Spectral band identification tool for hyperspectral imaging-based diagnostics.

Stream Technologies是一家开发高光谱成像设备的加拿大公司。当一个人

Stream Technologies is a Canadian company that has developed hyperspectral imaging equipment. When one takes an image of a subject with a conventional camera, the intensity of each pixel in the image is representative of the intensity of all wavelengths (colours) of light at a given point in the image. With hyperspectral imaging, this image is split into a suite of spatially identical images; however the signal at each pixel of each image is based on the intensity of light across a specific range of wavelengths (colours). Since different materials absorb and reflect/emit light at different wavelengths, depending on their chemistry, the different images can provide spatially-resolved chemical information about different regions of the image. The challenge lies in determining how to tune the system so that appropriate spectral bands will be collected such that the desired information can be gleaned from the data set. This research project will develop an experimental protocol to collect the data necessary to tune the imaging system and model the resulting data such that the desired information can be obtained in the future. Initial work will be with model systems (e.g. pebbles mixed in coffee beans, identifying different candies, etc). Once this protocol is established, more challenging systems will be explored including distinguishing mustard and canola seeds, measuring meat moisture/protein/fat, etc.