| Summary: | 碩士 === 國立彰化師範大學 === 工業教育學系 === 89 === Abstract
In the thesis, we combine the Images Management Techniques and Back-Propagation Network, simply named "BPN", to recognize true Chinese Chess chessmen. By this, it can be the interface between people and the machine for Digital Chinese Chess.
The reading management for images of Chinese Chess can be two parts. Firstly, we normalize and sample by features the images of Chinese Chess chessmen. Then, BPN is trained to identify Chinese Chess chessmen by getting distinctive data. The final goal we want is to make it to be the input of Digital Chinese Chess, instead of the mouse.
The lighting, the size of chessmen, and the miss of lines will cause to hardly recognize these images, when reading the true Chinese Chess by CCD. However, the Images Management Techniques on the thesis will resolve problems above. It will be also discussed that advantages and disadvantages of experiment about the best preprocess algorithm, as well as questions of learning inaccuracy and astringency while we use BPN to get the image data.
|
|---|
