Prediction of the Contrast between Target and Background based on an Improved Support Vector Machine

Abstract

In this paper, to avoid modeling the characteristic of infrared radiation and contrast between the target and the background, the apparent temperature difference (ATD) between the target and the background is used as an alternative method to evaluate the infrared radiation contrast. For static fixed targets, the ATD mostly depends on the external meteorological factors, which make it reasonable to use the meteorological information to predict the ATD. Thus, a support vector machine (SVM) algorithm based on an improved PSO algorithm is proposed to predict the ATD of two different static targets based on long-term testing. The improved PSO algorithm, called dynamic selection strategy based PSO, is proposed to search the optimal parameters of SVM for improving the performance of SVM. The experimental results show the feasibility and effectiveness of the proposed method.