RESUMEN
Due to the change of illumination environment and overlapping conditions caused by the neighboring fruits and other background objects, the simple application of the traditional machine vision method limits the detection accuracy of lychee fruits in natural orchard environments. Therefore, this research presented a detection method based on monocular machine vision to detect lychee fruits growing in overlapped conditions. Specifically, a combination of contrast limited adaptive histogram equalization (CLAHE), red/blue chromatic mapping, Otsu thresholding and morphology operations were adopted to segment the foreground regions of the lychees. A stepwise method was proposed for extracting individual lychee fruit from the lychee foreground region. The first step in this process was based on the relative position relation of the Hough circle and an equivalent area circle (equal to the area of the potential lychee foreground region) and was designed to distinguish lychee fruits growing in isolated or overlapped states. Then, a process based on the three-point definite circle theorem was performed to extract individual lychee fruits from the foreground regions of overlapped lychee fruit clusters. Finally, to enhance the robustness of the detection method, a local binary pattern support vector machine (LBP-SVM) was adopted to filter out the false positive detections generated by background chaff interferences. The performance of the presented method was evaluated using 485 images captured in a natural lychee orchard in Conghua (Area), Guangzhou. The detection results showed that the recall rate was 86.66%, the precision rate was greater than 87% and the F1-score was 87.07%.
RESUMEN
The maturity stage of bananas has a considerable influence on the fruit postharvest quality and the shelf life. In this study, an optical imaging based method was formulated to assess the importance of different external properties on the identification of four successive banana maturity stages. External optical properties, including the peel color and the local textural and local shape information, were extracted from the stalk, middle and tip of the bananas. Specifically, the peel color attributes were calculated from individual channels in the hue-saturation-value (HSV), the International Commission on Illumination (CIE) L*a*b* and the CIE L*ch color spaces; the textural information was encoded using a local binary pattern with uniform patterns (UP-LBP); and the local shape features were described by histogram of oriented gradients (HOG). Three classifiers based on the naïve Bayes (NB), linear discriminant analysis (LDA) and support vector machine (SVM) algorithms were adopted to evaluate the performance of identifying banana fruit maturity stages using the different optical appearance features. The experimental results demonstrate that overall identification accuracies of 99.2%, 100% and 99.2% were achieved using color appearance features with the NB, LDA and SVM classifiers, respectively; overall accuracies of 92.6%, 86.8% and 93.4% were obtained using local textural features for the three classifiers, respectively; and overall accuracies of only 84.3%, 83.5% and 82.6% were obtained using local shape features with the three classifiers, respectively. Compared to the complicated calculation of both the local textural and local shape properties, the simplicity and high accuracy of the peel color property make it more appropriate for identifying banana fruit maturity stages using optical imaging techniques.
Asunto(s)
Frutas/crecimiento & desarrollo , Musa/crecimiento & desarrollo , Imagen Óptica , Algoritmos , Teorema de Bayes , Color , Análisis Discriminante , Máquina de Vectores de SoporteRESUMEN
The segmentation of citrus trees in a natural orchard environment is a key technology for achieving the fully autonomous operation of agricultural unmanned aerial vehicles (UAVs). Therefore, a tree segmentation method based on monocular machine vision technology and a support vector machine (SVM) algorithm are proposed in this paper to segment citrus trees precisely under different brightness and weed coverage conditions. To reduce the sensitivity to environmental brightness, a selective illumination histogram equalization method was developed to compensate for the illumination, thereby improving the brightness contrast for the foreground without changing its hue and saturation. To accurately differentiate fruit trees from different weed coverage backgrounds, a chromatic aberration segmentation algorithm and the Otsu threshold method were combined to extract potential fruit tree regions. Then, 14 color features, five statistical texture features, and local binary pattern features of those regions were calculated to establish an SVM segmentation model. The proposed method was verified on a dataset with different brightness and weed coverage conditions, and the results show that the citrus tree segmentation accuracy reached 85.27% ± 9.43%; thus, the proposed method achieved better performance than two similar methods.
RESUMEN
Hyperspectral imaging was explored to detect Sclerotinia stem rot (SSR) on oilseed rape leaves with chemometric methods, and the influences of variable selection, machine learning, and calibration transfer methods on detection performances were evaluated. Three different sample sets containing healthy and infected oilseed rape leaves were acquired under different imaging acquisition parameters. Four discriminant models were built using full spectra, including partial least squares-discriminant analysis (PLS-DA), support vector machine (SVM), soft independent modeling of class analogies (SIMCA), and k-nearest neighbors (KNN). PLS-DA and SVM models were also built with the optimal wavelengths selected by principal component analysis (PCA) loadings, second derivative spectra, competitive adaptive reweighted sampling (CARS), and successive projections algorithm (SPA). The optimal wavelengths selected for each sample set by different methods were different; however, the optimal wavelengths selected by PCA loadings and second derivative spectra showed similarity between different sample sets. Direct standardization (DS) was successfully applied to reduce spectral differences among different sample sets. Overall, the results demonstrated that using hyperspectral imaging with chemometrics for plant disease detection can be efficient and will also help in the selection of optimal variable selection, machine learning, and calibration transfer methods for fast and accurate plant disease detection.
