Abstract: For the purpose of accurately detecting and recognizing objects containing complex edges,an object recognitive method is proposed based on a boundary fragment training model.Firstly,the method forms a weak classifier by extracting edge segments,and then the obtained weak classifier is further trained into a strong one by using an AdaBoost algorithm and is used to detect and recognize space objects.The simulation experiments dedicate that the method has an accurate result for detecting visible space objects,especially for the objects with complex edges.