Smart and ultra-compact hyperspectral camera

Abstract:
Hyper spectral imaging (HSI) serves many applications by gathering both the shape and chemical composition of objects. Unfortunately, the technology is associated with a high level of effort, especially for the spectral analysis which is based on near infrared (NIR) spectroscopy. However, there are applications where prior knowledge of the objects can be used to reduce this effort, e.g. for objects having a homogeneous material structure, like textiles. Using compact micro electro mechanical systems (MEMS) components and artificial intelligence-based analysis methods a new kind of cost-efficient and ultra-compact setup is enabled. High-resolution images can be captured and NIR spectral analysis is performed on automatically selected areas revealing their chemical composition. Preliminary experiments show reliable boundary recognition and material classification accuracy in controlled lab settings, highlighting the system’s potential for rapid, high-throughput industrial sorting.