Soil profile type recognition method based on three-dimensional spectrum curved surface partition features

Abstract

The invention discloses a soil profile type recognition method based on three-dimensional spectrum curved surface partition features. The method comprises the steps of 1, acquiring spectral reflectance data of a generation layer of any soil profile in a soil sample library, and performing depth interpolation treatment, depth unification treatment and continuum removal treatment; 2, generating a three-dimensional spectrum curved surface according to the spectral reflectance data processed by the step 1; 3, extracting statistical features of the three-dimensional spectrum curved surface per partition, and recording the features into a sub-feature set fk, wherein k is a current soil profile serial number; 4, repeatedly executing the steps 1 to 3 until all soil profiles in the soil sample library are processed, thus acquiring a feature set F that is equal to {fk / k=1, ..., K}, wherein K is the quantity of soil profiles in the soil sample library; 5, based on the feature set F, performing random forest training for many times, wherein each category sampling quantity with the same proportion is set during each training, and generating a random forest classification model set M; and 6, based on the random forest classification model set M, recognizing the category of the soil profiles to be recognized. The method provided by the invention is relatively good in recognition effect.

Description

technical field [0001] The invention relates to the technical field of geography and geographic information, in particular to a soil profile type identification method based on three-dimensional spectral curved surface partition features. Background technique [0002] Soil classification is based on the comprehensive differences in soil physical properties, classifying soil types at different levels, so as to promote agricultural technology and improve soil according to local conditions. The identification of soil types requires the acquisition of a large amount of diagnostic information such as soil morphology, physics, chemistry, and even biology. Some of the information can be obtained by visual observation or simple measurement, but most of the physical and chemical information traditionally requires laboratory testing and analysis. only available. Therefore, the identification and identification of soil types is usually costly and requires the involvement of soil taxon...

AI Technical Summary

Problems solved by technology

[0005] However, in the current application of soil type identification based on visible-near-infrared diffuse reflectance spectroscopy, due to the unbalanced number of different types of samples, the different depths of the occurrence layers of different sections, and the inconsistent collection depth of soil samples in different sections, the identification accuracy is not as good as that of other reasons. not ideal

Images