Principal curvature for infrared small target detection

• Principal curvature is introduced for infrared small target detection.
• The filter function is the product of positive Gaussian curvature and negative mean curvature.
• An approximate model is provided for optimizing the parameters and verifying the effectiveness of the method.
• The proposed method outperforms other popular methods.

Small target detection in infrared image with complex background and low signal–noise ratio is an important and difficult task in the infrared target tracking system. In this paper, a principal curvature-based method is proposed. The principal curvatures of target pixels are negative and their absolute values are larger than that of background pixels and noise pixels in a Gaussian-blurred infrared image. The proposed filter takes a composite function of the curvatures for detection. An approximate model is also built for optimizing the parameters. Experimental results show that the proposed algorithm is effective and adaptable for infrared small target detection in complex background. Compared with several popular methods, the proposed algorithm demonstrates significant improvement on detection performance in terms of the parameters of signal clutter ratio gain, background suppression factor and ROC.