Infrared cone-shaped spatial target recognition based on an improved MKELM.

This paper proposes a novel infrared cone-shaped spatial target recognition method, to the best of our knowledge, based on an improved multiple kernel extreme learning machine (MKELM) for the problem that radiation intensity sequence is the only data type available and is often contaminated by noise at long-range detection. Variational mode decomposition (VMD) and reconstruction are incorporated for radiation intensity sequence. Then, the whale optimization algorithm (WOA) is used to optimize parameters, and target recognition is tested on a simulated infrared radiation intensity sequence dataset using improved MKELM. The experimental results verify the effectiveness of the method and show its enhanced recognition accuracy and robustness.