Research on Space-based Sequence Image Recognition Based on Markov Model

Space target recognition is a challenging research topic. Enhancing the ability of space surveillance is the key to space control and space utilization. At present, space target recognition and cataloging are mainly based on synthetic aperture radar(SAR) image, inverse synthetic aperture radar(ISAR) image and ground optical image. In this paper, we proposed a space target recognition algorithm based on Markov feature of Space-based optical sequence images. By establishing the 3D model and orbit model of the space target, the geometric relationship between the space target and the observation satellite is simulated, and the simulation sequence image of the space target is obtained. The image noise is removed and the edge is extracted. Then the Markov feature is extracted after segmentation, and the Euclidean distance of Markov feature matrix of images is calculated. The length of image sequence L, the coincidence rate k of the nearest distance image block sequence and the nearest distance image block sequence of the candidate image block are trained step by step, and then the space target recognition is carried out by using this parameter. The simulation results show that this method can effectively recognize space targets, and can take into account the target recognition rate and accuracy. The recognition rate is higher than that of ellipse features, information entropy features, histogram features and wavelet texture features.