Infrared Small Target Detection Based on Improved DoG Filter

Accurate and rapid detection of infrared(IR) small target has great significance to military and national defense.It is a hot spot to improve detection rate and reduce false alarm in the field of IR small target detection.Local contrast methods of the human visual system(HVS) have been proven as an efficient method to improve IR image contrast and reduce false alarm rate in recent years.Most of existing algorithms(such as difference of Gaussian(DoG) filter) do not consider directional information.The DoG is easy to misdetect the background edge with a large contrast as the real target.So the quarter of the DoG(QDoG) algorithm based on the DoG filter is proposed in this paper.The QDoG is sensitive to orientations and can better suppress the complex background edges.The QDoG will use the global threshold extraction and local threshold extraction to get a lower false alarm rate and missed detection rate.