Leaf Pixels Research Articles

Automatic visual recognition, detection and classification of weeds in cotton fields based on machine vision

Crops and weeds are involved in a continuous competition for equal resources, which may result in a potential decrease in crop yields by up to 31% and an increase in the costs of agricultural inputs by up to 22% of cultivation. Weeds further impact crop production, and their detection is crucial for effective crop management. In this research, we targeted common weeds of cotton field, specifically i) Digitaria sanguinalis (L.) Scop, ii) Amaranthus retroflexus L., iii) Acalypha australis, L., iv) Cephalanoplos segetum, and v) Chenopodium album L. Additionally, image processing techniques such as grayscale conversion, binarization, and Gaussian and morphological filters were also utilized. These methods are based on machine vision and facilitate rapid and straightforward weed detection by segmenting, scrutinizing, and comparing input images. The plant height and area were obtained during cotton planting within 32 days and fitted to develop the growth law concerning planting days for achieving the function of distinguishing cotton from weeds. We conducted recognition experiments by dividing images into four quadrants and categorizing weeds as either inter-row or intra-row. Meanwhile, the inter-row planting information was used to identify weeds, and the leaf pixel area and circularity were used as the identification methods for intra-row weeds, which reduced the algorithm's running time and improved real-time performance. The experimental results indicated that the inter-row weed recognition rate was 89.4%, with an average processing time of 102ms. Whereas in the case of intra-row weeds, the recognition rate was measured at 84.6%, and the overall recognition rate for cotton was 85.0%, with a mean time consumption of 437ms. Furthermore, the present research underscores recent advancements such as machine vision and high-resolution imaging, which have significantly improved the accuracy of automated weed identification in cotton fields while acknowledging ongoing challenges and outlining future opportunities. By Integrating state-of-the-art technology with sustainable agricultural practices, implementing an intelligent system offers a viable approach toward efficient and environmentally friendly weed management in modern agriculture.

PDSE-Lite: lightweight framework for plant disease severity estimation based on Convolutional Autoencoder and Few-Shot Learning.

Plant disease diagnosis with estimation of disease severity at early stages still remains a significant research challenge in agriculture. It is helpful in diagnosing plant diseases at the earliest so that timely action can be taken for curing the disease. Existing studies often rely on labor-intensive manually annotated large datasets for disease severity estimation. In order to conquer this problem, a lightweight framework named "PDSE-Lite" based on Convolutional Autoencoder (CAE) and Few-Shot Learning (FSL) is proposed in this manuscript for plant disease severity estimation with few training instances. The PDSE-Lite framework is designed and developed in two stages. In first stage, a lightweight CAE model is built and trained to reconstruct leaf images from original leaf images with minimal reconstruction loss. In subsequent stage, pretrained layers of the CAE model built in the first stage are utilized to develop the image classification and segmentation models, which are then trained using FSL. By leveraging FSL, the proposed framework requires only a few annotated instances for training, which significantly reduces the human efforts required for data annotation. Disease severity is then calculated by determining the percentage of diseased leaf pixels obtained through segmentation out of the total leaf pixels. The PDSE-Lite framework's performance is evaluated on Apple-Tree-Leaf-Disease-Segmentation (ATLDS) dataset. However, the proposed framework can identify any plant disease and quantify the severity of identified diseases. Experimental results reveal that the PDSE-Lite framework can accurately detect healthy and four types of apple tree diseases as well as precisely segment the diseased area from leaf images by using only two training samples from each class of the ATLDS dataset. Furthermore, the PDSE-Lite framework's performance is compared with existing state-of-the-art techniques, and it is found that this framework outperformed these approaches. The proposed framework's applicability is further verified by statistical hypothesis testing using Student t-test. The results obtained from this test confirm that the proposed framework can precisely estimate the plant disease severity with a confidence interval of 99%. Hence, by reducing the reliance on large-scale manual data annotation, the proposed framework offers a promising solution for early-stage plant disease diagnosis and severity estimation.

