The Evaluation of Infrared Image Features Based on Modified RELIEF Algorithms

The infrared image features are the basis of the automatic target recognition (ATR) algorithm for seeker in the countermeasure environment. The complication of the IR countermeasure environment will inevitably lead to the constantly appearance of the new features and the features' combinations while lack of the importance evaluation of the new features. Aiming at this problem, this paper proposes a method of performance evaluation of infrared image features based on RELIEF algorithm. Firstly, the feature set is summed up by the analysis of the common ATR algorithms. Secondly, the feature space is formed by removing the redundant features using the method of information theory. In addition, the importance of the features is evaluated by the modified RELIEF algorithm. The experimental results show that this paper's algorithm is an effective performance evaluation metric for image features. It has certain theoretical and applicable values.