Feature-Based OCR

PART I - Intro & TOC

Screenshot

A screenshot of the OCR program prototype.

About

CSIE Project by Jerry Chen

Objective

To design and implement an optical character recognizer that can recognize hand-written characters from a specific finite domain (A-Z)

Specification

• PROBLEM: Classification
• INPUT: A bitmap of pixels representing the image of a character (can either be hand-drawn or loaded from file)
• OUTPUT: The ASCII code of the recognized character
• LANGUAGE: Microsoft Visual Basic

The Plan

1. Feature Extraction
2. Feature Analysis
3. Recognition
4. Error Correction

Table of Contents

1. Intro & TOC
2. Overview
3. Character Features
4. Skeletonization
5. Fragmentation
6. Character Bleeding
7. Feature Reduction
8. Problems

Next: Overview

Questions? Comments? Send them here. Or return to deXXed.