透過您的圖書館登入
IP:18.191.45.254
  • 期刊

使用Particle Filter之物體追蹤技術

Object Tracking Using Particle Filter

摘要


本文介紹一物體追蹤理論技術:Particle Filter。此技術使用影像顏色及影像紋理為特徵,在決定目標物體的模型後,經過預測,產生一樣本集合,然後接著量測和目標物體特徵值的相似性,以此相似性決定下一個時間的狀態,並更新目標物體的模型。由實驗結果得知,此技術對於快速移動以及大小改變顯著的物體,均有良好的追蹤效果。

並列摘要


We introduce an approach of object tracking: particle filter. We use color and texture of the image as the features. After choosing the model of target, it predicts a sample set of candidates. Then we measure the similarities between the target and each candidate. With these similarities, the next state can be decided and the model of target is updated. Experiments have shown that even if the object moves rapidly or the size of the object changes obviously, the tracking result is accurate.

被引用紀錄


Tsai, C. C. (2010). 智慧車巡航系統之車輛追蹤演算法與架構設計 [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU.2010.00454
羅安鈞(2007)。俯視型行人計數系統〔碩士論文,國立臺灣師範大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0021-2910200810534631

延伸閱讀