Infrared Small Target Detection Based on Derivative Dissimilarity Measure

Robust detection of a small target and effective suppression of a complex background are important issues in infrared searching and tracking systems. To address these issues, in this paper, a derivative dissimilarity measure (DDM) is proposed for infrared small target detection. First, derivatives of several directions from an infrared image are computed based on the facet model. Then, the derivative feature (DF) of a small target is extracted from each direction. Taking the continuity of background and discontinuity of small targets into consideration, a novel dissimilarity measure is proposed based on the model of derived kernel. Combined with DFs, a DDM is constructed and utilized to enhance the small target and suppress the background in different scenes. Finally, a result map is obtained by fusing sub-maps from all directions. Experimental results on six datasets demonstrate that the proposed small target detection method can not only suppress complex background clutters successfully regardless of noise interference, but can also locate weak and dim targets precisely with a high detection ratio and low false-alarm ratio compared to some other methods.