Shaokun Li

Research of Maize Leaf Disease Identifying Models Based Image Recognition

The methods of recognition and diagnosis of main maize leaf diseases using machine vision were studied in the paper. Threshold method was adopted to do image segmentation, and area-marking method was used calculating the num of disease as well as wiping off redundancy dots. And then Freeman link code was used to calculate form feature. Finally diseases were deduced according to bin-tree search method. In the research, the exclusive feature of main maize leaf diseases was presented, and the flow of disease diagnosis was confirmed and the recognition models were developed. The results indicate that the precision of five kinds of maize disease recognition is higher than 80%. It shows that this method is available for recognizing maize disease, and provides technique support for the automatic recognition of disease by compiling the system with reasonable process flow.

Automatic segmentation of crop leaf spot disease images by integrating local threshold and seeded region growing

At present, the region growing algorithm has been used as a segmentation technique of digital images. Most region growing algorithms are using fixed or determinate criterions to distinguish disease spots from leaf image with gray level differences between leaf and disease spot. But in practice, the objects in the disease leaf image have fuzziness and uncertainty, and edges of the objects are unclear. Whats more, the color of leaf and disease spots is uneven, and the gray level is overlapping, so it is difficult to use fixed threshold or determinate criteria to determine the uncertain objects in leaf disease spot images accurately. In order to improve the crop leaf spot disease image segmentation accuracy, an adaptive segmentation algorithm by integrating local threshold and seeded region growing (LTSRG) is proposed. The algorithm was implemented on VC6.0. The segmentation algorithm uses the pixels of which the R-channel gray level is more than the G-channel gray level as initial seed points (pixels), and then local threshold Ci is calculated for each connected seed region by Otsu. New seed pixels are included and the threshold C is re-calculated until no new seed pixel can be included. The results of LTSRG are compared with the results of threshold-based Otsu and clustering -based EM. The experiments show: The adapted segmentation method is satisfactory and highly efficient to separate disease spots from normal part of corn leaves. LTSRG algorithm is easy to realize, and can improve the precision of crop disease spot segmentation. Its image segmentation results have good region consistency and high efficiency. It is an adapting algorithm for image segmentation.

Spectrum Characteristics of Cotton Canopy Infected with Verticillium Wilt and Inversion of Severity Level

Verticillium wilt of cotton is one of the diseases of cotton with extensive occurrence and maximal harming in our country even in the world .Hyper spectrum remote sensing with the fine spectrum information has becoming the efficient method to monitor the Verticillium wilt of cotton. The research was conducted in Xinjiang, the largest cotton plant region of China. The paper used data which was collected both canopy spectrum infected with verticillium wilt and SL (severity level) in the year 2005 2006, the quantitative correlation were analyzed between SL and canopy reflectance spectrum derivative spectrum. The tested results indicated that spectrum characteristics of cotton canopy infected with verticillium wilt had better regularity with the increase of SL in different periods and varieties. Spectrum reflectance increased in visible light region (620 700nm) with the increase of the SL, which inverted in nearinfrared region, and extreme signification in 78

Evaluating the severity level of cotton Verticillium using spectral signature analysis

High spatial or spectral resolution remote sensing might be an efficient method for estimating Verticillium wilt incidence in cotton. The objectives of this study were to characterize leaf spectra and the physiological and biochemical parameters of cotton (Gossypium hirsutum) damaged by Verticillium dahliae Kleb. (simply, Verticillium) to determine the wavelengths of those leaves that were most responsive to cotton with Verticillium and to develop a spectral model to predict the severity levels (SLs) of Verticillium through evaluation of the SLs of cotton leaves with Verticillium at different growth stages using reflectance and the first derivative (FD) spectrum. The study revealed that the values of the physiological and biochemical parameters all gradually decreased with increasing SLs in cotton leaves infected with Verticillium. The spectral characteristics of cotton leaves infected with Verticillium were significant compared to healthy ones. The reflectance of cotton leaves increased with increasing SLs of SLs disease in the range of 400–2500 nm (excluding 700–900 nm). The values of FD spectrum changed significantly at the red edge of the chlorophyll absorption feature (680–740 nm). The wavelength position of the red edge shifted towards shorter wavelengths and the red-edge swing decreased with respect to increasing SLs. From this study, the raw spectral bands of 437–724 and 909–2500 nm and the FD spectra bands of 535–603 and 699–750 nm can be selected as sensitive bands for estimating the SLs of disease in cotton leaves. Inversion models have been established to estimate the SLs of cotton leaves infected with Verticillium. Of all models, the model of R 700nm/R 825nm was superior for quantitatively estimating the disease SLs of cotton leaves infected with Verticillium in practice: its root mean square error (RMSE) was 0.866 and relative error (RE) was only 0.012. Thus, both the selected wavelength ranges and the chosen reflectance models were good indicators of damage caused by Verticillium to cotton leaves. The results provide theoretical support for large-scale monitoring of cotton infected with Verticillium by air- and spaceborne remote sensing.

