Image Fusion Performance-gains at Different Fusion Levels

Abstract

Image fusion is a branch of multi-source information fusion, which plays an increasingly significant role in the military field. Since the environment is full of many interference factors, including light, dust, etc., the target object cannot be clearly identified. Image fusion based on visible image and infrared image is attractive and promising for the object detection applications. This paper analyzes and compares pixel-level, feature-level and decision-level image fusion, and summarizes the performance-gains of image fusion at different levels with examples. It is concluded that pixel-level fusion can be used to process more delicately than feature-level fusion, and the result of feature-level fusion is more delicate than decision-level fusion. Furthermore, we conclude a creative idea, that is pixel-level and feature-level methods can be combined in the future.