Automatic Hypercube Acquisition with high spatial and spectral resolution using a HSI Linescan Camera

Abstract

Push-broom HSI-cameras have a high spectral resolution but require scanning and stitching to get a correct spatial and spectral representation of a scene. The conventional approach is to calculate a scanning velocity that matches the camera’s frame rate so that the stitching becomes trivial. For this to work one needs to be able to apply the velocity steadily and know the distance between camera and subject. This contribution presents a feature-based approach that directly determines the distance the subject was moved between the images and uses it to correctly stitch the images. It is used in a self-contained environment consisting of a lightprotected box with a light source inside and a linear guide unit that moves the object under examination. The distance between camera and object can be altered to increase the spatial resolution for small objects. It is shown that our stitching approach works even if the velocity changes during scanning. If the distance of the camera is altered, e. g. to scan a larger object, the stitching is adjusted automatically without the need for manual intervention.

Keywords

• hyperspectral imaging
• stitching
• push-broom
• image acquisition