水稻是台灣賴以維生的糧食作物，對於都市環境有改善熱島效應和 綠化生活空間的優點。水稻的管理和分析一直以來是水土保持重要一環，在 傳統上水稻的判釋必須透過人工現場調查，費時費力，而現在科技進步採用 空間影像判釋是一個相當有效的方法。線性判別分析雖是眾所皆知的統計方 法，然而很少有方法將判別式進行優化，進而提高判釋準度；因此本研究以 高光譜影像為材料，研究與評估粒子群優化演算法在影像中的水稻判釋成 果，以粒子群演算法為探索式的演算法，將所有因子組合後以探索的方式進 行整合比較，獲取最佳的判釋規則。本研究以線性判別分析方式進行水稻影 像判釋；本研究使用的高光譜影像空間解析度為1 公尺，涵蓋可見光至近紅 外光72 個波段，研究圈選500 筆水稻樣本點，一半為訓練範例，另一半為 測試範例，之後使用粒子群優化演算法擇出高光譜72 個波段中對於水稻判 釋最有利的波段，再進行判別分析進行分類，最後使用誤差矩陣表以及主題 圖呈現出分類後之成果。結果顯示出，雖然優化後的分類器在誤差矩陣無明 顯優化，可在主題圖繪製時，減少雜訊並強化邊界的判釋率。

Paddy rice is the major crop of food in Taiwan. The estimation of its area is important because the national food policy, yearly crop yields calculation and post-disaster reimburse depend on it. In the past, it is performed through field exploration and rice pattern revising, which is a large amount of human power and time-consuming works. A more economic manner to estimate paddy rice area is desired. Accordingly, this study aims to design a paddy rice classifier by which the area of paddy rice in a remote sensing image can be calculated. In this study 250 samples of data (pixels in image) are selected and used for classification training process, and another 250 samples are selected for validation. Linear discrimination method with and without particle swarm optimization is used to construct the classification rules. It is noted that in the present study texture information and vegetation index are included as ancillary features. This is quite beneficial in constructing a more effective informational retrieval strategy for paddy rice classification. In the study the main contributions are summarized as follows, (1) retrieving the primary factors, (2) establishing a series of classification rules for paddy rice fields, and (3) constructing the thematic map.