Infrared imaging is the order of the day. Infrared imaging finds its applications extensively in defence related areas. Pedestrian detection in IR image is a challenging problem. This paper addresses this problem using fuzzy-based enhancement followed by top-hat transform. The proposed method is adaptive in the sense that the parameters required for the process are calculated adaptively based on the image characteristics. Experimental results on the standard dataset show the robustness of the proposed method with 90% detection rate of pedestrians under different conditions and outperform the other existing methods. © 2017 Inderscience Enterprises Ltd.