Profile plot-based rice bacterial leaf blight hyperspectral image dimensionality reduction method

Abstract

The invention discloses a profile plot-based rice bacterial leaf blight hyperspectral image dimensionality reduction method, and belongs to the field of hyperspectral information extraction. The method comprises the following steps: selecting characteristic waveband images in the spectral dimension of a hyperspectral image by using a successive projections algorithm, establishing corresponding profile plots by using the characteristic waveband images, calculating the gray value difference of the profile plots of different positions of rice leaves, and further choosing minimum characteristic wavebands needed by realization of detection of rice leaf bacterial leaf blight disease spots in order to reduce the dimensionality of the hyperspectral images needed by the disease spot detection. The method combining the successive projections algorithm with a profile plot technology realizes effective dimensionality reduction of high-dimensional data, and the obtained characteristic images can be used to accurately identify disease spots; and the disease spot area and the disease degree can be accurately calculated through combining image identification, and detection of the rice bacterial leaf blight disease spots is realized by using the minimum characteristic images, so the complex degree of a detection model is reduced, and the detection time is effectively shortened.

Description

technical field [0001] The invention belongs to the field of hyperspectral information extraction and relates to spectral data analysis, in particular to a dimensionality reduction method for hyperspectral images of rice bacterial blight based on line graphs. Background technique [0002] The hyperspectral image is a three-dimensional data cube (Image Cube), and the two-dimensional image records the spatial shape information of the sample, and each image corresponds to a spectral band. Hyperspectral image technology combines sample spectral information and image information to realize crop nutrient estimation and disease detection. When rice is infected with bacterial blight, the diseased spots on the leaves will gradually expand and the disease will worsen; in farmland management, it is necessary to spray pesticides according to the severity of the disease. One of the main problems of using hyperspectral images for disease detection is that the obtained data has a dimensio...

AI Technical Summary

Problems solved by technology

[0005] Since there are many bands involved in hyperspectral image detection, it is not easy to obtain the best combination of disease characteristics by only relying on the feature band extraction of the spectral dimension.

Images