Yubin Lan

Estimation of the age and amount of brown rice plant hoppers based on bionic electronic nose use.

The brown rice plant hopper (BRPH), Nilaparvata lugens (Stal), is one of the most important insect pests affecting rice and causes serious damage to the yield and quality of rice plants in Asia. This study used bionic electronic nose technology to sample BRPH volatiles, which vary in age and amount. Principal component analysis (PCA), linear discrimination analysis (LDA), probabilistic neural network (PNN), BP neural network (BPNN) and loading analysis (Loadings) techniques were used to analyze the sampling data. The results indicate that the PCA and LDA classification ability is poor, but the LDA classification displays superior performance relative to PCA. When a PNN was used to evaluate the BRPH age and amount, the classification rates of the training set were 100% and 96.67%, respectively, and the classification rates of the test set were 90.67% and 64.67%, respectively. When BPNN was used for the evaluation of the BRPH age and amount, the classification accuracies of the training set were 100% and 48.93%, respectively, and the classification accuracies of the test set were 96.67% and 47.33%, respectively. Loadings for BRPH volatiles indicate that the main elements of BRPHs volatiles are sulfur-containing organics, aromatics, sulfur- and chlorine-containing organics and nitrogen oxides, which provide a reference for sensors chosen when exploited in specialized BRPH identification devices. This research proves the feasibility and broad application prospects of bionic electronic noses for BRPH recognition.

Improved Algorithms for the Classification of Rough Rice Using a Bionic Electronic Nose Based on PCA and the Wilks Distribution

Principal Component Analysis (PCA) is one of the main methods used for electronic nose pattern recognition. However, poor classification performance is common in classification and recognition when using regular PCA. This paper aims to improve the classification performance of regular PCA based on the existing Wilks Λ-statistic (i.e., combined PCA with the Wilks distribution). The improved algorithms, which combine regular PCA with the Wilks Λ-statistic, were developed after analysing the functionality and defects of PCA. Verification tests were conducted using a PEN3 electronic nose. The collected samples consisted of the volatiles of six varieties of rough rice (Zhongxiang1, Xiangwan13, Yaopingxiang, WufengyouT025, Pin 36, and Youyou122), grown in same area and season. The first two principal components used as analysis vectors cannot perform the rough rice varieties classification task based on a regular PCA. Using the improved algorithms, which combine the regular PCA with the Wilks Λ-statistic, many different principal components were selected as analysis vectors. The set of data points of the Mahalanobis distance between each of the varieties of rough rice was selected to estimate the performance of the classification. The result illustrates that the rough rice varieties classification task is achieved well using the improved algorithm. A Probabilistic Neural Networks (PNN) was also established to test the effectiveness of the improved algorithms. The first two principal components (namely PC1 and PC2) and the first and fifth principal component (namely PC1 and PC5) were selected as the inputs of PNN for the classification of the six rough rice varieties. The results indicate that the classification accuracy based on the improved algorithm was improved by 6.67% compared to the results of the regular method. These results prove the effectiveness of using the Wilks Λ-statistic to improve the classification accuracy of the regular PCA approach. The results also indicate that the electronic nose provides a non-destructive and rapid classification method for rough rice.

A Semantic Labeling Approach for Accurate Weed Mapping of High Resolution UAV Imagery

Weed control is necessary in rice cultivation, but the excessive use of herbicide treatments has led to serious agronomic and environmental problems. Suitable site-specific weed management (SSWM) is a solution to address this problem while maintaining the rice production quality and quantity. In the context of SSWM, an accurate weed distribution map is needed to provide decision support information for herbicide treatment. UAV remote sensing offers an efficient and effective platform to monitor weeds thanks to its high spatial resolution. In this work, UAV imagery was captured in a rice field located in South China. A semantic labeling approach was adopted to generate the weed distribution maps of the UAV imagery. An ImageNet pre-trained CNN with residual framework was adapted in a fully convolutional form, and transferred to our dataset by fine-tuning. Atrous convolution was applied to extend the field of view of convolutional filters; the performance of multi-scale processing was evaluated; and a fully connected conditional random field (CRF) was applied after the CNN to further refine the spatial details. Finally, our approach was compared with the pixel-based-SVM and the classical FCN-8s. Experimental results demonstrated that our approach achieved the best performance in terms of accuracy. Especially for the detection of small weed patches in the imagery, our approach significantly outperformed other methods. The mean intersection over union (mean IU), overall accuracy, and Kappa coefficient of our method were 0.7751, 0.9445, and 0.9128, respectively. The experiments showed that our approach has high potential in accurate weed mapping of UAV imagery.