Canopy Spatial Distribution and Identification Using Hyperspectal Data in Winter Wheat

Winter wheat varieties (Triticum aestivum L.) with different leaf angle distributions (LADs) were used in this experiment. Results showed that varieties with planophile LADs had leaf orientation values (LOVs) ranging from 25.12 to 40.33, whereas varieties with erectophile LADs had LOVs ranging from 67.5 to 73.25. Canopy spectral reflectance was measured using a ground‐based spectroradiometer. Correlation analysis indicated that LOV affected canopy spectra more than leaf area index (LAI) before the jointing stage. The LAI had the greatest effect after the ground was nearly completely covered. Discriminant analysis showed that simultaneous measurements of normalized difference vegetation index (NDVI) and cover can differentiate LADs in those wheat varieties with similar population magnitude at the jointing stage. In addition, by using increments of the canopy spectral reflectance at different growth stages, the planophile varieties with low LAIs can be differentiated from the erectophile varieties with high LAIs, which cannot be achieved using NDVI and cover. Using ΔR890 as the reflectance increment of booting stage and jointing stage and R890 as the reflectance of 890‐nm energy at the jointing stage, different varieties presented distinctly different scatter plot representations (X = ΔR890, Y = R890). This analysis also indicted that varieties with different LADs can be clustered and identified qualitatively in the plot despite their population magnitude, also validated by discriminant analysis.

Extracting Winter Wheat Planting Area Based on Cropping System with Remote Sensing

Winter wheat is one kind of important crop in China. It’s planting area is one key element to explain yield change. To obtain winter wheat planting area as soon as possible can provide scientific reference for our country’s making related policy. Basing on the cropping system in Shandong province, winter wheat is divided into two kinds “winter wheat sowed by machine-maize” and “people broadcast winter wheat-rice”. Using MODIS data, NDVI characters of winter wheat, garlic, greenhouse vegetable, from sowing till overwintering stage were analyzed. Together with NDVI characters of former stubble crops in middle September, extracting requirements were set up for winter wheat planting area which was sowed by machine this year. In view of the spectrum similarity between rice wheat and greenhouse vegetable from sowing stage till overwintering stage, rice wheat planting area of former year was extracted relying on the character of biomass rapid growth at jointing stage. Because of the “people broadcast winter wheat-rice” cropping system is very fixed in Shandong province, then the rice wheat planting area of former year can take the place of the rice wheat planting area this year. Two kinds of winter wheat area were merged, and tested by 284 groups of located spots data, the accuracy reached 94.01%. The result showed that it is feasible to extract winter wheat area before overwintering stage, and the time is 4 months earlier than using jointing stage NDVI.

A Leaf Layer Spectral Model for Estimating Protein Content of Wheat Grains

The spectral signatures of crop canopies in the field provide much information relating morphological or quality characteristics of crops to their optical properties. This experiment was conducted using two winter-wheat (Triticum aestivum) cultivars, Jingdong8 (with erect leaves) and Zhongyou9507 (with horizontal leaves). We analysiced the relation between the direction spectral characteristics and the laeves nitrogen content(LNC). The result showed that the spectral information observed at the 0° angle mainly provided information on the upper canopy and the lower layer had little impact on their spectra. However, the spectral information observed at 30° and 60° angles reflected the whole canopy information and the status of the lower layer of the canopy had great effects on their spectra. Variance analysis indicated that the ear layer of canopy and the topmost leaf blade made greater contributions to CDS. The predicted grain protein content (GPC) model by leaf layers spectra using 0° view angle was the best with root mean squares (RMSE) of 0.7500 for Jingdong8 and 0.6461 for Zhongyou9507. The coefficients of determination, R2 between measured and estimated grain protein contents were 0.7467 and 0.7599. Thus, grain protein may be reliably predicted from the leaf layer spectral model.

