基于Kinect的手势图像识别研究

Research on Gesture Image Recognition Based on Kinect

针对在人机交互领域中基于视觉的手语识别的识别率低、实时性差等问题,采用Kinect深度摄像机对手语识别进行研究,并对不同的手语分类方式进行了分类和比较。首先通过Kinect设备采集手语图像,并提取出手部感兴趣区域,然后对手部图像进行预处理,得到手部边缘图像。随后用Hu矩方法对手部边缘图像进行特征提取。分别应用支持向量机、神经网络和随机森林三种分类方法对得到的数据进行分类和比较。最终结果表明,实验的平均识别率达到97.4%,同时发现在小样本情况下支持向量机方案的识别率最高,而随机森林的实时运行效果最好。

Aiming at the problems of low recognition rate and poor real-time performance of visionbased sign language recognition in the field of human-computer interaction, Kinect depth camera was used to study sign language recognition, and different sign language classification methods were classified and compared. Firstly, sign language images are acquired by Kinect equipment, and regions of interest in the hand are extracted. Then hand images are preprocessed to obtain hand edge images. Afterwards, Hu moment method is used to extract features from the edge image of the hand. Support vector machine,neural network and random forest are respectively used to classify the obtained data to obtain comparison results. The final result shows that the average recognition rate of the experiment is 97.4%, and at the same time it is found that the recognition rate of the support vector machine method is the highest in small samples, while the real-time running effect of the random forest is the best.