1 / 17

Text Recognition Techniques

Text Recognition Techniques. Group #2 Di Wu (d8wu) Ehren Choy (e3choy) Muhammad Qureshi (m2quresh) Mohammad Talha Khalid ( mtkhalid ). Problem. Recognize hand-written characters. Motivation. Hand-writing is a complex problem N eed AI techniques to help solve it. Remember A4?.

cael
Download Presentation

Text Recognition Techniques

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Text Recognition Techniques Group #2 Di Wu (d8wu) Ehren Choy (e3choy) Muhammad Qureshi(m2quresh) Mohammad TalhaKhalid (mtkhalid)

  2. Problem • Recognize hand-written characters

  3. Motivation • Hand-writing is a complex problem • Need AI techniques to help solve it

  4. Remember A4?

  5. Preprocessing • ColourConversion • Edge Detection • Canny Algorithm • Thinning & Skeletonization • Generate Predictor Variables

  6. Feature Extraction • Predictor Variables • Aspect Ratio • Junction and End Points • Loop • Ascenders and Descenders

  7. Application to Radial Basis • Input Layer • One neuron per predictor variable  • Hidden Layer • Calculate distance from center point • Output Layer • One output neuron for every category

  8. Fuzzy Systems • Pre-processing • Feature Extraction • “Structural recognition” • Extracting individual features • Fuzzy classification • Compare word structure with reference words

  9. Identifying structural features - English • Micro Vertical Line • Micro Horizontal Line • Micro Positive Slant • Micro Negative Slant

  10. Identifying structural features - Arabic Numerical decomposition of the word 4 sub-words 1 ascender 1 dot above 2 loops 2 descenders

  11. Compare word structure with reference words • Fuzzy classifier classifies word’s membership in different classes • Uses Fuzzy K nearest neighbor algorithm • Calculate distance between word & training samples

  12. Fuzzy Network: Pros and Cons • Advantages • Lower computational requirements • Disadvantages • Not widely used in handwriting recognition problems

  13. Genetic Programming • Generating graph from input image • Break down image to line segments • Compute fitness using fitness function • Edge Deviation • Graph Deviation • Crossover operation • Replace a path between two vertices in one graph with a path between two matching vertices in another graph

  14. Pros and Cons • Advantages • Easier to generate a large solution set by creating hybrids from a smaller initial set • Disadvantages • Long running time due to high number of possible combinations of graph pairs for computing fitness

  15. Demo

  16. Neural Networks: Pros and Cons • Advantages • Automatic Learning • Quick Classification • Disadvantages • Efficiency • Locality

  17. Questions?

More Related