基於可適性膚色切割之雙手手勢辨識系統

A Two Hands Gesture Recognition System Based on Adaptive Skin Color Model

張竣皓(Chun-Hao Chang)

指導教授 ： 張厥煒

國立臺北科技大學/電資學院/資訊工程系研究所/碩士(2010年)

摘要

近年來有越來越多人發展出使用手勢遙控周邊的電器，進一步拉近了人與機器之間的溝通與互動。而在手勢辨識系統中，光線所造成的反光或陰影將會大大的影響手勢切割品質，所以要如何在複雜背景中依然能有好的手勢切割品質，到現在是一項仍需研究的議題。 本論文提出一種可適應各人種膚色的動態手勢辨識技術，不需人工加以訓練的方式，在複雜背景環境下系統能自動取得使用者膚色取樣，並模擬出使用者的手勢膚色空間。首先，利用Haar-Like特徵找出雙手的所在位置，並判斷是否為我們所設定的起始動作。接著將偵測到的手進行統計與分析，藉以建立出使用者個人的膚色模型。利用此膚色模型切割出確切的手形出來，並分別進行靜態手勢與動態手勢辨識，而本論文是以iPhone的多點觸控作為模擬的目標，在同時偵測到雙手後，對其指尖的動態軌跡作辨識，以達到多點觸控的模擬。

關鍵字

類赫爾偵測 ； 可適性膚色模型 ； 手勢辨識

並列摘要

In recent years, more and more appliances are developed using hand gestures as remote controls. In gesture recognition systems, because the light reflection or shadow will affect on the quality of segmentation, how to maintain a good quality of segmentation in a complex background is still an issue that needs to study.
In this paper, we propose a hand gesture recognition technique that can be adapted to various skin color of different scenes without any training. At first, we use the Haar-Like feature to identify the location of user’s hands and determine if it matches the initial action. Then, we collect the hand’s color space to build an user’s skin color model. With this skin model, it could extract the shape of hands, and recognition the static and dynamic hand gestures separately. This paper focuses on a simulation of iPhone’s multi-touch technique. After detecting hands, we will analyze the trajectory of fingerprints in order to achieve the multi-touch simulation.

並列關鍵字

Haar-Like Detection ； Adaptive Skin Model ； Gesture Recognition

參考文獻

[1] Junwei Han, G.M Award, A. Sutherland, Hai Wu, "Automatic Skin Segmentation for Gesture Recognition Combining Region and Support Vector Machine Active Learning," Proc. of Int. Workshop on Automatic Face and Gesture Recognition, April 2006, pp.237-242.

[2] L. Gupta, S. Ma, "Gesture-based Interaction and Communication: Automated Classification of Hand Gesture Contours," IEEE Trans. on Systems Man and Cybernetics, Part C, vol.31, Feb. 2001, pp.114-120.

[4] H. Eglowstein, "Reach Out and Touch Your Data," Byte, vol. 15, 1990, pp.283-290.

[5] T. Takahashi and F. Kishino, "A Hand Gesture Recognition Method and Its Application," Systems and Computers, Japan, vol. 23, No.3, 1992, pp.38-48.

[6] T. Lorige and T. Kono, "Human-interface by Recognition of Human Gestures with Image Processing Recognition of Gesture to Specify Moving Direction," IEEE Int. Workshop on Robot and Human Communication, 1992, pp. 105-110.