Multispectral and hyperspectral machine vision systems deliver powerful inspection capabilities, but their complexity and cost can be a barrier to their use in cost-sensitive applications. Using two case studies, this talk presents hardware-level strategies for reducing system cost. These real-world applications utilize intelligent camera selection and application-matched LED illumination design.

The first approach consolidates a dual RGB-and SWIR-based machine vision solution, into a single system utilizing a multi-sensor, prism-based CMOS and InGaAs sensor combined with a configurable RGB-SWIR multispectral LED line light. The reduction in camera and emitter count, and associated electrical ancillaries and integration overhead, without sacrificing on performance, significantly improved the Performance-Cost ratio of this machine vision system.

The second approach uses a dual-sensor SWIR camera where each sensor targets a discrete spectral band. By carefully pairing, and strobing, specific wavelengths in the LED source in order to elicit a response to a known, possible material in only one sensor, the system is able to detect foreign material presence or absence without broadband illumination or spectral filtering hardware.

Attendees will leave with practical cost-reduction frameworks and an understanding of the lighting parameters required to implement them.