Spectral Characteristics of Cotton Infected with Verticillium Wilt and Severity Level of Disease Estimated Models

In order to elucidated characteristics of spectrum of cotton leaf infected with Verticillium wilt and estimated its severity level (SL) to provide theoretic foundation for further monitoring cotton Verticillium wilt at large scale using airborne and airspace remote sensing. The spectrum reflectance of cotton single leaf infected with Verticillium wilt was measured in cotton disease nursery and field at different growth phases, meanwhile, SL of single leaf infected with Verticillium wilt was investigated. The methods of first derivative spectraum were used to estimate accurately disease of cotton with Verticillium wilt when compared with the reflectance spectrum of different single leaf infected of Verticillium wilt. The results indicated that Spectral characteristic of cotton leaf of Verticillium wilt had better regularity with the increase of SL in different periods and varieties. Spectral reflectance increased significantly at visible light region (400–700nm) and near -infrared region (700–1300nm) with the increase of the SL, and specially signification at blue — violet to red regions(525–680nm). when SL got 25%, cotton leaf of Verticillium wilt could be used as a watershed and diagnosed index in early time. There were evident different characteristics of first derivative spectra in these disease leave, it changed significantly in red edge ranges(680–780nm) with different disease level, derivative spectra of red edge swing decreased, and red edge position equal moved to the blue. The thesis indicated that 434–724nm and 909–1600nm were selected out as sensitive bands region to SL of single leaf. Some inversion models for estimating cotton leaf diseased level of Verticillium wilt all reached the best significantly level. The model in which the first derivative spectra at 723nm could invert accurately the cotton leaf SL, and it may be used to forecasting the position of cotton leaf infected with Verticillium wilt in quantitatively.

Recognizing wheat plant-type using NDVI and cover degree

Plant-type is the main influential factor to canopy spectral for its influence on canopy structure, so it is very important to identify wheat plant-type for improving the estimating precision of leaf biochemical composition in the field by remote sensing. In this experiment, we selected ten winter wheat varieties of two plant-types and analyzed their canopy spectral reflectance (CSR), normalized difference vegetation index (NDVI), vegetation cover degree (COV) and leaf area index (LAI). The results show that varieties with lax leaves have higher canopy spectral reflectance in near infrared region than those with erect leaves from jointing to booting stage, and the differences are most obvious at jointing stage. Through changing the plant density by manual, the relationships between NDVI and COV were analyzed in different densities, and the results indicated that with COV increasing, NDVI increased but the difference between two plant-types became diminishing. COV of lax varieties is higher than that of erect ones under the same NDVI at jointing stage when their COV are low and obviously different between two plant-type varieties. Furthermore, there is significant difference of NDVI between erect and horizontal varieties (by testing, P<0.05) at jointing stage, which proved to be the best stage to identify plant-type using NDVI and COV, for NDVI has positive correlativity with LAI.

Estimating wheat grain protein content from ground-based hyperspectral data using a improved detecting method

Chinese Academy of Agricultural Sciences Room 104, Institute of Crop Sciences, No.12 South Street, Zhongguancun, Beijing, 100081,China. luylwheat@hotmail.com *shaokun0004@sina.com.cn. AbstractThat high correlation between nitrogen concentration and grain protein content (GPC) made it possible to predict grain quality indirectly by measuring nitrogen status in wheat. However, former studies all based on the canopy spectral reflectance (Rc) measurement method, by which mixed spectra such as ears, leaves and soil background spectrum were achieved. So it is very difficult to abstract useful information from the canopy spectra, and the precision can not be heightened authentically. Our objective was to test the feasibility of measuring ear-layer spectral reflectance (Rel) using a improved method and to realize prediction of wheat grain quality before harvest. The results showed that Rel measured by the improved method was proved more effective to estimate GPC of wheat than Rc measured by traditional method. In relation to Rc, Rel was much closer to pure ear spectral reflectance(Re), for the Rel measured by the improved method was not disturbed by factors such as soil background and plant-type. In this article, the correlations between ear nitrogen total content (ETNC) and 20 spectral characteristic parameters were also analyzed respectively, and the results indicated that ear layer spectral characteristic parameters have stronger correlativity with ETNC than canopy ones. Rel and GPC all had highly correlative relation with ETNC, so a new model was established to predict indirectly GPC using ratio vegetable index (RVI[890,670]) through two models linking, and by testing, R=0.6617, RSME=0.8509 by the improved method and R=0.8653 RSME=0.7339 by traditional method. it was proved that the improved method produced higher coefficient of determination and lower root mean square error (RMSE) than traditional method under the same condition, and the precision by RVI[890,670] calculated from Rel was 13.75 percent higher than from Rc, which indicated that the predictive model established with the improved measurement method is more reliable and practicability than with the traditional method. This study made it possible to predict wheat grain quality using RVI and laid the foundation for portable nitrogen and protein monitor exploiting.