在這個數位資訊爆炸化的時代，越來越多廠商開發出產品來提升我們日常生活的品質與方便性，近幾年來手勢控制的發展使人們逐漸從傳統鍵盤與滑鼠等操作介面，轉變為更符合直覺操作模式的遠距手勢控制，然而使用視覺影像來做手勢辨識還是有很多挑戰需要克服。本論文提出一個基於三維手勢辨識系統硬體架構，本系統使用低成本的雙攝影機來計算深度資訊，不但可以提供整張影像的深度資訊，也能在嚴峻複雜的背景中切割出感興趣物件來使系統辨識。 目前大多的手勢偵測研究採用膚色或運動量值做為前處理步驟，但只透過膚色資訊和運動量無法在複雜的環境下還能使系統正常的運作，本論文提出一個自適應性的膚色深度過濾，此方法可以有效分離出系統感興趣的部分「手」，並透過物件標籤法來去除剩餘雜訊並計算出手部的座標資訊，最後透過深度及座標資訊完成整體的手勢辨識，此系統包含一種靜態手勢與兩種動態手勢，且透過FPGA的實做驗證讓使用者更能體會到人機互動的概念。 ;In this digital information explosion era, more and more products are proposed to enhance the quality and convenience of our life. As the growing number of gesture control systems have been developed, keyboard and mouse have been replaced to remote control products which are more intuitive for users gradually. However, vision-based hand gesture recognition is still a challenging problem to overcome. In this paper, an architecture system was proposed with dual-camera to construct the depth map and segment the interesting object under complicated environment to recognize the dynamic hand gesture. Most of hand detection methods adopt skin filter or motion filter as the preprocessing. However, it can’t still segmentation the interesting object correctly under some complicated environment. The proposed system adopt adaptive depth filter which can separate foreground to segment the interesting object. We also proposed dynamic gesture recognition by using depth and coordinate information. The system contains one static gesture and two dynamic gestures and implement in FPGA which make users know more about HCI systems.
