文章詳目資料

地理學報 CSSCIScopusTSSCI

  • 加入收藏
  • 下載文章
篇名 區別分析於衛星影像土地利用分類之應用 以宜蘭沿海鄉鎮為例
卷期 29
並列篇名 Discriminant Analysis of SPOT Imagery for Land Use Classification A Case Study in I-Lain
作者 黃誌川徐美玲朱子豪
頁次 107-120
關鍵字 discriminant analysisland-use classificationunsupervised classificationISODATAkappa區別分析土地利用非監督性分類ScopusTSSCI
出刊日期 200105

中文摘要

本硏究利用衛星影像結合丨SODATA非監督性的土地利用分類及區別分 析,保留監督性分類與非監督性分類的優點,以宜蘭沿海五結鄕和頭城鎭爲 硏究區,進行模式的率定和檢核。並比較以SPOT衛星影像進行第一級的土 地利用分類(classl)與以地物特徵的分類的適用性。結果顯示,本分類模 式可以避免非監督性分類結果與實際土地利用配對的困擾及面積上的錯估, 不過,當某類土地利用總面積佔絕大部分時,此模式的整體預測結果雖可達 到相當不錯的正確率,但是對個別土地利用的分辨則不一定理想。此模式分 別以現行的第一級土地利用分類(classl)的類項和以地物特徵的分類類目 進行推測。整體而言,此模式在分辨地物特徵的類目有較佳的表現,不過個 別土地利用類目的預測正確度隨土地利用的種類而有差異。對於包含不同地 表影像特徵的種類,如「農業用地」,預測正確率不甚理想。但是對於具有 較一致之影像質地的土地利用,如養殖或水道等,則有相當不錯的預測結果。 因此,結合ISODATA和區別分析以SPOT衛星影像先就小面積樣區進行區 別函數的率定,再以之進行鄰近大範圍地區的土地利用分類應爲可行的,將 有助於土地利用變動的即時監測。

英文摘要

This study combined ISODATA and discriminant analysis for land classification, so as to retain the advantages of both supervised and unsupervised methods, and a case study was undertaken in I-Lain coastal zone. It showed that the proposed classification method avoided the cumbersome procedure of designating land use types. However, with a predominant land type in the study area, the proposed method resulted in a high overall accuracy rate, but poor performance in discerning minor land types. This study compared research results based respectively on the classification scheme presently adopted in Taiwan and a classification derived from land cover characteristics. It was found that the overall performance was better when the land cover classification scheme was used. The accuracy rates in discerning individual land use varied with different land use types. For example, those including land cover types, such as agriculture, were not classified as accurately as others. However, land use with more homogeneous imagery texture, such as aquaculture and waterways were estimated accurately. We concluded that our method which combine unsupervised ISODATA and discriminant analysis of SPOT imagery has great potential for monitoring land cover changes.

相關文獻