Method for identifying early-stage disease spots of sclerotinia sclerotiorum and botrytis of rape

Abstract

The invention discloses a method for identifying early-stage disease spots of sclerotinia sclerotiorum and botrytis of rape. The method comprises steps as follows: (1), a high-spectrum image of a to-be-detected leaf of the rape is taken, and the pixel of a local disease spot of the high-spectrum image is picked; (2), a band ratio operation value and a band differential operation value of the pixel at the band of 681.95 nm-748.23 nm are calculated and obtained; (3), a feature vector composed of a spectrum value of the pixel at the band of 555.29 nm, the band ratio operation value and the band differential operation value is input into a trained least square support vector machine model, and the type of the disease spot is judged according to an output result. A band with a characteristic difference is obtained from the high-spectrum image information with the early-stage disease spots of sclerotinia sclerotiorum and botrytis through analysis, the early-stage disease spots of sclerotinia sclerotiorum and botrytis are accurately classified by combination of a characteristic wavelength, the band ratio operation value, the band differential operation value and a least square support vector machine, and the identification accuracy is higher.

Description

technical field [0001] The invention relates to the technical field of digital agriculture, in particular to a method for distinguishing early lesions of rape sclerotinia and botrytis cinerea. Background technique [0002] Rapeseed is an important oil crop in my country. It is often threatened by various diseases during the planting process, resulting in reduced yield. The main diseases of rapeseed are sclerotinia and gray mold. Sclerotinia is caused by S. sclerotiorum infection, and gray mold is caused by Botrytis cinerea infection. When the two diseases infect rapeseed leaves in the early stage, water-soaked lesions appear on the rapeseed leaves, and it is difficult to distinguish the two diseases, which is easy to cause misjudgment and miss the best control time. If these two diseases are accurately identified in a fast and non-destructive way, farmers can be guided to apply pesticides for specific diseases, improving efficiency and protecting the environment to the gre...

AI Technical Summary

Problems solved by technology

[0005] However, few studies have used hyperspectral imagery to identify different diseases with unusually similar symptoms on the same plant

T. Rumpf uses hyperspectral information combined with vegetation indices (VegetationIndices) and support vector machines (SupportVectorMachines) to distinguish healthy sugar beet leaves from leaves infected with three types of sugar beet diseases, such as rust, cercospora leaf spot and powdery mildew. The total identification accuracy exceeds 86%, but the accuracy is not ideal when the three kinds of diseases are identified in pairs (RumpfT., MahleinA.K., SteinerU., et al. Early detection and classification of plant diseases with Support Vector Machines based on hyperspectral reflection [J]. Computers and Electronics in Agriculture, 2010, 74 (1): 91-99.)

This is an unresolved problem in the research of hyperspectral image technology for identification of plant diseases

Images