An effective space target electromagnetic signature extraction and recognition method is presented. By dealing with the average range which is acquired from diverse attitude angles, target attitude angles' dual range is obtained based on dispectrum transformation. Then the algorithm calculates the eigenvalue of the dual range matrix, and obtains a fusion feature vector that can identify the space object with information fusion technology. Finally, by means of validation with measurement data, this method can recognize and classify space objects exactly, and is better than the usual BP neural network classification method statistically.