Using analytic hierarchy process to evaluate deep learning for infrared target recognition
Laixiang Xu,Bingxu Cao,Xinhong Wei,Bei Li,Jingfeng Su,Shiyuan Ning,Junmin Zhao
DOI: https://doi.org/10.1007/s11042-024-20373-x
IF: 2.577
2024-10-12
Multimedia Tools and Applications
Abstract:In order to promote the evaluation performance of deep learning infrared automatic target recognition (ATR) algorithms in the complex environment of air-to-air missile research, we proposed an analytic hierarchy process (AHP) evaluation system consisting of the target recognition ability, the generalization recognition ability, the anti-interference ability, the background suppression ability, and the cost metrics ability. According to detailed quantitative analysis and experimental results, we establish probability evaluation models and get their probability values. Compared with traditional evaluation systems, our proposed method implementation is showing high reliability. Similar superior performances are also achieved, such as quick, comprehensive, and scientific. Through the above research, the key technology of deep learning performance evaluation in automatic target recognition has been broken through, which provides some basis and guidance for the analysis of infrared imaging-guided weapon systems.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering