Vis-and NIR-Based Instruments for Detection of Black-Tip Damaged Wheat Kernels: A Comparative Study

摘要

Black-tip (BT) is a non-mycotoxic fungus that attacks wheat kernels, forming a dark brown or black sooty area at the tip of the kernel. Visual inspection, which is the approved reference method for determining the amount of BT in wheat, requires substantial time and has high potential for subjective evaluation. Three spectrometers covering the spectral ranges 950-1636 nm (Sped), 600-1045 nm (Sped), and 380-780 nm (Spec3) were evaluatedfor their ability to predict the presence ofBT. Kernels were quantified into four levels: (A) sound, (B) low black-tip symptoms (BTS), (C) high BTS, and (D) BT damaged (BTD). Discriminant classification models were developed to evaluate combinations of levels. The combinations were (1) levels A, B, C, and D separately;(2) A, B+C andD; and (3) A +B and C+D. Spectral data for 2,760 kernels obtainedfrom 23 hard red winter (HRW) wheat samples, each comprising 30 kernels that were visually selectedfor each of the four levels of black-tip severity (A, B, C, and D), were collected with each spectrometer. Discriminant calibration models for each spectrometer and classification category were developed based on (I) three combinations of 17 HRW wheat samples, with the six remaining samples usedfor independent validation, and (2) combinations of 20 randomly selected kernels from each of the 23 HRW wheat samples as calibration samples, with the remaining ten kernels used as validation samples. Discriminant analysis was based on five wavelengths for each model. Spectra pretreatment was the standard normal variate (SNV). Results showed that all three spectrometers were capable of detecting BT damage on wheat kernels. BT classification accuracy was observed to have been affected by wheat varieties for Sped and Spec2 (both with NIR wavelengths) but not for Spec3, which was entirely in the visible region. The two-category classification (A+B, C+D) provided higher accuracy than the three-category (A, B+C, D) andfour-category (A, B, C, D) classifications. Based on the percent correct classification and Youden s index, Spec2 performed better in detecting sound and BTD wheat kernels, with classification accuracies of the best two-category classification calibration model ranging from 85.6% to 87.5%, compared to Sped at 74.8% to 78.4%andSpec3 at 76.7% to 79.2%. This study also showed the potential of using a five-wavelength model, which equates to the potential for developing simple, less expensive, high-speed photoelectric detection instruments. These instruments can serve as important tools in plant breeding, grading, or grain processing facilities to enable BT detection and, with proper selection of wavelengths, may also find applications in simultaneous single-kernel detection, measurement, and segregation of other chemical characteristics, such as protein and starch content.