In modern aerospace, according to star pattern distribution status obtained by star sensors, the attitudes of space vehicle are obtained. It has widely application for high reliability, confidentiality, accuracy and widely suitable range. Firstly, traditional recognition algorithm is introduced. And then, improved algorithms referred in recent years are summarized. And improved triangle, vector index, string index, parameter identification without calibration and based on neural network and support vector machine algorithm are respectively introduced in detail. These methods in the area of recognition speed, success rate and real-time ability are improved to some extent. At the condition of low false alarm, the robustness of the algorithm is improved. Finally, the shortcomings of current star sensor identification methods are summarized and the development trends are forecasted.