查詢結果分析
來源資料
相關文獻
- 類神經網路與SPOT衛星影像分類之研究
- 類神經網路於多光譜影像分類之應用
- 動態學習法則類神經網路法在闊葉林區土地利用型分類上應用之研究
- 類神經網路應用於遙測影像分類之研究
- A Gary Level Hamming Net for Gray Level Image Classification
- 監督式分類方法於遙測影像判釋之研究
- 以樹狀倒傳遞類神經網路於田埂判釋研究
- Adaptive Filtering Approaches to Multispectral Image Classification
- 非監督式自組織映射圖類神經網路應用於影像分類探討
- Neural Network Procedures for Taguchi's Dynamic Problems
頁籤選單縮合
題 名 | 類神經網路與SPOT衛星影像分類之研究=Study in Neural Network and SPOT Image Classification |
---|---|
作 者 | 蕭百齡; 陳朝圳; | 書刊名 | 國立屏東科技大學學報 |
卷 期 | 9:3 2000.09[民89.09] |
頁 次 | 頁187-197 |
分類號 | 436.718 |
關鍵詞 | 影像分類; 類神經網路; SPOT; Image classification; Neural network; |
語 文 | 中文(Chinese) |
中文摘要 | 本研究以涵蓋墾丁國家公園陸域範圍之恆春半島為研究區,選擇SPOT HRV LEVEL 10影像為分類材料,取其SPOT影像及加入綠度指數(NDVI)之影像,分別檢測最短距離法(Minimum Distance to Means Classifier)、最大概似法(Maximum Likelihood Classifier)及倒傳遞類神經網路(Back-Propagation Neural Network)分類法則,對地表九種分類別之分類準確度及Kappa值。結果以類神經網路分類法對SPOT影像之分類準確度81.2%,Kappa值0.774為最高。而各種分類法則,對不同之地被類別之準確度,卻各有其擅長之處。本研究最後取各分類別準確度較高者為優先,依序疊合出分類準確度最高之地被類別分佈圖,其總體準確度達82.3%,總體Kappa值0.787。 |
英文摘要 | The area around the Kenting National Park in the Hengchuen Peninsula, Pingtung County, is selected as a test site for this study. The SPOT HRV multi-spectral image (original image) and superimposing the original image with the NDVI image were used as the test materials for different classification algorithms; i.e., the Minimum Distance to Means Classifier (MINDIS), the Maximum Likelihood Classifier (MLC) and the Back-Propagation Neural Network (BP). Under this classification, the land images were classified, on the basis of the extent of its ground cover, into nine categories; i.e., high-density forest, mid-density forest, low-density forest, grass land, aquatic land, agricultural land, built-riverbed-rock area, sandy beach, and barren soil after development. The accuracy is evaluated by overall, omission and commission accuracy and the KAPPA. Results indicate that using the BP in classification achieved the highest accuracy in classifying the original image, with an overall accuracy rate of 81.2% and a KAPPA value of 0.774. The ground cover distribution map of the research area generated by using the above six different combinations in the image classification was overlaid with the image in a sequence of their degrees of accuracy, starting with that of the highest degree of accuracy. The result of the total accuracy rate was 82.3% with a total KAPPA value of 0.787. |
本系統中英文摘要資訊取自各篇刊載內容。