Investigations on image fusion techniques for surveillance applications

Abstract

In the recent years, image fusion has become a discipline where a single sensor image is not enough to provide the required information due to the environmental constraint or either by the specific design of the sensors. Image fusion is the technique of integrating two or more images of the same scene into a single image. It contains enhanced depiction of the region of interest or scene than the one provided by any of the source images, thus the fused image assists to better human visual perception or further machine learning process. A human observer may not constantly combine and view a composite image from the multiple image captures. The complementary details of these images should be combined for accurate depiction of the scene. This thesis discusses with the research work involved in the development of efficient fusion techniques for surveillance applications. In surveillance applications, same modality or multi modality images are combined depending on the required application for enhancing image quality and content. The enhanced image is helpful for better analysis and decision making. Fusion also reduces the randomness and redundancy present in the source images. Complementary information obtained from the source images can be analyzed for taking necessary control actions. newline