Region Object-oriented Classification Model of Automatic Mapping for Paddy Rice Thematic Map

• Main Authors: Che-yuan Li, 李哲源
• Other Authors: 雷祖強
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/41776446078990896107

碩士 === 逢甲大學 === 環境資訊科技碩士學位學程 === 98 === In Taiwan, the paddy rice is very important crop. The government surveys the paddy rice area from aerial photography every year. However, to make GIS paddy rice maps information needs a lot of manpower、time and funding. How to quick and correctly extraction paddy rice area is very valued to research. In the image classification, in general, is used to using pixel-based algorithm to extraction information in image, but result of by pixel-based algorithm have Salt-and-Pepper Effect, so the results can’t be used. Therefore, this research purposed region-based model of Region Object-oriented Classification (ROC) to extraction paddy rice fields, the ROC model combined high-resolution multispectral numerical Aerial photographs grid data and GIS’s vector information. Grid data can representative of status of the surface information；Vector information can representative of hill blocks information of over the years. Combined Grid and Vector information extraction results can retain the original block structure of paddy rice fields, and can extract new or fallow paddy fields on hill blocks of information. The study result shows ROC model can eliminate Salt-and-Pepper Effect, and rise of overall accuracy (the overall accuracy from 0.94 rise to 0.98). The ROC extracted paddy rice fields is completed more than the MLC, and let the extraction result more valuable and helpful for automated mapping and maps modify. The result of ROC model produced vector compare with manual results are very similar, so ROC model used in automated cartography is quite feasible.