Abstract: A new method for recognizing 3-D objects and measuring its attitude is introduced in this pa-per. The spatial models of 3-D objects are used in the technique. The artificial intelligence techniques in-cluded in the approach, such as constraint manipulation and geometric reasoning, can deduce and predicateimage models when only the partial 3-D information of object are known. Through the measurement of theimage features back constraints can be set up, by which the 3-D information implied in the image can be in-ferred and eventually it is realized to recognize the 3-D objects and to measure its attitude. The system isbuilt in C programming language on VICOM image processing system. The satisfactory experiment resultsare enclosed.