Abstract: Glory is an important meteorological resource for tourism. Due to the lack of automatic observation technology, the difficulty of artificial observation and the missing measurement, there is a lack of observation data in Glory, and the resource evaluation is almost blank. According to the principle of glory observation of “three points of the sunlight-the line of sight of the human eye or the camera-glory（imaging area of cloud and fog）must be distributed in a line”, a real-time automatic observation glory technology based on camera is developed, that is, every 10 seconds, the camera angle is adjusted to the current solar azimuth angle and height angle, so that the camera orientation is kept parallel with the changing sunlight, and the image is captured and saved. The glory recognition model is established by using the deep learning technology, and the real-time photos are automatically recognized. The application shows that the technique can solve the real-time observation problem of glory, and the observation results (photos) are the important basis for the evaluation and census of glory tourism resource, which can provide technical support for glory tourism resources development and the viewing of glory.