具有現場膚色偵測以指尖與夾角為基礎的手勢辨識方法

A Gesture Recognition Method based on the Finger Tips and Angles with Local Skin Color Detection

呂緯昱(Wei-Yu Lu)

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)

https://doi.org/10.6841/NTUT.2010.00129

摘要

手勢辨識其應用範圍非常廣泛，常用於人機介面，透過手勢辨識系統將自然的手勢訊號轉變成電腦的控制訊號，進而取代以往電腦與人的互動方式。
本論文先以畫面差異偵測在畫面中有變動的像素，以人臉偵測排除掉人臉之後，對有變動的像素進行較大範圍膚色偵測，保留屬於廣域膚色的部份進行標記演算法，再找出最大的連通區塊後，將此區域設置成感興趣區域。對區域的顏色進行分析統計，以此區域顏色的分佈機率及位置，做為追蹤演算法Camshift 的初始追蹤目標；並以手部輪廓為特徵，利用計算幾何中的凸多邊形來描述手部輪廓，由這些多邊形在空間上的關係，以凹處的夾角為特徵，對手勢進行辨識。

關鍵字

人臉偵測 ； 膚色偵測 ； 物件追蹤 ； 凸多邊形

並列摘要

Gesture recognition can be applied to various fields. Usually, it is applied to Human-Computer Interface. By using gesture recognition, the system could transform the gesture signals into computational control signal to replace the interaction model of Computer-Human.
In this paper, we use frame difference to detect the changing pixel in the frame. After masking the face region by face detection, the changing pixel will be examined by skin color detection to keep the parts for labeling. The maximum connected component is found out to set as the region of interest. Statistics and analysis will be conducted to examine the colors in the component, and the color probability distribution and location will become the initial target of Camshift.
In the part of hand feature, convex hull of computational geometry is used to describe the hand contour. From the relationships among polygons in the space, the gestures are recognized by the angles of concave parts.

並列關鍵字

Face Detection ； Skin Color Detection ； Camshift ； Convex Hull

參考文獻

[19] 張厥煒、許倫維，”即時性連續手勢追蹤與辨識，” 2006數位生活科技研討會，台南 台灣，2006。

[20] 張厥煒、陳良傑，”應用階層式手勢模型的連續手勢追蹤與辨識，” 2008數位生活科技研討會，台南 台灣，2008。

[1] D. C. S. Allison1 and M. T. Noga “Some performance tests of convex hull algorithms,”Department of Computer Science, Virginia Polytechnic Institute and State University, 24061 Blacksburg, Virginia, USA.

[2] E. J. Holden and R. Owens “Recognizing Moving Hand Shapes,”Proceedings of the 12th International Conference on Image Analysis and Processing, 2003.

[4] Hee-Sung Kim, Gregorij Kurillo, Ruzena Bajcsy, “Hand tracking and motion detection from the sequence of stereo color image frames,” In Proceedings of IEEE International Conference on Industrial Technology (ICIT ’08), April 21-24, 2008, Chengdu, China.