An approach for locating surface mounted devices based on phase correlation

This paper presents a model-based method for localization and measurement of various surface mounted devices. The proposed method uses all the image information and without feature extraction, it can be very accurate, efficient and robust. In our algorithm, the phase correlation is used for scale, rotation and translation estimation. In the part of scale and rotation estimation, we implement log-polar transform of magnitude spectrum; the scale and rotation are extracted by using phase correlation in log-polar space. In the part of translation estimation, we firstly generate the ideal model of surface mounted devices and adopt scale and rotation transforms to the model, and then the translation estimation can be obtained by phase correlation. Finally, representative experimental results on ideal and real gray-level surface mounted devices images show that the proposed approach is applicable to Scale-roto-translation estimation within a specified degree of accuracy. The proposed method can become an effective technique to inspect all kinds of surface mounted devices with arbitrary shapes which is suitable for a pick-and-place system used by semiconductor assembly.