Fast Recognition Algorithm for Human Motion Posture Using Multimodal Bioinformation Fusion

To address the problems of low feature extraction accuracy, large bias human motion pose recognition and posture error, poor effect, rate traditional fast algorithm, we propose a algorithm using multimodal bioinformation fusion. First, wavelet packet decomposition with sample entropy is used to extract hand features such as kurtosis, time domain skewness, frequency electromyogram (EMG) integral value mean, standard deviation, interquartile distance leg amplitude. Second, after normalizing two features, set obtained, finally construct model based on fusion, input into model, which completes fusion information by improving typical correlation analysis method, result minimum classifier achieve recognition. The results show that proposed has high accuracy extraction, small recognition, error -0.21∼0.02, always above 95%, practical application effect good.