Detection of infrared targets based on Adaboost by feature extraction using KPCA

According to the shortages of conventional infrared target detection method, an algorithm combining kernel principal component analysis (KPCA) and Adaboost classifier was presented. Firstly, KPCA was used to extract the features of the training samples of the target. Then the training samples of the background and the samples to be detected were projected onto the feature vectors of the target in kernel space as the features of themselves respectively. The features of both target and background samples were then used to train the classifier. Finally, the above classifier was applied to detect the target in IR images. The performances of the algorithm presented were compared with support vector machine (SVM) and quadratic correlation filter (QCF). Experimental results show that the proposed algorithm can achieve a robust and accurate detection and the parameter setting of the detection algorithm has a certain degree of adaptability.