Fast image registration algorithm for automated inspection of laser micromachining

We investigated the possibility of fast quality control of laser micromachining of surface based on optical measurements. The main objective is to match CAD-model and 3D geometry of the machined surface. We found that standard matching algorithms have either low performance or are ineffective at high noise level or at presence of geometrical distortion. The algorithm based on Ciratefi algorithm, which previously wasn’t used for image registration, was developed. The performance of algorithm was increased by using an iterative search of optimum based on image pyramid. The testing of algorithm on height maps of objects formed by laser methods showed its high accuracy and performance.