【正文】 2000, 4(4): 146150.[9]吳健平, 楊星衛(wèi). 遙感數據分類結果的精度分析[J]. 遙感技術與應用, 1995, 10(1): 1724.[10]冉有華, 李文君, 陳賢章. TM圖像土地利用分類精度驗證與評估—以定西縣為例[J]. 遙感技術與應用, 2003, 18(2): 8l86.[11]李全, 王海燕, 李霖. 基于最大似然分類算法的土地覆蓋分類精度控制研究[J]. 國土資源科技管理, 2005, 4(8): 4245.[12]于炳飛, 李增華, 王輝. 基于Lansat7 ETM+的從化市遙感影像信息提取研究[J]. 軟件技術研究, 2006, 21(4): 2430.[13]徐冬青, 唐新軍, 賀錚. 遙感圖像處理軟件ENVI應用中的若干技巧[J]. 黑龍江水專學報, 2004, 12, 31(4): 3638.[14]盧曉峰. 基于RS和GIS的鄭州市城市擴展研究[J]. 河南理工大學學報, 2008, 27(2): 2427.[15]彭光雄, 徐兵, 沈蔚, 等. TM圖像的城鎮(zhèn)用地信息提取方法研究[J]. 遙感技術與應用, 2006, 21(2): 1520.[16]梅安新, 彭望琭, 秦其明, 劉慧平. 遙感導論[M]. 高等教育出版社, 2007, 1.Analysis on the characteristics of TM remote sensing image in Tianshui city in different periods （College of Lifescience and Chemistry, Tianshui Normal University, Tianshui, Gansu,741000）Abstract: Combined with ground data, based mainly on the information of landsat TM images in 1992 and 2001, and centered TianShui Airport, this paper makes full use of software study the main the major surface features of puter classification in some areas in TianShui City, and develop a classification index, using a variety of supervised classification to study the classification map of area. This paper also pares the accuracies of various classifications, finding out that the maximum likelihood method has better classification results. The ground survey inspection shows that sorting results are consistent with the actual land surface conditions, with high classification accuracy. After the linkage of classification diagram in two years, the results show that the study area has experienced some significant changes in land use in recent ten years, for instances, the increases of building land area, reduction of cultivated land area, and reduction of vegetation areas etc.Key words: TM remote sensing。最后非常感謝論文答辯委員會諸位老師在百忙之中審閱我的論文并出席論文答辯會。根據波譜響應特征是進行地物分類的一個主要的方法。分類結果如圖7所示：圖7 馬氏距離法分類結果 The result of Mahalanobis distance classification分類后精度評價結果如下表：表5 研究區(qū)馬氏距離法分類精度評價結果Table5:The evaluation results of classification accuracy of Mahalanobis distance in study areas河流植被居民區(qū)裸地耕地用戶精度（%）河流690132226植被5482341居民區(qū)46147101裸地20386016耕地17113167生產精度（%）總體精度：% Kappa系數：3 分類結果與分析對于分類結果只要采用混淆矩陣進行分類精度評價，將影像中的不同類和感興趣區(qū)進行匹配，計算出混淆矩陣。然后選取感興趣區(qū)，對研究區(qū)選擇不同的方法進行分類。圖1 天水市TM432波段合成遙感影像圖 Synthetic remote sensing image in the