查詢結果分析
來源資料
頁籤選單縮合
題 名 | 超矩形學習模式應用於遙測影像分類之研究=A Study of Applying Hyper-rectangle Learning Model to the Classification of Remote Sensing Image |
---|---|
作 者 | 陳莉; 盧文鴻; 江柏寬; | 書刊名 | 農業工程學報 |
卷 期 | 47:3 2001.09[民90.09] |
頁 次 | 頁75-85 |
分類號 | 440.98 |
關鍵詞 | 超矩形學習模式; 遙感探測; 地表覆蓋判釋; Hyper-rectangles learning model; Remote sensing; Land cover interpretation; |
語 文 | 中文(Chinese) |
中文摘要 | 近年來許多國家大量使用遙感探測之方法,作為分析及決策規劃的重要資料來源。由於遙測資料涵蓋面積廣泛且具有即時的特性,可做為建立環境資源資料庫之一種有效且使用方便的量測工具。 本研究以人工智慧領域中的超矩形學習模式進行影像分類,此模式藉由不斷的訓練達到回饋修正之目的,將過去曾發生的資料以超矩形的結構加以儲存,不僅節省記憶且更具系統化。 本模式先以數學函數作測試,再應用於曾文水庫集水區進行地表覆蓋判釋。研究中以專家學者所提出的六個分類特徵作為輸入變數,並與倒傳遞類神經網路(BPN)作比較,結果顯示經由超矩形學習模式所得分類結果之正確率比BPN尤佳,因而展現本模式優異的分類效能及高度的學習效果。 |
英文摘要 | Recently, remote sensing methods are widely applied to serve as the important data resource analysis and decision planning comprehensively. The application in this aspect has also become more and more popular. The remote sensing data has both broad coverage area and up-to-date characteristics, and it will be capable of being served as a kind of effective survey tool for constructing of environmental resource database with convenient service. This paper focuses on the image classification by means of Hyper-Rectangles Learning Model based on artificial intellectual field. This model has an experience feedback training procedure. It stores the past data in hyper rectangles structure, not only saving memory but also with systematic significance. In order to verify the classification ability of the model, theoretical math function is tested firstly. Then the model is applied to the land-use classification of Tzengwen Reservoir catchment. Six classification features provided by specialists are used as the input variables. As a result, this model has better accuracy than the BP neural network method. The results demonstrate that the model is a very powerful and efficient tool using in the field of remote sensing image categorization. |
本系統中英文摘要資訊取自各篇刊載內容。