Multi-modal High-Resolution Hyperspectral Object Detection System Based on Lightweight Platform

To solve the problems of large volume and low detection efficiency of UAV hyperspectral imagers, a light and small multi-modal high-resolution hyperspectral imager is proposed. This study primarily introduces the optical system design, data acquisition, and real-time processing modules of the hyperspectral imager. The proposed imager satisfies the detection mode requirements in different fields, such as spectral characteristic analysis and target detection, by switching the scanning mode. A low distortion, high throughput, and compact spectroscopic optical system design is adopted to meet the weight and detection accuracy requirements of the UAV platform for the spectral imager. The product was processed according to the design requirements and the performance was tested simultaneously. Among these, the MTF reached 0.19 and spectral resolution was 3.5–5.4 nm. The ability of the system to detect real-time targets was verified through detection of impurities in various pipelines. The implementation results showed that the system achieved high-precision spectral anomaly target detection in 2048 pixel×2048 pixel scenes per second with a detection accuracy greater than 87%.