查詢結果分析
來源資料
頁籤選單縮合
題名 | 植基於隱匿式馬可夫模式之軍事影像分類法研究=A Study of Hidden Markov Model Based Methodology for Military Imagery Classification |
---|---|
作者 | 李奇奎; 左杰官; Lee, C. K.; Tso, Brandt; |
期刊 | 國防管理學報 |
出版日期 | 20071100 |
卷期 | 28:2 2007.11[民96.11] |
頁次 | 頁23-31 |
分類號 | 312.1 |
語文 | chi |
關鍵詞 | 內容服務; 影像擷取; 隱匿式馬可夫; 軍事圖資; Content service; Image retrieval; Hidden markov model; Military image; |
中文摘要 | 影像擷取在近年來和文字資訊一樣成為網際網路普及之後的一門顯學,本研考將針對軍事圖資進行主要研究對象,逐步蒐集網路上具授權性之軍事圖資,進行多解析度特徵值擷取及正規化等前置處理後,將配合開發之觀測密兖及序列調整式隱匿馬可夫模式(HMM)進行圖像自動歸屬斤研究,以軍事圖資為例,可區分為飛機、船籃、潛艇、坦克等。經研究發現,本研究所研提之觀測序列調整法可成功將二維影像納入一維隱匿式馬可夫模式中,本研究完成後將可成為我國防軍情蒐集重要之一環。 |
英文摘要 | Image retrieval has shared the same importance as information extraction in recent year to become a significant research ground. This research therefore concentrates on developing new methodologies in image retrieval field. This study will be focused on military imagery. The imagery are obtained from the web with authorization. A pre-processing step is applied to these imagery to extract image characteristics and to perform image normalizations. In this research, we will concentrate on developing observation-sequence and observation-density methodology to automatically achieve imagery classification. In the case of military imagery classifications, the imagery are classified into airplane, boat, submarine, and tanks. The results show that the developed observation-sequence approach can successfully arrange two dimensional imagery into one-decisional hidden Markov model. This research once completed will become an important part in national defense intelligence collection. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。