StripeRust-Pocket: A Mobile-Based Deep Learning Application for Efficient Disease Severity Assessment of Wheat Stripe Rust.

Wheat stripe rust poses a marked threat to global wheat production. Accurate and effective disease severity assessments are crucial for disease resistance breeding and timely management of field diseases. In this study, we propose a practical solution using mobile-based deep learning and model-assisted labeling. StripeRust-Pocket, a user-friendly mobile application developed based on deep learning models, accurately quantifies disease severity in wheat stripe rust leaf images, even under complex backgrounds. Additionally, StripeRust-Pocket facilitates image acquisition, result storage, organization, and sharing. The underlying model employed by StripeRust-Pocket, called StripeRustNet, is a balanced lightweight 2-stage model. The first stage utilizes MobileNetV2-DeepLabV3+ for leaf segmentation, followed by ResNet50-DeepLabV3+ in the second stage for lesion segmentation. Disease severity is estimated by calculating the ratio of the lesion pixel area to the leaf pixel area. StripeRustNet achieves 98.65% mean intersection over union (MIoU) for leaf segmentation and 86.08% MIoU for lesion segmentation. Validation using an additional 100 field images demonstrated a mean correlation of over 0.964 with 3 expert visual scores. To address the challenges in manual labeling, we introduce a 2-stage labeling pipeline that combines model-assisted labeling, manual correction, and spatial complementarity. We apply this pipeline to our self-collected dataset, reducing the annotation time from 20 min to 3 min per image. Our method provides an efficient and practical solution for wheat stripe rust severity assessments, empowering wheat breeders and pathologists to implement timely disease management. It also demonstrates how to address the "last mile" challenge of applying computer vision technology to plant phenomics.

Classification of the Nutritional Status of Peach Trees Using Indexes from Hyperspectral Images

This paper presents a procedure for the comparison of two technologies developed to classify peach trees according to their nutritional status. The first technology uses the leaf SPAD-502 meter value to characterize tree classes as indicated by agronomist experts: sound, intermediate, and strong chlorosis trees. It is used as a reference for the second technology, which uses a combination of two multispectral indexes computed from reflectance hyperspectral images. Specifically, R_NDVI = (R800 − R670)/(R800 + R670) and HyperSPAD = (R940/R650) are computed for each leaf pixel. An automated methodology is proposed that sets two optical thresholds (three hyperspectral categories) in view of the outliers according to a normal distribution, together with an iterative optimization of the bounding that determines the best assignment of trees to one of the three SPAD_502 levels of nutritional status, as required for practical agronomical purposes such as fertilization. The Chi 2 distribution is used to confirm the similarity of both nutritional classifications. These results encourage the use of on-board multispectral cameras to monitor the nutritional status of trees and to establish a more efficient fertilization strategy where inputs are applied according to individual status, with the consequent reduction in losses of fertilizers such as nitrogen to the atmosphere, soil, and water resulting from over-application.

Image analysis reveals environmental influences on the seagrass-epiphyte dynamic relationship for Thalassia testudinum in the northwestern Gulf of Mexico

Spatiotemporal patterns in seagrass-epiphyte dynamics for Thalassia testudinum in the northwestern Gulf of Mexico were evaluated through biomass measurements and scanned-image based metrics to investigate the potentially harmful impact of excessive epiphyte accumulations on seagrass condition. Image analysis with Spectral Angle Mapper algorithms distinguished epiphyte and uncovered seagrass leaf pixels to generate a normalized metric of leaf area coverage (epiphyte pixels/total leaf pixels). Imaging metrics were compared to biomass-based metrics seasonally, among three locations with different environmental conditions (depth, salinity, temperature and nutrient levels inferred from sediment porewater measurements) near Redfish Bay, Texas, USA. Image analysis, in conjunction with biomass measures, provides enhanced insight into the seagrass-epiphyte dynamic relationship and how it varies with environmental conditions. Compared with the biomass and morphological measures, image analysis may be more informative as an indicator of environmental changes. Variation in linear regressions of epiphyte biomass vs. epiphyte area (pixels) suggested changes in the thickness and/or density of accumulated epiphytes across environmental contexts and seasons. Two different epiphyte colonization patterns were presented based on the correlation between the normalized metrics of epiphyte load and epiphyte leaf coverage. The epiphyte load was highest at low temperatures and locations with elevated DIN:P ratio in sediment porewater. Conversely, the mean leaf coverage by epiphytes stayed relatively constant (± 10%) across seasons but differed by location (25% ~55% in this case), suggesting that leaf growth in this study is regulated to maintain the proportion of uncolonized leaf surface and that epiphyte coverage plays a role in its regulation.

