|Delwiche, Stephen - Steve|
|YANG, I-CHANG - National Science Council|
Submitted to: Computers and Electronics in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/4/2013
Publication Date: 10/1/2013
Citation: Delwiche, S.R., Yang, I., Graybosch, R.A. 2013. Multiple view image analysis of freefalling U.S. wheat grains for damage assessment. Computers and Electronics in Agriculture. 98:62-73.
Interpretive Summary: Inspection of wheat in the United States for grade and class is currently performed through human visual analysis. This is a time consuming operation typically taking several minutes for each grain lot. Traditional digital imaging research applied to the cereals has been successful in detecting wheat from non-wheat and wheat of differing wheat classes. An unmet challenge, however, exists in detecting kernel defects from damage and disease. This study addresses the challenge through an innovative design that utilizes one high speed camera to view freefalling kernels from multiple viewing angles. Size, shape, and texture properties at each view are calculated and form the basis for deciding whether the kernel is damaged or sound, with an experimentally determined 95% level of accuracy. Beneficiaries of this research include federal, state and private grain inspection agencies, manufacturers of grain quality instruments and, in the long run, the public through improvement in wholesomeness of the country’s wheat supply.
Technical Abstract: Currently, inspection of wheat in the United States for grade and class is performed by human visual analysis. This is a time consuming operation typically taking several minutes for each sample. Digital imaging research has addressed this issue over the past two decades, with success in recognition of differing wheat classes and distinguishing wheat from non-wheat species. Detection of wheat kernel defects, either by damage or disease, has been a greater challenge. A study has been undertaken that uses high-speed digital imaging to detect damaged kernels in freefall, one kernel at a time. The system is composed of hardware (camera, lighting, power supplies, and data acquisition card) and associated software for instrument control, data collection, and analysis. It is designed to capture images of freefalling kernels at opposing angles through the use of optical grade mirrors. Parameterization is performed on kernel morphological and textural characteristics, whereupon these terms are used to develop classification models for sound and damaged classes. Fifty samples of hard red and white wheat subjected to weather-related damage during plant development were examined. Parametric (linear discriminant analysis, LDA) and non-parametric (k-nearest neighbor, KNN) classification models were tested to determine the image features that best foster recognition of kernel damage (mold, sprout, and black tip). The morphological features used in classification included area, projected volume, perimeter, ellipse eccentricity, and major and minor axis lengths. Textural features from calculated gray level co-occurrence matrices (including contrast, correlation, energy, homogeneity) as well as entropy were also considered, as were elliptical Fourier descriptors (truncated Fourier series functions that defined the contour of border in each view). The results indicate that with a combination of two morphological and four texture properties, classification levels attain 91 to 94 percent accuracy, depending on the type of classification model (LDA or KNN). The research findings are intended to lead to the streamlining of feature extraction in image-based grain inspection as well as to design criteria for high speed sorting.