A Physical Information Neural Network Algorithm for Centroid Solution of Cross Strip Anode Detector

With the rapid development of aerospace technology, the detection ability of remote dim target has become a key task. Cross strip anode (XS) photoelectric imaging detector can be used to capture weak target signal, has become a powerful tool for remote target detection. Due to the complex and changeable space environment, various noises will inevitably enter the detection system, thus interfering with the accurate detection of targets and reducing signal resolution. In order to improve the precision of calculating the centroid of cross strip anode under strong noise, a model based on physical information neural network (PINN) is proposed. Compared with the traditional centroid method, the physical information based neural network method does not need to set specific parameters according to the type of noise, so it is hardly affected by noise interference and can obtain high precision centroid localization.