基于手指角度特征的静态手势识别算法

Static Gesture Recognition Algorithm Based on Characteristics of Finger Angle

提出一种基于手指角度特征的静态手势识别算法。以指尖到手掌中心的连线构成手势骨架,计算手指间的角度;以角度的大小和指间数量进行分类,把手势定义为一、二、三、四、五、六、七、八、九等9种。该算法不受手势的方向和尺度的影响,仅通过判断手指间的角度大小来识别手势。对900幅手势图进行分析识别的实验结果表明:该算法正确率达96.8%,准确性高;平均用时不超过0.05 s,实时性好。

This paper puts forward a static gesture recognition algorithm based on finger angle characteristics. With the fingertips attachment the authors constitute the gesture framework by the line from finger to the center of palm, and calculate the angle between your fingers. To view the number and size of classification, the gestures are defined as one, two, three, four, five, six, seven, eight, nine. The algorithm is not affected by gestures to the direction and scale, and only by judging the fingers point size to recognize hand gestures. In experiments, 900 gesture image recognition are analyzed, the accuracy rate is 96.8%, the average is less than 0. 05 sec- onds, which indicate that the algorithm is of high accuracy, good real-time performance.