适用于老年公寓的动作识别设计方法

为帮助老年公寓监护人员及时发现老年人摔倒等动作,提出了一种基于视频监控的动作识别方法。对监控视频,首先通过基于HSV空间的混合高斯背景建模方法提取前景图像,然后利用所提出的运动特征和形态特征相结合的方式进行特征提取,最后通过具有高斯输出的HMM模型实现动作类型的识别。提出的方法能够适应光照变化影响,对不同动作的动作方向和动作幅度变化具有很好的鲁棒性,实验中动作的识别准确率达到90%。结果表明,本方法能够满足老年公寓动作识别的基本要求,具有一定的实用价值。

Design method for action recognition applied to senile apartment

To help nursing staff in senile apartment find the elderly fall and other actions timely,an action recognition method based on video surveillance is proposed.Firstly,the foreground images are extracted by the GMM background modeling method in HS color space.Feature extraction is performed by combining the motion features and morphological features.And action recognition can be achieved by HMM with Gaussian output.The method proposed in this paper can adapt to the changes of illumination.The method also has good robustness to the change of motion direction and motion range,and the recognition accuracy rate reaches 90％.The result shows that the method can meet the basic requirements of action recognition and the method has certain practical value.