done together with Jaana.
Presentation:
Resources:
- Project report.
- Pictures of 242 words that were used in the experiments.
- OpenCV C++ code used for preprocessing and feature extraction.
- Purified data matrix.
- Column 1: character from a-z.
- Column 2: aspect ratio (height divided by weight)
- Columns 3-130: 16 x 8 normalized pixel values.
