An efficient method for abnormal behavior recognition based on feature merging

In order to solve the problem of poor real-time recognition and low recognition accuracy in the current smart home environment, this paper introduces the idea of feature merging into the MCRF(Multiple Conditional Random Field Ensemble) model, which effectively reduces the number of observed features, improves the recognition rate of abnormal behavior based MCRF model. The results show that the MCRF anomaly recognition based on feature merging can recognize the anomalies in real time and accurately. The algorithm, which sends the detected abnormal results to the children or medical staff through the mobile terminal in real time, can greatly reduce the harm caused to the elderly living alone due to unexpected accidents (such as falls and fainting).