Abstract: Hyperspectral imaging technology combines the advantages of spectral detection and image detection, and has obvious advantages in crop nutrient diagnosis and disease and insect pest degree judgment. In recent years, some scholars at home and abroad have successfully applied hyperspectral imaging technology to the diagnosis of crop nutrients and disease stress, and have made significant progress.
Because hyperspectral imaging has the advantage of combining maps and spectra, one point on the leaves can be accurately detected to detect the characteristics of different stress symptoms of crops, and the spectral information of the stressed crops can be obtained, which comprehensively reflects the degree of stress on the crops. . Therefore, hyperspectral imaging has become a hot research topic at home and abroad. At present, scholars use hyperspectral imaging technology to quantitatively extract various stress characteristics suffered by crops, analyze the leaves and local regions of the leaves according to high-resolution images, and finally characterize the degree of stress, so that they can be more microscopic. Mechanism research on the scale. Using the SOC710 / SOC730 hyperspectral imaging spectrometer of the United States SOC company, the hyperspectral images of wheat leaves under stress of nutrients and pests and diseases can be collected, and the spectral characteristics are enhanced by the pixel-by-pixel average method. Then, the spectral characteristics of the leaves at different leaf positions and different diseases 1. Extraction and analysis of leaf spectral characteristics of insect stress, providing theoretical exploration for hyperspectral imaging technology for crop stress diagnosis.
1. Spectral feature extraction and analysis of leaves (leaves at different leaf positions) under nutrient stress
According to the growth and development rules of crops, the nutrients of leaves at different leaf positions will differ in levels, and at the same time, the spectral information of the leaves will also be different. The purpose of this paper is to extract the spectral information of the leaves in different leaf positions from the hyperspectral image of the leaves. The purpose is to eliminate the point data errors caused by the uneven distribution of the leaves, and to more accurately detect the spectral differences caused by the lack of nutrients on the vertical gradient of the crop. In the visible light band, the reflectivity of the lower leaves is the highest, the middle layer is the lowest, and the red and green bands at 550-650nm are particularly obvious; secondly, there are absorption valleys near 680nrn, and the difference between the lower layer and the middle and upper layers is large; 700-760nm The band is a characteristic red edge of vegetation. The middle and upper leaves are red-shifted due to good growth conditions, while the lower leaves are yellowed due to lack of nutrients, and the red edge moves toward the blue wave. The main reason for the difference in spectral characteristics is the difference in chlorophyll content in the leaves. This is consistent with previous research results, that is, chlorophyll content is a sensitive factor for analyzing spectral changes. The visible light band is characterized by strong reflection of chlorophyll a and chlorophyll b in the blue-green light region and slow absorption in the red light region.
The near infrared band of 780 ~ 900nm is opposite to the visible band. The upper and middle leaves have strong vitality, the cell structure is intact, and the light forms multiple scattering inside the leaves, so the reflectivity value is high; while the nutrient deficiency of the lower leaves leaves yellow, the chlorophyll content decreases, and the cell structure changes, causing its reflectance to be significantly lower than the middle upper layer. This is consistent with the research results under different fertilization treatments, that is, the near-infrared wave band is mainly judged by the difference in multiple scattering caused by the leaf cell structure.
To sum up, in this study, when using SOC710 portable visible-near infrared imaging spectrometer to diagnose the leaf nutrient stress of different leaf positions, it can not only extract the hyperspectral differences of the leaves in different leaf positions, but also visually visualize it according to the imaging map. Judgment, this provides an advantageous means for the accurate diagnosis of crop nutrient stress.
2. Extraction and analysis of leaf spectral characteristics under powdery mildew stress
Because the hyperspectral cube of the leaf contains two-dimensional spectral information, each pixel on the image has continuous spectral information in the band interval, which has a unique advantage for quantitative qualitative analysis when crops are subjected to disease stress. With the help of SOC710 hyperspectral imaging spectrometer with high spatial resolution during near-ground observation, the effect of the number of lesions and the area of ​​infection on the leaves can be quantitatively studied. In the visible band, the spectral properties of crops are mainly affected by chlorophyll content. Normal leaves will have a reflection peak of 550nm and an absorption valley of 680nm. After being infected with powdery mildew, the chlorophyll content in the leaves decreases, which intuitively appears as light yellow and white. Under sunlight, the absorption value decreases, so the reflection value is higher. In the near infrared band of 780-900nm, the spectral reflection intensity of the normal part of the leaf is significantly different from that of the diseased part, but the difference between the number of different diseased spots is narrower than that of the visible part. The general rule is normal (0%)> mild (5%)> q1 degree (15%)> severe (30%)> severe (50%), which is the opposite of visible light. This is because the spectral characteristics of the blade in the near-infrared band are determined by the cell structure inside the blade. The cells in the normal blade will scatter light multiple times, so the reflectivity is high, and the powdery mildew affects the internal structure of the blade, so the reflection The rate value decreases.
3. Spectral feature extraction and analysis of leaves under aphid stress
The SOC710VP visible-near infrared imaging spectrometer can extract and analyze the spectrum of aphids and even individual aphids, which provides a new method for quantitatively studying the damage of aphids to wheat leaves. Normal leaves have the most significant absorption valley in the red light band at 680nm, followed by attached aphids and damaged leaves. Aphids also have similar spectra. Normal leaves have the highest reflectivity value in the near infrared band, followed by damaged leaves and leaves with aphids, indicating that the internal cell structure of normal leaves is normal and forms multiple reflections, so the reflection value is the highest. The reflectivity of aphids gradually increases with the increase of the wave band, but it is lower than the reflectivity value of the leaves before and after the pest.
The reflectance of the normal leaves and the leaves damaged by aphids in the 450-500nm, 560-680nm, and 750-900nm bands is obviously different, which can be used as the identification band for judging the aphid stress. The reflectance of leaves with aphids and leaves damaged by aphids is not much different, and they can be collectively referred to as the stress state of crops suffering from insect pests. In the 450 ~ 700nm band, the aphids have the highest reflectance, but they also show a reflection peak of 550nm and an absorption valley of 680nm, indicating that the leaves have more or less a certain contribution; but at the red edge of 700 ~ 780nm, it is obviously and green. Spectral characteristics of plants are different, so the red edge can also be used as a characteristic area for judging aphid diseases.
4. Discussion
The natural growth of crops is a dynamic process, and the stresses that affect their growth and development are diverse and complex. We first remove other influencing factors and study from the single-leaf scale. The purpose is to find sensitive bands that affect their changes, and then build nutrient diagnosis. 1. Early warning and prediction models of diseases and insect pests, and then verification at the field scale, can be truly used in the production management of precision agriculture.
When crops are affected by stress, they will show different spectral responses, and capturing the sensitive spectral differences is the basis of remote sensing research. On the data cube of the combined map, not only the spectral information of spot-like lesions and aphids is obtained, but also the information of the planar leaves infected by it is extracted, and the combination of points and planes is used to explain the spectral changes of the leaves after stress. feature. This is the advantage of the near-ground application of the imaging spectrometer.
Pastry Tools,Baking Tools,Cake Decorating Tools,Cake Decorating Kit
Xiongyang Household Co., Ltd , https://www.xyutensil.com