Asunto(s)
Ascomicetos/aislamiento & purificación , Brassica napus/microbiología , Enfermedades de las Plantas/microbiología , Hojas de la Planta/microbiología , Algoritmos , Ascomicetos/patogenicidad , Brassica napus/crecimiento & desarrollo , Tallos de la Planta/microbiología , Análisis de Componente Principal , Espectroscopía Infrarroja Corta , Máquina de Vectores de SoporteRESUMEN
Variety classification is an important step in seed quality testing. This study introduces t-distributed stochastic neighbourhood embedding (t-SNE), a manifold learning algorithm, into the field of hyperspectral imaging (HSI) and proposes a method for classifying seed varieties. Images of 800 maize kernels of eight varieties (100 kernels per variety, 50 kernels for each side of the seed) were imaged in the visible- near infrared (386.7â»1016.7 nm) wavelength range. The images were pre-processed by Procrustes analysis (PA) to improve the classification accuracy, and then these data were reduced to low-dimensional space using t-SNE. Finally, Fisher's discriminant analysis (FDA) was used for classification of the low-dimensional data. To compare the effect of t-SNE, principal component analysis (PCA), kernel principal component analysis (KPCA) and locally linear embedding (LLE) were used as comparative methods in this study, and the results demonstrated that the t-SNE model with PA pre-processing has obtained better classification results. The highest classification accuracy of the t-SNE model was up to 97.5%, which was much more satisfactory than the results of the other models (up to 75% for PCA, 85% for KPCA, 76.25% for LLE). The overall results indicated that the t-SNE model with PA pre-processing can be used for variety classification of waxy maize seeds and be considered as a new method for hyperspectral image analysis.
RESUMEN
Rapid detection of soil nutrient elements is beneficial to the evaluation of crop yield, and it's of great significance in agricultural production. The aim of this work was to compare the detection ability of single-pulse (SP) and collinear double-pulse (DP) laser-induced breakdown spectroscopy (LIBS) for soil nutrient elements and obtain an accurate and reliable method for rapid detection of soil nutrient elements. 63 soil samples were collected for SP and collinear DP signal acquisition, respectively. Macro-nutrients (K, Ca, Mg) and micro-nutrients (Fe, Mn, Na) were analyzed. Three main aspects of all elements were investigated, including spectral intensity, signal stability, and detection sensitivity. Signal-to-noise ratio (SNR) and relative standard deviation (RSD) of elemental spectra were applied to evaluate the stability of SP and collinear DP signals. In terms of detection sensitivity, the performance of chemometrics models (univariate and multivariate analysis models) and the limit of detection (LOD) of elements were analyzed, and the results indicated that the DP-LIBS technique coupled with PLSR could be an accurate and reliable method in the quantitative determination of soil nutrient elements.
RESUMEN
This work provides the experimental and theoretical fundamentals for detecting the molecular fingerprints of six kinds of pesticides by using terahertz (THz) time-domain spectroscopy (THz-TDS). The spectra of absorption coefficient and refractive index of the pesticides, chlorpyrifos, fipronil, carbofuran, dimethoate, methomyl, and thidiazuron are obtained in frequencies of 0.1â»3.5 THz. To accurately describe the THz spectral characteristics of pesticides, the wavelet threshold de-noising (WTD) method with db 5 wavelet fucntion, 5-layer decomposition, and soft-threshold de-noising was used to eliminate the spectral noise. The spectral baseline correction (SBC) method based on asymmetric least squares smoothing was used to remove the baseline drift. Spectral results show that chlorpyrifo had three characteristic absorption peaks at 1.47, 1.93, and 2.73 THz. Fipronil showed three peaks at 0.76, 1.23, and 2.31 THz. Carbofuran showed two peaks at 2.72 and 3.06 THz. Dimethoate showed three peaks at 1.05, 1.89, and 2.92 THz. Methomyl showed five peaks at 1.01, 1.65, 1.91, 2.72, and 3.20 THz. Thidiazuron showed four peaks at 0.99, 1.57, 2.17, and 2.66 THz. The density functional theory (DFT) of B3LYP/6-31G+(d,p) was applied to simulate the molecular dynamics for peak analyzing of the pesticides based on isolated molecules. The theoretical spectra are in good agreement with the experimental spectra processed by WTD + SBC, which implies the validity of WTD + SBC spectral processing methods and the accuracy of DFT spectral peak analysis. These results support that the combination of THz-TDS and DFT is an effective tool for pesticide fingerprint analysis and the molecular dynamics simulations.
