Target detection is the fundamental step in image processing area and designed differently by the object of application. Specially increased demands on FLIR imagery processing require target detection in FLIR imagery. This paper proposes a target detection algorithm in FLIR imagery with large intensity variation. Generally, object in FLIR imagery usually has one fixed heat source. So in object image, each pixel has various intensities but intensity’s sequence sorted by intensity value is usually alike. This paper introduces an intensity ranking-feature that is generated by intensity’s sequence, and a detection algorithm that is influenced by proposed feature and reinforced with averaging. Experiments show that the proposed intensity ranking-feature is robust to intensity variation and that detection algorithm which uses the proposed feature has a high accuracy, robustness to noise and low computational power.