NLCS - A novel coordinate system for spatial analysis on hyperspectral leaf images and an improved nitrogen index for soybean plants

Hyperspectral imaging (HSI) has been increasingly applied in plant phenotyping projects. However, most HSI systems’ imaging quality is compromised by various noise factors such as the changing ambient light, leaf slopes, and so on. In recent years, new HSI devices such as LeafSpec have been introduced to provide a higher signal-over-noise ratio along with higher spectral and spatial resolutions. However, most of the previous image processing software only calculated the averaged spectrum over the whole leaf, but rarely include the spatial distribution analysis on the leaf level. Meanwhile, different nutrient stresses could result in different spatial distribution patterns on the leaf which can be used to elevate the quality of plant phenotyping. This study focused on the development of a new methodology for spatial distribution analysis of the leaf-level HSI images. Firstly, a novel way of encoding the soybean leaf pixels to a new coordinate system called the Natural Leaf Coordinate System (NLCS) was introduced. NLCS defined the coordinates of every leaf pixel relative to the venation structure of the leaf so that the spatial distribution analysis could be conducted more intuitively. Second, a new nitrogen index based on NLCS called NLCS-N was developed and able to outperform the whole leaf averaged NDVI in terms of predicting the nitrogen content of the soybean plants and distinguishing the nitrogen-sufficient plants from the nitrogen-deficient ones more significantly.

Deep learning-based segmentation and classification of leaf images for detection of tomato plant disease.

Plants contribute significantly to the global food supply. Various Plant diseases can result in production losses, which can be avoided by maintaining vigilance. However, manually monitoring plant diseases by agriculture experts and botanists is time-consuming, challenging and error-prone. To reduce the risk of disease severity, machine vision technology (i.e., artificial intelligence) can play a significant role. In the alternative method, the severity of the disease can be diminished through computer technologies and the cooperation of humans. These methods can also eliminate the disadvantages of manual observation. In this work, we proposed a solution to detect tomato plant disease using a deep leaning-based system utilizing the plant leaves image data. We utilized an architecture for deep learning based on a recently developed convolutional neural network that is trained over 18,161 segmented and non-segmented tomato leaf images—using a supervised learning approach to detect and recognize various tomato diseases using the Inception Net model in the research work. For the detection and segmentation of disease-affected regions, two state-of-the-art semantic segmentation models, i.e., U-Net and Modified U-Net, are utilized in this work. The plant leaf pixels are binary and classified by the model as Region of Interest (ROI) and background. There is also an examination of the presentation of binary arrangement (healthy and diseased leaves), six-level classification (healthy and other ailing leaf groups), and ten-level classification (healthy and other types of ailing leaves) models. The Modified U-net segmentation model outperforms the simple U-net segmentation model by 98.66 percent, 98.5 IoU score, and 98.73 percent on the dice. InceptionNet1 achieves 99.95% accuracy for binary classification problems and 99.12% for classifying six segmented class images; InceptionNet outperformed the Modified U-net model to achieve higher accuracy. The experimental results of our proposed method for classifying plant diseases demonstrate that it outperforms the methods currently available in the literature.

Automated, image-based disease measurement for phenotyping resistance to soybean frogeye leaf spot

BackgroundFrogeye leaf spot is a disease of soybean, and there are limited sources of crop genetic resistance. Accurate quantification of resistance is necessary for the discovery of novel resistance sources, which can be accelerated by using a low-cost and easy-to-use image analysis system to phenotype the disease. The objective herein was to develop an automated image analysis phenotyping pipeline to measure and count frogeye leaf spot lesions on soybean leaves with high precision and resolution while ensuring data integrity.ResultsThe image analysis program developed measures two traits: the percent of diseased leaf area and the number of lesions on a leaf. Percent of diseased leaf area is calculated by dividing the number of diseased pixels by the total number of leaf pixels, which are segmented through a series of color space transformations and pixel value thresholding. Lesion number is determined by counting the number of objects remaining in the image when the lesions are segmented. Automated measurement of the percent of diseased leaf area deviates from the manually measured value by less than 0.05% on average. Automatic lesion counting deviates by an average of 1.6 lesions from the manually counted value. The proposed method is highly correlated with a conventional method using a 1–5 ordinal scale based on a standard area diagram. Input image compression was optimal at a resolution of 1500 × 1000 pixels. At this resolution, the image analysis method proposed can process an image in less than 10 s and is highly concordant with uncompressed images.ConclusionImage analysis provides improved resolution over conventional methods of frogeye leaf spot disease phenotyping. This method can improve the precision and resolution of phenotyping frogeye leaf spot, which can be used in genetic mapping to identify QTLs for crop genetic resistance and in breeding efforts for resistance to the disease.

Effect of the Shadow Pixels on Evapotranspiration Inversion of Vineyard: A High-Resolution UAV-Based and Ground-Based Remote Sensing Measurements

Due to the proliferation of precision agriculture, the obstacle of estimating evapotranspiration (ET) and its components from shadow pixels acquired from remote sensing technology should not be neglected. To accurately detect shaded soil and leaf pixels and quantify the implications of shadow pixels on ET inversion, a two-year field-scale observation was carried out in the growing season for a pinot noir vineyard. Based on high-resolution remote sensing sensors covering visible light, thermal infrared, and multispectral light, the supervised classification was applied to detect shadow pixels. Then, we innovatively combined the normalized difference vegetation index with the three-temperature model to quantify the proportion of plant transpiration (T) and soil evaporation (E) in the vineyard ecosystem. Finally, evaluated with the eddy covariance system, we clarified the implications of the shadow pixels on the ET estimation and the spatiotemporal patterns of ET in a vineyard system by considering where shadow pixels were presented. Results indicated that the shadow detection process significantly improved reliable assessment of ET and its components. (1) The shaded soil pixels misled the land cover classification, with the mean canopy cover ignoring shadows 1.68–1.70 times more often than that of shaded area removal; the estimation accuracy of ET can be improved by 4.59–6.82% after considering the effect of shaded soil pixels; and the accuracy can be improved by 0.28–0.89% after multispectral correction. (2) There was a 2 °C canopy temperature discrepancy between sunlit leaves and shaded leaves, meaning that the estimation accuracy of T can be improved by 1.38–7.16% after considering the effect of shaded canopy pixels. (3) Simultaneously, the characteristics showed that there was heterogeneity of ET in the vineyard spatially and that E and T fluxes accounted for 238.05 and 208.79 W·m−2, respectively; the diurnal variation represented a single-peak curve, with a mean of 0.26 mm/h. Our findings provide a better understanding of the influences of shadow pixels on ET estimation using remote sensing techniques.

