Human body behavior recognition method based on time-frequency domain information

The invention discloses a human body behavior recognition method based on time-frequency domain information. The method comprises the following steps: 1, acquiring a three-axis acceleration sequence sample of a human body behavior and preprocessing the three-axis acceleration sequence sample; 2, segmenting the preprocessed human body behavior three-axis acceleration sequence sample into a training set and a test set; 3, constructing an LSTM-CNN network model; 4, fusing the feature vectors extracted based on the time-frequency domain and carrying out human body behavior recognition; and 5, updating LSTM-CNN network parameters through a gradient descent back propagation algorithm. According to the invention, the accuracy of human behavior recognition can be effectively improved. 本发明公开了一种基于时频域信息的人体行为识别方法，其步骤包括：1、采集人体行为的三轴加速度序列样本并进行预处理；2、将预处理后的人体行为三轴加速度序列样本分割为训练集和测试集；3、构建LSTM-CNN网络模型；4、融合基于时频域提取的特征向量并进行人体行为识别；5、通过梯度下降反向传播算法更新LSTM-CNN网络参数。本发明能够有效提高人体行为识别的准确率。