Accurate Weed Mapping and Prescription Map Generation Based on Fully Convolutional Networks Using UAV Imagery

Chemical control is necessary in order to control weed infestation and to ensure a rice yield. However, excessive use of herbicides has caused serious agronomic and environmental problems. Site specific weed management (SSWM) recommends an appropriate dose of herbicides according to the weed coverage, which may reduce the use of herbicides while enhancing their chemical effects. In the context of SSWM, the weed cover map and prescription map must be generated in order to carry out the accurate spraying. In this paper, high resolution unmanned aerial vehicle (UAV) imagery were captured over a rice field. Different workflows were evaluated to generate the weed cover map for the whole field. Fully convolutional networks (FCN) was applied for a pixel-level classification. Theoretical analysis and practical evaluation were carried out to seek for an architecture improvement and performance boost. A chessboard segmentation process was used to build the grid framework of the prescription map. The experimental results showed that the overall accuracy and mean intersection over union (mean IU) for weed mapping using FCN-4s were 0.9196 and 0.8473, and the total time (including the data collection and data processing) required to generate the weed cover map for the entire field (50 × 60 m) was less than half an hour. Different weed thresholds (0.00–0.25, with an interval of 0.05) were used for the prescription map generation. High accuracies (above 0.94) were observed for all of the threshold values, and the relevant herbicide saving ranged from 58.3% to 70.8%. All of the experimental results demonstrated that the method used in this work has the potential to produce an accurate weed cover map and prescription map in SSWM applications.

Study of the similarity and recognition between volatiles of brown rice plant hoppers and rice stem based on the electronic nose

Abstract The volatiles of Brown rice plant hopper (BRPH) itself is an important evidence for BRPH electronic nose detection. However, during infestation, BRPH always sucks the juice from the rice stem, therefore, a study on the similarity between BRPH’s volatiles and undamaged rice stem volatiles might help determine whether the volatile contents of BRPH would be influenced by the sucking of the rice stem juice. If so, recognizing BRPH from rice stem should be a crucial step to reduce the misjudgment of BRPH occurrence prediction by using electronic nose, which has not been reported until now. This paper used an electronic nose (PEN3) sample of the volatile of U3IN (under the 3th-instar nymphs), O3IN (over the 3th-instar nymphs) and healthy rice stem. Hierarchical clustering analysis (HCA), Loading analysis (Loadings), principal component analysis (PCA), k-nearest neighbor (KNN), probabilistic neural network (PNN), and support vector machine (SVM) were used for data analysis. HCA, Loadings, and PCA results proved that certain similarities exist between volatiles of rice stem and BRPH, Loadings and PCA results also indicated the volatile similarity between O3IN and rice stem is stronger than the volatile similarity between U3IN and rice stem. To reduce the redundant information and improve computation efficiency, according to Loadings and PCA results, sensor R5 of electronic nose has been be removed, then, the fist four principle components has been kept as the feature values. KNN, PNN and SVM all can recognize rice stem, O3IN, and U3IN effectively, however, KNN and PNN are more fit to solve the problem of rice stem and BRHP recognition than SVM. This experiment results proved that certain similarities exist between volatiles of rice stem and BRPH, also figured out the feasible way to recognize rice stem and BRPH, which could provide a reference for further research of BRPH prediction.

Spectral Properties of Crops at Different Growth Stages

Timely detection and remediation of volunteer cotton plants in both cultivated and non-cultivated habitats is critical for completing boll weevil eradication in Central and South Texas. However, timely detection of cotton plants over large areas and habitats is a challenging process. We examined the spectral reflectance properties of cotton, corn, soybean, and grain sorghum during different growth stages in 2009 to determine whether the spectral properties of plants could be used to distinguish cotton from other crops. Two blocks were set up according to the soil types in the TAMU farm, which are Belk clay (BaA) and Ships clay (ShA). Cotton, corn, soybean and sorghum were planted and managed in each block using conventional production practices for the area. Spectral information was collected from all crops at different growth stages from May to July. Reflectance spectra and the first derivative of the spectra were analyzed to characterize the spectral properties of crop varieties and compare the crops grown in different soil types. The results showed that the reflectance spectra of different crops could be differentiated at the early vegetative and late growth stages. At the vegetative stage, cotton could be distinguished from other crops; however, the reflectance spectra of soybean and sorghum were not significantly different between 410 and 560 nm. At the reproductive stage, cotton could be distinguished from soybean and sorghum between 510 and 590 nm and 650 and 900 nm. The red edge position could also be used to distinguish cotton, corn, soybean and sorghum at the vegetative growth stage and cotton, soybean and sorghum at the reproductive growth stage. The red edge points of cotton, soybean and sorghum shifted with the growth stages of development. No correlation was found between crop height and individual wavelength, but the difference of the heights of cotton plants grown in two blocks may explain that the reflectance of cotton plants in block BaA was more than those in block ShA.