一种高效人体行为特征提取方法

An Efficient Method for Human Behavior Feature Extraction

针对现有人体行为识别方法中特征提取的计算负荷重的问题,提出一种新颖的基于光流的特征描述方法。由于背景的光流包含一定的噪声,提出几种方法来减弱噪声影响。首先,将固定数量的帧捆成一个块并提取光流的中值,其次,对基于整体量级的直方图进行正则化,最后在频率域进行低通滤波。利用傅立叶变换将时间域转化成频率域,保证时间变换时特征的不变性。在构造光流直方图时在直方图中分配行为的方向,因此可以保证行为表示时方向的不变性。最后,在KTH实验数据集与自建数据集中进行对比实验,结果表明该方法的有效性。

In order to reduce high-computational of the existing human behavior recognition methods,proposes a novel feature based on optical flow for action recognition.The feature is quite simple and has much lower load than the existing features for action recognition algorithms.Since raw optical flow is noisy on the background,several methods for noise reduction are presented.Firstly,bundles up the fixed number of frames as a block and take the median value of optical flow(median flow).Secondly,takes the normalization of histogram depending on the total magnitude.Lastly,does low-pass filtering in frequency domain.Converting the time domain to frequency domain based on Fourier transform makes the feature invariant to shifted time duration of action.In constructing the histogram of optical flow,aligns the direction of an action so that we can get direction invariant action representation.Experiments on benchmark action dataset(KTH)and our own dataset for smart class show that the proposed method gives a good performance comparable to the state-of-the-art approaches and has applicabili?ty to actual environments with smart class dataset.