Asunto(s)
Plaguicidas/análisis , Plaguicidas/química , Análisis de los Mínimos Cuadrados , Simulación de Dinámica Molecular , Refractometría , Análisis Espectral/métodos , Espectroscopía de TerahertzRESUMEN
A high-performance medical image segmentation model based on deep learning depends on the availability of large amounts of annotated training data. However, it is not trivial to obtain sufficient annotated medical images. Generally, the small size of most tissue lesions, e.g., pulmonary nodules and liver tumours, could worsen the class imbalance problem in medical image segmentation. In this study, we propose a multidimensional data augmentation method combining affine transform and random oversampling. The training data is first expanded by affine transformation combined with random oversampling to improve the prior data distribution of small objects and the diversity of samples. Secondly, class weight balancing is used to avoid having biased networks since the number of background pixels is much higher than the lesion pixels. The class imbalance problem is solved by utilizing weighted cross-entropy loss function during the training of the CNN model. The LUNA16 and LiTS17 datasets were introduced to evaluate the performance of our works, where four deep neural network models, Mask-RCNN, U-Net, SegNet and DeepLabv3+, were adopted for small tissue lesion segmentation in CT images. In addition, the small tissue segmentation performance of the four different deep learning architectures on both datasets could be greatly improved by incorporating the data augmentation strategy. The best pixelwise segmentation performance for both pulmonary nodules and liver tumours was obtained by the Mask-RCNN model, with DSC values of 0.829 and 0.879, respectively, which were similar to those of state-of-the-art methods.
Asunto(s)
Neoplasias Hepáticas , Nódulos Pulmonares Múltiples , Humanos , Procesamiento de Imagen Asistido por Computador/métodos , Redes Neurales de la ComputaciónRESUMEN
Intelligent detection and localization of mature citrus fruits is a critical challenge in developing an automatic harvesting robot. Variable illumination conditions and different occlusion states are some of the essential issues that must be addressed for the accurate detection and localization of citrus in the orchard environment. In this paper, a novel method for the detection and localization of mature citrus using improved You Only Look Once (YOLO) v5s with binocular vision is proposed. First, a new loss function (polarity binary cross-entropy with logit loss) for YOLO v5s is designed to calculate the loss value of class probability and objectness score, so that a large penalty for false and missing detection is applied during the training process. Second, to recover the missing depth information caused by randomly overlapping background participants, Cr-Cb chromatic mapping, the Otsu thresholding algorithm, and morphological processing are successively used to extract the complete shape of the citrus, and the kriging method is applied to obtain the best linear unbiased estimator for the missing depth value. Finally, the citrus spatial position and posture information are obtained according to the camera imaging model and the geometric features of the citrus. The experimental results show that the recall rates of citrus detection under non-uniform illumination conditions, weak illumination, and well illumination are 99.55%, 98.47%, and 98.48%, respectively, approximately 2-9% higher than those of the original YOLO v5s network. The average error of the distance between the citrus fruit and the camera is 3.98 mm, and the average errors of the citrus diameters in the 3D direction are less than 2.75 mm. The average detection time per frame is 78.96 ms. The results indicate that our method can detect and localize citrus fruits in the complex environment of orchards with high accuracy and speed. Our dataset and codes are available at https://github.com/AshesBen/citrus-detection-localization.
RESUMEN
The novel coronavirus disease 2019 (COVID-19) pandemic has led to a worldwide crisis in public health. It is crucial we understand the epidemiological trends and impact of non-pharmacological interventions (NPIs), such as lockdowns for effective management of the disease and control of its spread. We develop and validate a novel intelligent computational model to predict epidemiological trends of COVID-19, with the model parameters enabling an evaluation of the impact of NPIs. By representing the number of daily confirmed cases (NDCC) as a time-series, we assume that, with or without NPIs, the pattern of the pandemic satisfies a series of Gaussian distributions according to the central limit theorem. The underlying pandemic trend is first extracted using a singular spectral analysis (SSA) technique, which decomposes the NDCC time series into the sum of a small number of independent and interpretable components such as a slow varying trend, oscillatory components and structureless noise. We then use a mixture of Gaussian fitting (GF) to derive a novel predictive model for the SSA extracted NDCC incidence trend, with the overall model termed SSA-GF. Our proposed model is shown to accurately predict the NDCC trend, peak daily cases, the length of the pandemic period, the total confirmed cases and the associated dates of the turning points on the cumulated NDCC curve. Further, the three key model parameters, specifically, the amplitude (alpha), mean (mu), and standard deviation (sigma) are linked to the underlying pandemic patterns, and enable a directly interpretable evaluation of the impact of NPIs, such as strict lockdowns and travel restrictions. The predictive model is validated using available data from China and South Korea, and new predictions are made, partially requiring future validation, for the cases of Italy, Spain, the UK and the USA. Comparative results demonstrate that the introduction of consistent control measures across countries can lead to development of similar parametric models, reflected in particular by relative variations in their underlying sigma, alpha and mu values. The paper concludes with a number of open questions and outlines future research directions.
