A New Registration Method of Infrared and Visible Images based on Improved Edge Extraction and Revised Measure Function

摘要

An improved method based on traditional one of matching feature points is presented in order to realize the registration of infrared image and visual image. In the improved method, firstly, image edges are extracted by the improved edge extraction method and Harris operator is selected to detect feature points. The points prepared for matching are gained by the AND operation between feature points and image edges. Thus the feature points not on the edges can be removed to reduce false match rate and computation of matching feature points. Secondly, in the points prepared for matching, two matching point pairs are extracted by the Revised Measure Function(RMF), then used to calculate registration parameters of rigid transformation model for images registration. The method combines the advantages of features and gray-based method, and has high registration accuracy. Experimental results show its feasibility and practicality, and it is applicable to real-time visible and infrared images registration with any zoom scale, rotation angle and translational quantity.