Urban building extraction method based on deep learning

A technology of deep learning and extraction methods, applied in neural learning methods, biological neural network models, instruments, etc., can solve the problems of poor detail perception, low semantics, and less convolution, etc., to improve feature expression and recognition ability, Increase the feature receptive field and improve the effect of semantics

Pending Publication Date: 2022-01-11
HAI YING GROUP OF AEROSPACE IND
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0002] Model construction of urban buildings to present a three-dimensional model structure, which can effectively design the existing urban layout and planning, and the technology used in the existing model construction method has a higher resolution of the low-level feature space and includes more locations , detailed information, but due to fewer convolutions, its semantics are lower and the noise is more. High-level features have stronger semantic information, but the spatial resolution is very low, and the perception of details is poor. Therefore, we A deep learning-based extraction method for urban buildings is proposed

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Urban building extraction method based on deep learning
  • Urban building extraction method based on deep learning
  • Urban building extraction method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0023] A method for extracting urban buildings based on deep learning, comprising the following steps:

[0024] S1. Perform data enhancement processing on the training samples;

[0025] S2. Propose a deep semantic segmentation model of encoding layer-intermediate layer-decoding layer;

[0026] S3. The middle layer uses dilated convolution for cascading mode stacking;

[0027] S4. Deconvolution and upsampling is performed on the decoding layer, and features of different scales are added and fused.

[0028] Specifically, the specific enhancement processing method of step S1 is to perform random horizontal flipping of the samples through the column transformation of the matrix, random vertical flipping through the row transformation of the matrix, and random rotation, random translation and random scaling through the affine transformation of the matrix.

[0029] Specifically, the model segmentation in step S2 is based on adding a residual network model in the network coding lay...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an urban building extraction method based on deep learning. The method comprises the following steps: S1, carrying out data enhancement processing on a training sample; S2, proposing a coding layer-middle layer-decoding layer depth semantic segmentation model; S3, performing cascade mode stacking on the middle layer by using expansion convolution; S4, performing deconvolution up-sampling on the decoding layer, and adding and fusing different scale features; expansion convolution is used for cascading mode stacking, feature receptive fields are increased, and features of different scales and features of different resolutions are fused, so that through multiple times of convolution and hole injection, original low-layer features are guaranteed, high-layer semantics is improved, noise is reduced, and the building feature expression and recognition capacity is improved for high-layer pairs.

Description

technical field [0001] The invention relates to the technical field of urban building model construction, in particular to a method for extracting urban buildings based on deep learning. Background technique [0002] Model construction of urban buildings to present a three-dimensional model structure, which can effectively design the existing urban layout and planning, and the technology used in the existing model construction method has a higher resolution of the low-level feature space and includes more locations , detailed information, but due to fewer convolutions, its semantics are lower and the noise is more. High-level features have stronger semantic information, but the spatial resolution is very low, and the perception of details is poor. Therefore, we A method for extracting urban buildings based on deep learning is proposed. Contents of the invention [0003] The object of the present invention is to provide a method for extracting urban buildings based on deep...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06V10/80G06V10/46G06V20/10G06V10/774G06V10/82G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/2148G06F18/253
Inventor 苏晓玉宋歌朱俊志张东映罗蔚然洪志明
Owner HAI YING GROUP OF AEROSPACE IND
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap