High-resolution image classification and denoising method and system for disease and dead wood of pine trees

Abstract

The present invention discloses a high-resolution image classification and denoising method and system for the disease and dead wood of pine trees. The method comprises: according to classification training of ground object color features of verified pest tree spectrum information, performing corresponding ground object color feature extraction on all pixel points in a to-be-classified image sample; selecting a target pixel set conforming to the multi-level disease pine tree feature from the to-be-classified image sample after being subjected to the ground object color feature extraction; performing denoising on the target pixel set conforming to the multi-level disease tree feature for the first time; performing denoising on the target pixel set conforming to the multi-level disease treefeature for the second time according to the distribution features of the disease tree; and finally, performing denoising on the target pixel set conforming to the multi-level disease tree feature forthe third time according to the forest area background, and generating a classification result of the disease and dead wood state of pine trees. Thus, according to the technical scheme of the presentinvention, a high-resolution image of the disease and dead wood of pine trees can be identified with high detection accuracy of different disease areas and different disease degrees, and the processing speed is fast and extremely stable.

Description

technical field [0001] The invention relates to a method for identifying diseased and dead pine trees, in particular to a method and system for classifying and denoising high-resolution images of diseased and dead pine trees. Background technique [0002] The spectral feature method is the most common method for identifying pine trees with pests and diseases. It is based on the fact that when the plant is infested by pests and diseases, the difference in physiological changes will be reflected in the spectral characteristics, especially the difference in spectral characteristics between the red region and the near-infrared region. Use imaging equipment such as spectrometers to obtain spectral data information of reflection and radiation, and on this basis, discover the changing rules of reflection spectra in different bands. [0003] Compared with the traditional manual field survey method, obtaining the spectral feature information of objects in remote sensing images to ide...

AI Technical Summary

Problems solved by technology

[0005] The support vector data description multi-classification method is based on image pixels for classification and recognition. One of the disadvantages is that a large number of samples need to be trained, so the execution efficiency is slow; moreover, this method cannot realize the identification of multi-level diseased pine trees, such as early infection, mid-term Infection, late infection and dead pine trees; in addition, it has the disadvantages of high misjudgment rate and relatively low recognition accuracy when solving multi-classification recognition problems

Images