查詢結果分析
來源資料
相關文獻
- 應用類神經網路於氣象衛星圖中雲之識別
- 運用奇異值分解法於影像處理
- Image Compression Using Sub-Band DCT Features for Self-organizing Map
- A Facial Expression Recognition System Using Facial Feature Extraction and Artificial Neural Networks
- Neural Network Procedures for Taguchi's Dynamic Problems
- A Fast and Efficient Competitive Learning Design Algorithm Based on Weight Vector Training in Transform Domain
- 專家系統振動訊號圖型判別之研究
- 反傳遞模糊類神經網路於流量推估之應用
- 類神經網路(Neural Networks)的種類及其在影像處理上的應用
- C++Fuzzy類神經網路物件導向發展系統之建立
頁籤選單縮合
題 名 | 應用類神經網路於氣象衛星圖中雲之識別=Detection of Cloud Data from GMS-5 Satellite using Artificial Neural Networks |
---|---|
作 者 | 劉仁俊; 莊金曉; | 書刊名 | 大氣科學 |
卷 期 | 25:2 1997.06[民86.06] |
頁 次 | 頁195-210 |
分類號 | 328.62 |
關鍵詞 | 類神經網路; 雲識別; 奇異值分解; K-L轉換法; 特徵擷取; 自組織特徵映射; Artificial neural network; Cloud detection; Singular value decomposition; Karhunen-loeve transform; Feature extraction; Kohonen self-organizing feature maps; |
語 文 | 中文(Chinese) |
中文摘要 | 雲的變化為許多大自然現象的指標,因而有許多研究是投入在雲的分析上。本文 提出以類神經網路的方法來進行衛星圖中雲的識別,期望能突破傳統所利用的統計或門檻法 則而得到更好的結果。 處理的步驟是先以奇異值分解法或 K-L 轉換法來擷取影像之紋理特 徵,有效地利用最少的資訊來表達圖像之意義;再將擷取出來的特徵送入一屬於自組織特徵 映射之 KOHONEN 類神經網路來作自動化雲之識別。 類神經網路的技術擁有普及性、容錯能 力和並行處理的優點,因此大幅提升了影像處理之速度和準確性。 本文是以 GMS-5 紅外線 衛星雲圖來作驗證,結果中清楚的說明此方法有效地彌補了傳統方法之缺點,達到相當高的 辨識率和低的誤判率,且對雲團能作到某種程度的分類,對於未來的大氣科學的研究發展應 有相當的助益。 |
英文摘要 | The variations of clouds indicate many global phenomena. Thus many efforts have been dedicated on cloud analysis. In this research we propose a method using artificial networks for cloud detection in the hope to obtain better results as opposed to many conventional methods, which utilize thresholding or statistical type of approaches. The process of this research first uses singular value Decomposition and Karhunen-Loeve transform to extract image textural features in addition to mean value methodologies. The extracted features are then presented of a self-organizing feature map or Kohonen network for automatic detection and classification of cloud areas. Neural network techniques have the capabilities of generalization, fault tolerance, and parallel processing which can increase the speed and accuracy of the process. We used GMS-5 satellite data in the infrared (IR) rage for demonstration. The results clearly indicate the superiority of this approach under many situations which are considered difficult for the conventional method. High detection rates and low false alarm rates were achieved. the processed results also show some interesting classification capabilities which can facilitate future studies on this topic |
本系統中英文摘要資訊取自各篇刊載內容。