A Method for Detection of Corn Kernel Mildew Based on Co-Clustering Algorithm with Hyperspectral Image Technology.

Hyperspectral imaging can simultaneously acquire spectral and spatial information of the samples and is, therefore, widely applied in the non-destructive detection of grain quality. Supervised learning is the mainstream method of hyperspectral imaging for pixel-level detection of mildew in corn kernels, which requires a large number of training samples to establish the prediction or classification models. This paper presents an unsupervised redundant co-clustering algorithm (FCM-SC) based on multi-center fuzzy c-means (FCM) clustering and spectral clustering (SC), which can effectively detect non-uniformly distributed mildew in corn kernels. This algorithm first carries out fuzzy c-means clustering of sample features, extracts redundant cluster centers, merges the cluster centers by spectral clustering, and finally finds the category of corresponding cluster centers for each sample. It effectively solves the problems of the poor ability of the traditional fuzzy c-means clustering algorithm to classify the data with complex structure distribution and the complex calculation of the traditional spectral clustering algorithm. The experimental results demonstrated that the proposed algorithm could describe the complex structure of mildew distribution in corn kernels and exhibits higher stability, better anti-interference ability, generalization ability, and accuracy than the supervised classification model.

Rice Seed Purity Identification Technology Using Hyperspectral Image with LASSO Logistic Regression Model.

Hyperspectral technology is used to obtain spectral and spatial information of samples simultaneously and demonstrates significant potential for use in seed purity identification. However, it has certain limitations, such as high acquisition cost and massive redundant information. This study integrates the advantages of the sparse feature of the least absolute shrinkage and selection operator (LASSO) algorithm and the classification feature of the logistic regression model (LRM). We propose a hyperspectral rice seed purity identification method based on the LASSO logistic regression model (LLRM). The feasibility of using LLRM for the selection of feature wavelength bands and seed purity identification are discussed using four types of rice seeds as research objects. The results of 13 different adulteration cases revealed that the value of the regularisation parameter was different in each case. The recognition accuracy of LLRM and average recognition accuracy were 91.67-100% and 98.47%, respectively. Furthermore, the recognition accuracy of full-band LRM was 71.60-100%. However, the average recognition accuracy was merely 89.63%. These results indicate that LLRM can select the feature wavelength bands stably and improve the recognition accuracy of rice seeds, demonstrating the feasibility of developing a hyperspectral technology with LLRM for seed purity identification.

Comparative Performance of Spectral Reflectance Indices and Multivariate Modeling for Assessing Agronomic Parameters in Advanced Spring Wheat Lines Under Two Contrasting Irrigation Regimes.

The incorporation of nondestructive and cost-effective tools in genetic drought studies in combination with reliable indirect screening criteria that exhibit high heritability and genetic correlations will be critical for addressing the water deficit challenges of the agricultural sector under arid conditions and ensuring the success of genotype development. In this study, the proximal spectral reflectance data were exploited to assess three destructive agronomic parameters [dry weight (DW) and water content (WC) of the aboveground biomass and grain yield (GY)] in 30 recombinant F7 and F8 inbred lines (RILs) growing under full (FL) and limited (LM) irrigation regimes. The utility of different groups of spectral reflectance indices (SRIs) as an indirect assessment tool was tested based on heritability and genetic correlations. The performance of the SRIs and different models of partial least squares regression (PLSR) and stepwise multiple linear regression (SMLR) in estimating the destructive parameters was considered. Generally, all groups of SRIs, as well as different models of PLSR and SMLR, generated better estimations for destructive parameters under LM and combined FL+LM than under FL. Even though most of the SRIs exhibited a low association with destructive parameters under FL, they exhibited moderate to high genetic correlations and also had high heritability. The SRIs based on near-infrared (NIR)/visible (VIS) and NIR/NIR, especially those developed in this study, spectral band intervals extracted within VIS, red edge, and NIR spectral range, or individual effective wavelengths relevant to green, red, red edge, and middle NIR spectral region, were found to be more effective in estimating the destructive parameters under all conditions. Five models of SMLR and PLSR for each condition explained most of the variation in the three destructive parameters among genotypes. These models explained 42% to 46%, 19% to 30%, and 39% to 46% of the variation in DW, WC, and GY among genotypes under FL, 69% to 72%, 59% to 61%, and 77% to 81% under LM, and 71% to 75%, 61% to 71%, and 74% to 78% under FL+LM, respectively. Overall, these results confirmed that application of hyperspectral reflectance sensing in breeding programs is not only important for evaluating a sufficient number of genotypes in an expeditious and cost-effective manner but also could be exploited to develop indirect breeding traits that aid in accelerating the development of genotypes for application under adverse environmental conditions.