Applying Image Processing Technology to Pattern Recognition of Poker and Invoice
碩士 === 國立彰化師範大學 === 電機工程學系 === 105 === This thesis proposes a method of pattern recognition for two applications. One is to recognize the number and suits of playing cards, and the other is for the invoice number. The system applies edge detection, image rotation, and morphology to character enhance...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2017
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/tcg9rd |
| Summary: | 碩士 === 國立彰化師範大學 === 電機工程學系 === 105 === This thesis proposes a method of pattern recognition for two applications. One is to recognize the number and suits of playing cards, and the other is for the invoice number. The system applies edge detection, image rotation, and morphology to character enhancement. We use connected-region labeling to capture the region of interest of poker. In the other hand, we take the width of invoice as a scale to capture the region of interest of invoice. The regions of interest for two applications are captured automatically.
We use Optical Character Recognition (OCR) for pattern recognition. Based on comparing the normalized character with database, the system can find the correct character. According to the experimental results the proposed method has high efficiency. The character recognition accuracy of poker is 100%. The processing time of recognizing poker takes about 2.774 seconds. After adding new module to the database, the character recognition accuracy of invoice is about 99.24%. The processing time of recognizing invoice takes about 3.381 seconds.
|
|---|