Machine learning-based prediction of nutritional status in oil palm leaves using proximal multispectral images

This study evaluated the application of proximal multispectral images accompanied by 4 machine learning approaches for estimating the nutritional status of oil palm leaves. The image responded for five bands: blue, green, red, red edge, and near-infrared regions with a center wavelength of 475, 560, 668, 717, and 840 nm. Average and standard deviation (SD) values from the leaf pixels of each band were extracted, obtaining 5 average and 5 SD values from 5 bands. Thirty-four vegetation variables were generated based on those average and SD values. In total, forty-four variables consisted of 10 average-and SD-based features, and 34 vegetation variables were used as the input candidates for analyses against 10 target variables: nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), iron (Fe), manganese (Mn), zinc (Zn), boron (B), and chlorophyll (SPAD). No significant input came out for modeling with P and Zn based on the stepwise selection. Therefore, 8 nutritional models were proposed in this study. A training set with 50 samples was used to be modeled for each target, and a test set with 15 samples was employed to evaluate the models' performances. Based on random forest (RF), support vector regression (SVR), partial least square regression (PLSR), and artificial neuron network (ANN) applied to be modeled, the models for chlorophyll, N, and Ca predictions were acceptable for screening, and those for K and Mg predictions were acceptable for rough screening. The chlorophyll model developed based on the RF had the predictive statistics in terms of coefficient of determination for prediction (r2), root mean square error of prediction (RMSEP), and standard error of prediction (SEP) of 0.752, 5.46 SPAD, and 5.65 SPAD, respectively. The other 2 screening models developed based on SVR and RF for N and Ca, respectively, gave the performances with the r2, RMSEP, and SEP ranging from 0.655 to 0.718, 0.12 to 0.17%, and 0.12 to 0.18%, respectively. In the case of the 2 rough screening models established using the RF algorithm, the predictive statistics ranged from 0.496 to 0.530 for the r2 and 0.07–0.16% for both RMSEP and SEP. In this study, the Fe, Mn, and B models had poor results presenting the range of r2, RMSEP, and SEP of 0.308–0.491, 2.39–72.9 ppm, and 2.45–62.8 ppm, respectively. Based on the results, this study confirmed that the proximal multispectral information of oil palm leaves had enough significance to account for the status of chlorophyll and macro-nutrients: N, K, Ca, and Mg in the leaves.

Exploring the potential of hyperspectral imaging to detect Esca disease complex in asymptomatic grapevine leaves

• A pixel-based classification of esca-affected grapevine leaves was achieved by HSI. • Multiple pre-processing and variable selection methods were tested using PLS-DA. • Feature wavelengths were identified using automatic and interval PLS methods. • Asymptomatic leaves belonging to esca-affected vines were successfully identified. • NIR-HSI could detect foliar symptoms of grapevine Esca disease complex. Precise and reliable identification of specific plant diseases is a challenge within precision agriculture nowadays. This is the case of esca, a complex grapevine trunk disease, that represents a major threat to modern viticulture as it is responsible for large economic losses annually. The lack of effective control strategies and the complexity of esca disease expression make essential the identification of affected plants, before symptoms become evident, for a better management of the vineyard. This study evaluated the suitability of a near-infrared hyperspectral imaging (HSI) system to detect esca disease in asymptomatic grapevine leaves of Tempranillo red-berried cultivar. For this, 72 leaves from an experimental vineyard, naturally infected with esca, were collected and scanned with a lab-scale HSI system in the 900-1700 nm spectral range. Then, effective image processing and multivariate analysis techniques were merged to develop pixel-based classification models for the distinction of healthy, asymptomatic and symptomatic leaves. Automatic and interval partial least squares variable selection methods were tested to identify the most relevant wavelengths for the detection of esca-affected vines using partial least squares discriminant analysis and different pre-processing techniques. Three-class and two-class classifiers were carried out to differentiate healthy, asymptomatic and symptomatic leaf pixels, and healthy from asymptomatic pixels, respectively. Both variable selection methods performed similarly, achieving good classification rates in the range of 82.77-97.17% in validation datasets for either three-class or two-class classifiers. The latter results demonstrated the capability of hyperspectral imaging to distinguish two groups of seemingly identical leaves (healthy and asymptomatic). These findings would ease the annual monitoring of disease incidence in the vineyard and, therefore, better crop management and decision making.

