Pedestrian Detection in Infrared Video Images

Abstract

Abstract. This paper introduces a method to detect pedestrians in infrared video images and studies its application into intelligent video surveillance. As part of a far infrared thermometric system in public places, our algorithm is aimed at the detection of human bodies in fixed area. According to the characteristics of infrared imaging technology, pedestrian detection depends on the difference of infrared radiation between the human body and the environment. Detection is achieved by the extraction and recognition of simple features of pedestrians which reduces the complexity of tradition methods and avoids using quantities of samples for training classifier. Our method not only reduces noise effectively， but also reaches high detection rate(DR) and low fault detection rate(FDR). Experimental results are presented to demonstrate the excellent performance of our algorithms.