Infrared small target detection using directional highpass filters based on LS-SVM

An infrared small target detection method based on the least squares support vector machine (LS-SVM) is proposed. The LS-SVM model is applied to generate a row and a column directional highpass template filters. By filtering the original infrared image with the two templates, respectively, and fusing the two filtered images, the possible small target can be extracted. Experimental results demonstrate that the presented method is robust and efficient.