A multi-objective anomaly abnormal detection method based on the infrared and optical image fusion

Due to the impact of COVID-19, people are required to wear masks and take the temperature in most public places. A multi-objective detection method for abnormal people is proposed to solve the problem of screening for abnormal people in high crowd density effectively. Based on the infrared and visible light fusion image, the method screens abnormal people with high body temperature or without wearing masks. Firstly, the Maximum A Posterior（MAP） algorithm is used to compensate for the temperature of the acquired infrared grayscale image. Then the compensated infrared image is scaled, translated, colored, and fused with the visible light image through the screen to show the change and situation of body temperature directly. After that, machine learning plays a significant role in recognizing faces and determining whether the target is wearing a mask. Experiments show that this method has a high success rate for mask recognition at different angles and distances, and can effectively screen abnormal persons with high body temperature.