Asunto(s)
Inteligencia Artificial , COVID-19/terapia , COVID-19/epidemiología , COVID-19/virología , Humanos , SARS-CoV-2/aislamiento & purificación , España/epidemiologíaRESUMEN
The Asian citrus psyllid, Diaphorina citri Kuwayama (Hemiptera: Liviidae), can cause direct damage to citrus trees and is the main vector for the devastating disease, citrus greening disease or huanglongbing. Most molecular studies on this important insect pest use real-time reverse-transcription quantitative polymerase chain reaction (RT-qPCR) to quantify gene expression, including analyzing molecular basis for insecticide resistance in field populations. One critical factor to cause inaccuracy in RT-qPCR results is the lack of appropriate internal reference genes for optimal data normalization. In this study, the expression levels of 10 selected reference genes were evaluated in different tissue samples of psyllid adults and in the insects treated with different temperatures and insecticides. Data were analyzed using different computational algorithms, including Delta Ct, BestKeeper, NormFinder, geNorm, and RefFinder. According to our results, at least two reference genes should be used for the normalization of RT-qPCR data in this insect. The best choices of reference genes for different samples are as follows: ACT1 and Ferritin for different tissue samples, RPS20 and Ferritin for samples treated with different temperatures, TBP and EF1α for samples treated with imidacloprid, and Ferritin and TBP for samples treated with beta-cypermethrin. The reference genes identified in this study should be useful for future studies to analyze the expression patterns of target genes, especially for genes linked with temperature adaptability and insecticide resistance in this insect species in the future.
Asunto(s)
Genes de Insecto , Hemípteros/genética , Algoritmos , Animales , Hemípteros/metabolismo , Reacción en Cadena en Tiempo Real de la Polimerasa/normas , Estándares de ReferenciaRESUMEN
There are a few common species and many rare species in a biological community or a multi-species collection in given space and time. This hollow distribution curve is called species abundance distribution (SAD). Few studies have examined the patterns and dynamics of SADs during the succession of forest communities by model selection. This study explored whether the communities in different successional stages followed different SAD models and whether there existed a best SAD model to reveal their intrinsic quantitative features of structure and dynamics in succession. The abundance (the number of individuals) of each vascular plant was surveyed by quadrat sampling method from the tree, shrub and herb layers in two typical communities (i.e., the evergreen needle- and broad-leaved mixed forest and the monsoon evergreen broad-leaved forest) in southern subtropical Dinghushan Biosphere Reserve, South China. The sites of two forest communities in different successional stages are both 1 ha in area. We collected seven widely representative SAD models with obviously different function forms and transformed them into the same octave (log2) scale. These models are simultaneously confronted with eight datasets from four layers of two communities, and their goodness-of-fits to the data were evaluated by the chi-squared test, the adjusted coefficient of determination and the information criteria. The results indicated that: (1) the logCauchy model followed all the datasets and was the best among seven models; (2) the fitness of each model to the data was not directly related to the successional stage of forest community; (3) according to the SAD curves predicted by the best model (i.e., the logCauchy), the proportion of rare species decreased but that of common ones increased in the upper layers with succession, while the reverse was true in the lower layers; and (4) the difference of the SADs increased between the upper and the lower layers with succession. We concluded that the logCauchy model had the widest applicability in describing the SADs, and could best mirror the SAD patterns and dynamics of communities and their different layers in the succession of forests. The logCauchy-modeled SADs can quantitatively guide the construction of ecological forests and the restoration of degraded vegetation.
Asunto(s)
Bosques , Modelos BiológicosRESUMEN
Early detection of foliar diseases is vital to the management of plant disease, since these pathogens hinder crop productivity worldwide. This research applied hyperspectral imaging (HSI) technology to early detection of Magnaporthe oryzae-infected barley leaves at four consecutive infection periods. The averaged spectra were used to identify the infection periods of the samples. Additionally, principal component analysis (PCA), spectral unmixing analysis and spectral angle mapping (SAM) were adopted to locate the lesion sites. The results indicated that linear discriminant analysis (LDA) coupled with competitive adaptive reweighted sampling (CARS) achieved over 98% classification accuracy and successfully identified the infected samples 24 h after inoculation. Importantly, spectral unmixing analysis was able to reveal the lesion regions within 24 h after inoculation, and the resulting visualization of host-pathogen interactions was interpretable. Therefore, HSI combined with analysis by those methods would be a promising tool for both early infection period identification and lesion visualization, which would greatly improve plant disease management.