In the field of computer vision and automatically intelligence, the approach to recognize an object can be regarded as the task to determine the similarity between the unknown object and the known reference object. This concept is the fundamental theory of generic object recognition system. There is several problems imposed in the object recognition task. The problem of different view point projection is one of the major difficulty in the shape recognition and pose estimation . Besides that , the performance of object recognition algorithm in terms of success rate and access time can be used to evaluate the robustness and the efficiency of the object recognition approach. In this thesis, particle swarm optimization technique (PSO) integrated with migrant principle (Mig) is proposed to employ into the affine invariant object recognition approach. There are two major objectives in this suggested approach. The first objective is that this object recognition can solve the difficulties caused by various forms of unknown image scenes especially the broken boundaries object, noisy distorted object and occluded objects in different view point projection. The second objective is the performance evaluation of object recognition. One of the major difference between the matching in simple closed boundaries objects and complicated noisy broken boundaries objects is that the complexity of matching algorithm and the searching space will be increased in the later case. The enhancement result in the rate of success match and the access iteration can demonstrate that the applicability of proposed algorithm is significantly strengthened. It is also benefit to the feasibility and the robustness to object recognition approach.