Early-Stage Brown Spot Disease Recognition in Paddy Using Image Processing and Deep Learning Techniques

India is an agricultural country. Paddy is the main crop here on which the livelihood of millions of people depends. Brown spot disease caused by fungus is the most predominant infection that appears as oval and round lesions on the paddy leaves. If not addressed on time, it might result in serious crop loss. Pesticide use for plant disease treatment should be limited because it raises costs and pollutes the environment. Usage of pesticide and crop loss both can be minimized if we recognize the disease in a timely manner. Our aim is to develop a simple, fast, and effective deep learning structure for early-stage brown spot disease detection by utilizing infection severity estimation using image processing techniques. The suggested approach consists of two phases. In the first phase, the brown spot infected leaf image dataset is partitioned into two sets named as early-stage brown spot and developed stage brown spot. This partition is done on the basis of calculated infection severity. Infection severity is computed as a ratio of infected pixel count to total leaf pixel count. Total leaf pixel counts are determined by segmenting the leaf region from the background image using Otsu's thresholding technique. Infected pixel counts are determined by segmenting infected regions from leaf regions using Triangle thresholding segmentation. In the second phase, a fully connected CNN architecture is built for automatic feature extraction and classification. The CNN-based classification model is trained and validated using early-stage brown spot, developed stage brown spot, and healthy leaves images of rice plants. Early-stage brown spot and developed stage brown spot images used in training and validation are the same images that are obtained in phase 1. The experimental analysis shows that the proposed fully connected CNN-based early-stage brown spot disease recognition model is an effective approach. The classification accuracy of the suggested model is found to be 99.20%. The result of the suggested method is compared with those existing CNN-based disease recognition and classification methods that have used leaf images to recognize the diseases. It is observed that the performance of our method is significantly better than compared methods.

TheLNet270v1 - A Novel Deep-Network Architecture for the Automatic Classification of Thermal Images for Greenhouse Plants.

The real challenge for separating leaf pixels from background pixels in thermal images is associated with various factors such as the amount of emitted and reflected thermal radiation from the targeted plant, absorption of reflected radiation by the humidity of the greenhouse, and the outside environment. We proposed TheLNet270v1 (thermal leaf network with 270 layers version 1) to recover the leaf canopy from its background in real time with higher accuracy than previous systems. The proposed network had an accuracy of 91% (mean boundary F1 score or BF score) to distinguish canopy pixels from background pixels and then segment the image into two classes: leaf and background. We evaluated the classification (segment) performance by using more than 13,766 images and obtained 95.75% training and 95.23% validation accuracies without overfitting issues. This research aimed to develop a deep learning technique for the automatic segmentation of thermal images to continuously monitor the canopy surface temperature inside a greenhouse.

Plot-level rapid screening for photosynthetic parameters using proximal hyperspectral imaging.

Photosynthesis is currently measured using time-laborious and/or destructive methods which slows research and breeding efforts to identify crop germplasm with higher photosynthetic capacities. We present a plot-level screening tool for quantification of photosynthetic parameters and pigment contents that utilizes hyperspectral reflectance from sunlit leaf pixels collected from a plot (~2 m×2 m) in <1 min. Using field-grown Nicotiana tabacum with genetically altered photosynthetic pathways over two growing seasons (2017 and 2018), we built predictive models for eight photosynthetic parameters and pigment traits. Using partial least squares regression (PLSR) analysis of plot-level sunlit vegetative reflectance pixels from a single visible near infra-red (VNIR) (400-900 nm) hyperspectral camera, we predict maximum carboxylation rate of Rubisco (Vc,max, R2=0.79) maximum electron transport rate in given conditions (J1800, R2=0.59), maximal light-saturated photosynthesis (Pmax, R2=0.54), chlorophyll content (R2=0.87), the Chl a/b ratio (R2=0.63), carbon content (R2=0.47), and nitrogen content (R2=0.49). Model predictions did not improve when using two cameras spanning 400-1800 nm, suggesting a robust, widely applicable and more 'cost-effective' pipeline requiring only a single VNIR camera. The analysis pipeline and methods can be used in any cropping system with modified species-specific PLSR analysis to offer a high-throughput field phenotyping screening for germplasm with improved photosynthetic performance in field trials.

