Soil texture image classification and spatial prediction method and system based on deep learning

CN121921580BActive Publication Date: 2026-06-09NORTHWEST A & F UNIV

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Patents(China)
Current Assignee / Owner: NORTHWEST A & F UNIV
Filing Date: 2026-03-27
Publication Date: 2026-06-09

AI Technical Summary

Technical Problem

Existing technologies struggle to quickly and accurately identify soil texture types, especially lacking the ability to distinguish between similar types and the ability to identify the vertical hierarchical structure of soil profiles. They also cannot optimize image acquisition quality under complex lighting conditions in the field.

Method used

A deep learning-based method for soil texture image classification and spatial prediction is adopted. By standardizing image acquisition, multi-scale visual feature recognition, profile-layered texture classification, and environmental covariate fusion, combined with confidence feedback control, a complete technical closed loop is formed to achieve automatic identification of soil texture types and prediction of regional spatial distribution.

Benefits of technology

It achieves automated identification of soil texture types and prediction of regional spatial distribution, improves the ability to distinguish fine-grained soil types from adjacent transitional types, eliminates image differences, improves classification accuracy and stability, and breaks through the limitations of traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN121921580B_ABST

Patent Text Reader

Abstract

The present application relates to the technical field of computer vision and soil science, and discloses a soil texture image classification and spatial prediction method and system based on deep learning, which comprises standard image acquisition, multi-scale visual feature recognition network classification, image type judgment, profile layering and layer-by-layer texture determination, environmental covariate fusion spatial prediction and confidence feedback regulation, and comprises a standard image acquisition module, a soil texture visual feature recognition module, a texture layering recognition module, an environmental covariate fusion prediction module and a soil texture archive and mapping output module. Through the collaborative design of a multi-scale double-branch network structure and a channel attention mechanism, comprehensive extraction and effective fusion of multi-dimensional visual features such as soil color, structure morphology, pore distribution and particle fineness are realized, and the fine-grained differentiation ability for adjacent transition type soils is significantly improved.

Need to check novelty before this filing date? Find Prior Art