In view of the issue of automatic target recognition met by infrared imaging guided missiles in the war, an automatic target recognition algorithm which uses several 213 images to recognize 3-D targets is put forward and a Support Vector Machine is used as a classifier. The simulation experiment results have shown that the algorithm can recognize the target at any angle in the 3-D space successfully and can solve the 3-D target recognition problem which is difficult to be solved by many 2-D recognition algorithms. At the same time, compared with the traditional recognition algorithms, this algorithm has a higher target recognition ratio.