Assessing the Impact of Spatial Resolution on the Estimation of Leaf Nitrogen Concentration Over the Full Season of Paddy Rice Using Near-Surface Imaging Spectroscopy Data.

Timely monitoring nitrogen status of rice crops with remote sensing can help us optimize nitrogen fertilizer management and reduce environmental pollution. Recently, the use of near-surface imaging spectroscopy is emerging as a promising technology that can collect hyperspectral images with spatial resolutions ranging from millimeters to decimeters. The spatial resolution is crucial for the efficiency in the image sampling across rice plants and the separation of leaf signals from the background. However, the optimal spatial resolution of such images for monitoring the leaf nitrogen concentration (LNC) in rice crops remains unclear. To assess the impact of spatial resolution on the estimation of rice LNC, we collected ground-based hyperspectral images throughout the entire growing season over 2 consecutive years and generated ten sets of images with spatial resolutions ranging from 1.3 to 450 mm. These images were used to determine the sensitivity of LNC prediction to spatial resolution with three groups of vegetation indices (VIs) and two multivariate methods Gaussian Process regression (GPR) and Partial least squares regression (PLSR). The reflectance spectra of sunlit-, shaded-, and all-leaf leaf pixels separated from background pixels at each spatial resolution were used to predict LNC with VIs, GPR and PLSR, respectively. The results demonstrated all-leaf pixels generally exhibited more stable performance than sunlit- and shaded-leaf pixels regardless of estimation approaches. The predictions of LNC required stage-specific LNC~VI models for each vegetative stage but could be performed with a single model for all the reproductive stages. Specifically, most VIs achieved stable performances from all the resolutions finer than 14 mm for the early tillering stage but from all the resolutions finer than 56 mm for the other stages. In contrast, the global models for the prediction of LNC across the entire growing season were successfully established with the approaches of GPR or PLSR. In particular, GPR generally exhibited the best prediction of LNC with the optimal spatial resolution being found at 28 mm. These findings represent significant advances in the application of ground-based imaging spectroscopy as a promising approach to crop monitoring and understanding the effects of spatial resolution on the estimation of rice LNC.

Non-destructive measurement for estimating leaf area of Bellis perennis

Non-destructive measurement of approaches of modeling can be very convenient and useful for plant growth estimation. This study, digital image processing was evaluated as a non-destructive technique to estimate leaf area of Bellis perennis. The plant samples were growing in the greenhouse and the images were taken every day using Kinect camera. The proposed method used combination of L*a*b* color space, Otsu’s thresholding, morphological operations and connected component analysis to estimate leaf area of Bellis perennis. L* channel was used to distinguish the leaves and background. Calibration area uses a pot of known area in each image as a scale to calibrate the leaves area. The results show that the algorithm is able to separate leaf pixels from soil or pot backgrounds, and also allow it to be implemented in greenhouse automatically. This algorithm can be used for other plants in assumption that there is not too much leaf overlapped during measurement.

飛行ロボットによる広域自動モニタリングに関する研究

In the recently, a lot of research and development by UAV has been done. The flying robot has an advantage of being able to shoot a moving image from a viewpoint that people cannot normally see. In this paper, I propose the system for automatic monitoring of almond fields using flying robots. Specifically, I do a growth study by comparing almond leaf pixels and growth form.

