Location: Food Quality LaboratoryTitle: Does spatial region of interest (ROI) matter in multispectral and hyperspectral imaging of segmented wheat kernels
Submitted to: Biosystems Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/6/2021
Publication Date: 10/23/2021
Citation: Delwiche, S.R., Baek, I., Kim, M.S. 2021. Does spatial region of interest (ROI) matter in multispectral and hyperspectral imaging of segmented wheat kernels. Biosystems Engineering. 212:106-114.
Interpretive Summary: Seeds of cereal grains, such as wheat, may undergo a post-harvest sorting operation for removal of foreign material or physiological damage. Existing commercial sorters typically use the capture of light reflected from the object surface for accepting or rejecting seed, with an emphasis placed on speed of operation. Advances in spectral imaging and computation are now making it feasible to perform rudimentary image (1-3 wavebands) analysis in seed sorting operations. This study was conducted to examine the effect of size of the region of a seed undergoing image analysis, with size ranging from the entire viewed surface down to a small centrally located subregion (approximately 5 percent) of the surface. The objective was to determine how small a size can be used while still maintaining the ability to categorize wheat seed into a accept or reject condition for sorting purposes. In a test case involving sound wheat kernels and kernels infected with Fusarium head blight (a fungal disease), it was found that image subregions representing less than 10 percent of the exposed seed surface produced reasonably accurate classifications, thus allowing for simplified image classification algorithm design. This research is intended to benefit developers of multispectral and hyperspectral image software for use in image-based cereal sorters. Ultimately, the work will lead to the adoption of image analysis in cereal seed sorting.
Technical Abstract: High-speed optical sorting of seeds for damage and foreign material is a well-developed technology, with commercial sorters readily available. Advances in optics technology and computational processing have brought multispectral and hyperspectral imaging to commercial sorting of fruits and vegetables, yet the application of imaging to single cereal seeds has lagged due to the enormity in numbers of seeds and challenges posed by lighting, shadowing, and seed curvature that are less problematic with larger objects. This study examined the effect of region of interest (ROI) size on the seed surface with respect to the ability to sort seed into accept and reject categories. With the predication that light reflection variation is greater between seeds of different class than within seed of the same class, sorting potential was examined on a set and subset of hard red wheat samples (n = 87 and 5, respectively) with varying mixtures of sound and fusarium-damaged seed. Regions of interest (ROI) size ranged from 5 centrally located pixels arranged in a cross to all pixels (typically 100) contained in the viewed surface of a kernel. Two modeling structures were used; the first involving all 87 samples (approximately 220 kernels per sample), in which mixture level of sound and fusarium-damaged kernels is known, but individual kernel class is unknown; and the second involving 5 samples, of which, an equal number of 287 known sound and known fusarium-damaged kernels were used. Accordingly, the larger set model characterized the dispersion (by standard deviation) of kernel-to-kernel reflectance at a single representative wavelength, while the smaller set was used to develop linear discriminant analysis classification models using one to three wavelengths. With either structure it was found that the smaller ROIs should be sufficient for a two-class (accepts, rejects) image-based sorting operation.