A Fast Target Imaging Model Based on Electromagnetic Scattering Solution and Machine Learning

Traditional Inverse Synthetic Aperture Radar (ISAR) image simulation methods based on electromagnetic (EM) scattering calculation often face the challenge of high time costs, making it difficult to generate high-resolution ISAR image samples of the target in real-time. To address the low efficiency in constructing a dataset of image samples for complex targets, this paper proposes a machine learning-based rapid prediction model for ISAR image. This model utilizes a small amount of ISAR echo data by EM calculation of the complex target as input. Through feature engineering to expand the feature information of the echo data, it further employs ensemble techniques, combining five base models including Multilayer Perceptron (MLP), decision trees, k-Nearest Neighbor algorithm (KNN), Support Vector Machine (SVM) and gradient boosting. This ensemble approach enables quick predictions of the ISAR echo data, reducing the number of EM simulation calculations and significantly improving the efficiency of sample generation. Experimental results demonstrate that the proposed model can accurately predict the entire data required to generate images using only 10% of the echo data, achieving an overall efficiency improvement of approximately 90%. Especially, as the complexity and resolution of the target increase, the time required for image generation using simulation calculations will significantly rise. In such cases, our proposed model demonstrates even greater advantages.