Improved infrared image binarization by using a combination of global method and local method

In this paper, we present a novel binarization method for infrared images. Such images are characterized by low contrast, fuzzy edges and more complex with their histogram distributions. If we only apply a global or local threshold, these images will lose a lot of edges and detail information, or ignore the contours of objects. To address this issue, the proposed method combines the advantages of global method and local method; it not only retains rich details, but also the profile information. The experiment results show that our method has a better performance than Ostu binarization approach (global) and Bernsern binarization approach (local) for infrared images.