查詢結果分析
來源資料
頁籤選單縮合
題名 | 具備自動追蹤與光學放大功能的非校正式智慧型監視系統=An Automatic Pan/Tilt/Zoom Tracking Mechanism for a Calibration-free Intelligent Surveillance System |
---|---|
作者 | 彭守道; 許毅然; 蔡明宏; | 書刊名 | 南臺科技大學學報 |
卷期 | 32 2007.12[民96.12] |
頁次 | 頁11-25 |
分類號 | 448.94 |
關鍵詞 | 智慧型監視系統; 影像伺服系統; 徑向基底函數神經網路; 主動式視覺法; Intelligent surveillance system; Visual-servoing; Co-operative learning neural network; Self-organizing maps; Radial basis function neural network; |
語文 | 中文(Chinese) |
中文摘要 | 本研究旨在發展一類兼具目標物追蹤與自動光學放大(Optical zoom)的智慧型監視系統。研究中採用了固定式攝影機(Static camera)資訊引導移動式攝影機(moving camera)的整合方案,以強化移動式攝影機跟拍目標物時的穩定性,並取得高解析度的目標物影像。在方案中,固定式攝影畫面的『目標物影像框』與移動式攝影機『Pan/Tilt/Zoom』參數間的映射關係,是藉由訓練一組內含多個徑向基底函數神經網路架構所建立。本研究的特色在於能運用類神經網路架構整合PTZ攝影機的Pan/Tilt/Zoom三項功能。從實驗中可發現,此智慧型監視系統經過充份的學習次數後,在PTZ攝影機的導引與自動光學放大上皆有一定的準確度。 |
英文摘要 | This paper presents an intelligent surveillance system that is able to simultaneously perform auto-tracking and auto-zooming for the suspect targets. This surveillance system uses image information from static camera to guide PTZ camera for enhancing its tracking stability. The visual-servoing relationship between the static camera (Image data) and PTZ camera (Pan/Tilt/Zoom parameters) is constructed by using the co-operative learning neural network in which the supervised and unsupervised learning neural networks are coordinated together. Experiments show that, after a sufficient number of times training, the system can perform well as expected in auto-tracking and auto-zooming. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。