以無軀幹方式分析人像活動

Abstract

對於電腦視覺的這個領域來說，人像活動的辨識一直都是被熱中研究的議題。在以往大多數的研究方法中，通常是將人體整體的資訊全部考慮進來。然而我們可以觀察到，其實只要得知頭部與四肢等較關鍵部位的變化，就能夠大略的了解人像的動作，而足以應付許多的實際運用。本論文即是提出一個新的方法，用來在一連串的影像中，持續的追蹤頭部與四肢末端的變化，然後在無軀幹資料的情況下，建立簡易的圖形來呈現人像活動的情況，亦可驗證實驗的結果。以這種方式來進行監控時，由於不需要直接檢視受監控者，故可以保護受監控者的隱私，同時因為僅具有頭部與四肢末端等的資料，故極少的資料更可以減輕需要進行即時遠端傳輸監控系統的頻寬負荷。

Human activity recognition is a popular topic in the field of computer vision. While most analysis algorithms take the whole human body into consideration, the movements of merely the head and limbs are often informative enough in many practical applications. In this paper, a novel approach is proposed to track these extruding parts of a human body in consecutive images. Accordingly, a simplified torso-less pattern of gesture is proposed to represent human activities and the effectiveness of the proposed approach can be observed from some experimental results. Such a representation can not only ensure the privacy of the person being monitored, but is also suitable for real-time surveillance based on bandwidth-limited communication since only a very small amount of data are used if compared to conventional approaches.