Please use this identifier to cite or link to this item:
標題: 衛星影像紋理分析在山坡地監測管理之應用
Texture Analysis of Satellite Images for Slopeland Monitoring and Management
作者: 傅桂霖
關鍵字: Slopeland Monitoring;山坡地監測;Texture Analysis;Descriptor;紋理分析;描述子
出版社: 水土保持學系
The spatial distribution of gray-tone of remotely sensed spectral images were calculated in this study for discussing the texture characteristics. The changed site extracted from the images were used as the samples to compute the co-occurrence matrix and the texture descriptors by applying the methods of SGLDM(Spatial Grey Level Dependence Matrix)The remotely sensed images of hillside in Nantou were chosen for discussing the texture descriptors such as angular second moment, contrast, correlation, homogeneity, entropy, sum entropy, difference entropy, cluster tendency and probability to segment images of different classes of land cover.
The result showed that texture analysis can be efficiently used to distinguish and classify the landuse for the better wave band images by screening the suitable texture descriptor and other factors of the images in advance. It shows better results using CON, ENT, HOMO, PRL texture descriptor, but still need to study for further distinguishing the similar texture is legal or illegal.

由於山坡地違規查報工作繁重加上人情壓力等問題,嚴重影響山坡地之監測管理工作。利用SPOT衛星影像,結合衛星定位系統,能有效的監測山坡地之開發行為,避免違規使用,為目前山坡地監測之首要工作。本研究以南投山區為例,針對山坡地土地利用型態變遷分析影像,探討影像灰階值之空間關係,選取不同的變異點,以共生矩陣紋理分析法(Spatial Grey Level Dependence Matrix,SGLDM),計算其共生矩陣及紋理特徵值,探討光譜空間影像之角二次距、對比、相關度、熵、熵總合、熵差、群集傾向度及行長機率等紋理特性,作為辨識變異點地物紋理特徵之參考,藉以了解山坡地變異點之紋理趨勢。分析結果顯示影像紋理可作為地物類別辨識之先驅工作,於多譜影像中能篩選較佳的波段使辨識分類更有效率,在分析中CON、ENT、HOMO、PRL,這些紋理描述子在一般條件下,都可以達到較佳的辨識結果,惟地物種類呈像紋理極為相似時,如何判定為違規使用或非違規使用仍須其他方法輔助比對。
Appears in Collections:水土保持學系

Show full item record
TAIR Related Article

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.