Remote sensing image forest land extraction method and device

By using the ResNet network to replace the encoder in the U-Net network in remote sensing image processing, and introducing a jump connection structure and attention module, problems such as missed division and missing division in the forest land extraction process in the prior art are solved, and a more accurate and coherent forest land extraction effect is achieved.

CN119723330APending Publication Date: 2025-03-28CHINA SURVEY SURVEYING & MAPPING TECH
0 Cites 1 Cited by

Patent Information

Application Number
CN202411766890.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-04
Publication Date
2025-03-28

AI Technical Summary

Technical Problem

The existing remote sensing image processing method based on convolutional neural network model has problems such as missed division and missing division, weak coherence, large block boundary deviation, many point-like fragmented map spots, and inaccurate image recognition and segmentation effect in the forest land extraction process.

Method used

The ResNet network is used to replace the encoder in the U-Net network, and a jump connection structure and attention module are introduced into the image segmentation model to build an image segmentation model to improve the accuracy of forest land extraction.

Benefits of technology

It effectively reduces the phenomenon of missed division and missed division, enhances coherence, the block boundaries are more consistent with the actual boundaries, and the dot-like fragmented plots are reduced, so the segmentation results are more accurate.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a remote sensing image forest land extraction method and device, and belongs to the technical field of remote sensing image processing, and the extraction method comprises the steps: carrying out the preprocessing of a remote sensing image, and obtaining a plurality of sample images of a preset size; constructing an image segmentation model, taking a decoder in the U-Net network structure as a decoder of the image segmentation model, and taking a structure in which an average pooling layer, a full connection layer and an activation function layer are removed in the ResNet network as an encoder of the image segmentation model; training an image segmentation model by using the sample image to obtain a target image segmentation model; and inputting a remote sensing image to be extracted into the target image segmentation model to obtain forest land extraction information. According to the remote sensing image forest land extraction method, the ResNet network is used for replacing an encoder in the U-Net network, the classification performance of an image segmentation model can be effectively improved, the phenomena of wrong classification and missing classification are effectively reduced, the continuity is enhanced, the boundary of the image block is matched with the actual boundary, the point block broken image spots are reduced, and the segmentation result is more accurate.
Need to check novelty before this filing date? Find Prior Art