基于Kinect深度数据的人物检测

摘要

In order to realize the computer vision-based motion recognition,human detection must be realised first,and in the past few years,there are a lot of research and methods on human detection. And in this paper,introduce a brandnew human detection method based on the depth data acquired by Microsoft Kinect as human scene collector,and then it can remove background depth area from human in the image,remaining human depth area by processing the acquired depth data. It was verified in the matlab experiment platform and showed fine results in which background can be removed totally,and the calculation complexity of this method is low. In conclusion,the human detection method based on depth data can achieve human detection well.%　　为了实现基于计算机视觉的人体动作识别，首先必须实现人物检测，过去的几年中，有很多关于人物检测的研究和方法。而在文中，介绍了一种较新的人物检测方法，该方法利用微软Kinect体感器作为人物场景采集器，将其采集到的深度数据进行处理，将人物与背景分割，去除背景深度区域，保留人物所在深度区域，进一步检测出人物。该方法已经在matlab实验平台中得到验证，取得了良好效果，背景能够被完全去除，并且该方法计算复杂度低。由此得出结论，基于Ki-nect深度数据的人物检测方法能够很好地实现人物检测。