Method and device for detecting cotton diseases based on multisource light information technology

Abstract

The invention relates to a method and device for detecting cotton diseases based on a multisource light information technology. The device comprises a computer, a control module, a light information sensor, an LED combination light source, an electric control displacement table and an instrument rack. The method comprises the steps as follows: acquiring a multisource light information data cube of cotton leaves, comparing differences between sample images and backgrounds and differences between scab images and normal parts at different wavelengths, and dividing the backgrounds and scabs by utilizing a normalized difference vegetation map; extracting a main component image, and determining the characteristic wavelength distribution of the diseases; obtaining the polarization characteristics at the characteristic wavelengths by applying a VirtualLab software, and extracting the reflection strength distribution of a visual spectrum in a target region and the grey-scale and texture features of the image; and dividing the characteristic information of the cotton diseases into three types, establishing polarization characteristic space, strength characteristic space and image characteristic space respectively, and carrying out fusion decision judgment by utilizing three SVM (Support Vector Machine) sub-classifiers and a D-S (Dempster-Shafer) evidence theory reasoning method. The method can be used for accurately identifying and evaluating the cotton diseases.

Description

technical field [0001] The invention relates to a detection method for cotton diseases, in particular to a non-destructive detection method for cotton diseases based on multi-source optical information technology, and also relates to a device for implementing the detection method. Background technique [0002] In the process of cotton field management, disease is one of the main problems that plague the growth of cotton. It seriously endangers the growth of cotton and affects the yield and quality of cotton. Traditional control methods mainly rely on the experience of farmers or experts to identify the type and severity of the disease, and then apply chemical agents. This method of disease identification is subjective, slow, high-intensity, high in misrecognition rate, and poor in real-time performance. Often lead to waste of resources and environmental pollution. Cotton diseases can cause changes in the color, texture, and internal tissue structure of leaves, plants, and c...

AI Technical Summary

Problems solved by technology

At present, there are no relevant patents and reports at home and abroad

Images