Discriminating Leaves Affected with Tomato Yellow Leaf Curl through Fluorescence Imaging Using Texture and Leaf Vein Features

<abstract> Tomato yellow leaf curl disease (TYLCD), one of the most common diseases in tomato production, spreads rapidly and results in huge yield losses. This study aims to investigate the feasibility of discriminating TYLCD through fluorescence imaging. Fluorescence images (FI) of healthy and infected plants were collected with a high-speed camera, filters (690 nm), and three excitation LEDs (430 nm). Automatic iterative threshold selection and region filling methods were used to separate the leaf pixels of each image from the background. The segmented subimages containing 100% leaf pixels were selected as central images (CI). Leaf vein images (LI) were obtained from CIs through hit-or-miss transformation. In contrast to the vein morphologies of the infected leaves, those of the CIs and LIs of the healthy leaves were uniformly distributed. Gray-level co-occurrence matrix was used to extract texture features (TF) from CIs and leaf vein features (LF) from LIs. Sixteen features were evaluated using the receiver operating characteristic (ROC) curve, and rational features were selected by the area under the curves (AUC). All features were categorized into three groups: TF, LF, and their combination. For each group, features without ROC selection (TF, LF, and TF+LF) and after ROC selection (selected TF, selected LF, and selected features) were used as inputs to three classifiers: k-nearest neighbor (KNN), liner discrimination analysis (LDA), and binary logistic regression (BLR). Finally, six TF and six LF were selected. Without ROC selection, all classifiers performed with accuracies ranging from 95% to 100% in identifying infected leaves. For the TF+LF group, BLR exhibited the highest overall accuracy (100%), whereas LDA showed the lowest. After ROC selection, BLR had significantly increased classification accuracy, with higher accuracy for identifying infected leaves than for healthy leaves. This method could be used with TF and LF extracted from FI to discriminate TYLCD.

Pixel classification methods for identifying and quantifying leaf surface injury from digital images

Plants exposed to stress due to pollution, disease or nutrient deficiency often develop visible symptoms on leaves such as spots, colour changes and necrotic regions. Early symptom detection is important for precision agriculture, environmental monitoring using bio-indicators and quality assessment of leafy vegetables. Leaf injury is usually assessed by visual inspection, which is labour-intensive and to a considerable extent subjective. In this study, methods for classifying individual pixels as healthy or injured from images of clover leaves exposed to the air pollutant ozone were tested and compared. RGB images of the leaves were acquired under controlled conditions in a laboratory using a standard digital SLR camera. Different feature vectors were extracted from the images by including different colour and texture (spatial) information. Four approaches to classification were evaluated: (1) Fit to a Pattern Multivariate Image Analysis (FPM) combined with T2 statistics (FPM-T2) or (2) Residual Sum of Squares statistics (FPM-RSS), (3) linear discriminant analysis (LDA) and (4) K-means clustering. The predicted leaf pixel classifications were trained from and compared to manually segmented images to evaluate classification performance. The LDA classifier outperformed the three other approaches in pixel identification with significantly higher accuracy, precision, true positive rate and F-score and significantly lower false positive rate and computation time. A feature vector of single pixel colour channel intensities was sufficient for capturing the information relevant for pixel identification. Including neighbourhood pixel information in the feature vector did not improve performance, but significantly increased the computation time. The LDA classifier was robust with 95% mean accuracy, 83% mean true positive rate and 2% mean false positive rate, indicating that it has potential for real-time applications.

Easy Leaf Area: Automated digital image analysis for rapid and accurate measurement of leaf area.

• Premise of the study: Measurement of leaf areas from digital photographs has traditionally required significant user input unless backgrounds are carefully masked. Easy Leaf Area was developed to batch process hundreds of Arabidopsis rosette images in minutes, removing background artifacts and saving results to a spreadsheet-ready CSV file.• Methods and Results: Easy Leaf Area uses the color ratios of each pixel to distinguish leaves and calibration areas from their background and compares leaf pixel counts to a red calibration area to eliminate the need for camera distance calculations or manual ruler scale measurement that other software methods typically require. Leaf areas estimated by this software from images taken with a camera phone were more accurate than ImageJ estimates from flatbed scanner images.• Conclusions: Easy Leaf Area provides an easy-to-use method for rapid measurement of leaf area and nondestructive estimation of canopy area from digital images.