Research on the identification method of micro assembly part

In view of the assembly process characteristic of non-silicon flat micro parts in MEMS devices, a micro assembly control system is developed based on an established experimental platform. According to the characteristics of the micro parts, micro parts recognition and positioning method is studied, and a new matching algorithm based on shape template is proposed, then an identification module has been developed. Firstly, by acquiring the image of template and target, edge information of the target parts can be acquired after image filtering, binarization, denoising and sub-pixel edge detection. Secondly, edge direction vector is calculated by Sobel filter. Finally, using image pyramid method, image matching speed has been accelerated. Experimental results of typical parts assembly show that the new method has high recognition rate and fast matching speed.