A Kernel-Based Nonparametric Regression Method for Clutter Removal in Infrared Small-Target Detection Applications

Infrared small target detection based on directional zero-crossing measure

“…In order to further demonstrate the superiority of the proposed method to the other five methods, the receiver operator characteristic (ROC) curve [9] is applied for above eight different scenes, respectively. The ROC curve represent the relationship between the detection probability P d and false alarm rate P f , where P d is defined as the ratio of the detected target numbers to the real target numbers and P f is defined as the ratio of false alarm numbers to the total images.…”

“…They preserve the good performance of 2D LMS filter while using the edge information to estimate background. Gu et al [9] proposed a kernel-based nonparametric regression method to predict the background. In this method each pixel is represented by using a linear mixture model.…”

Small target detection using edge-preserving background estimation based on maximum patch similarity

“…Another important indicator to demonstrate the effect of methods is receiver operating characteristic (ROC) curve, 7 which is an intuitionistic display of the relationship between the false alarm rate and detection rate in the detection results. The ROC often determined by the clutter in the target image.…”

Kernel regression in mixed feature spaces for spatio-temporal saliency detection

“…In [29], kernel regression is adopted to estimate the cluttered background in infrared images so that the background part can be suppressed and the small targets can stand out. Kim et al deduce a refined weighted center-surround saliency measurement for noisy image using kernel regression [30], and they achieve impressive experimental results.…”