Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Construction waste remote sensing image recognition method based on deep learning

A construction waste, remote sensing image technology, applied in neural learning methods, character and pattern recognition, instruments, etc., can solve the problems of inability to mine overall information, low recognition accuracy, and remote sensing image recognition methods are easily interfered, and achieve dynamic tracking. Monitor the effect of purifying the urban environment, shortening the revisit cycle, and saving manpower

Inactive Publication Date: 2021-10-22
HARBIN UNIV OF SCI & TECH
View PDF4 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problem that the existing remote sensing image recognition method is susceptible to interference, and these cannot mine the overall information, resulting in low recognition accuracy, and propose a remote sensing image recognition method based on deep learning for construction waste

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
  • Construction waste remote sensing image recognition method based on deep learning
  • Construction waste remote sensing image recognition method based on deep learning
  • Construction waste remote sensing image recognition method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0026] A deep learning-based remote sensing image recognition method for construction waste in this embodiment, such as figure 1 As shown, the method is realized through the following steps:

[0027] Step 1. Preprocessing the acquired remote sensing image to obtain a remote sensing image dataset;

[0028] Step 2, expand the sample of the remote sensing image data set, add L2 regularization penalty term in the seventh layer of neural network, use the expanded data set to train the network model adding L2 regularization penalty term, and obtain the target recognition model;

[0029] Step 3. Improve the semantic segmentation algorithm by calculating the mIOU ratio of the intersection and union of the real value and the predicted value in DeepLab's semantic segmentation method;

[0030] Step 4, using the improved recognition model and algorithm for image recognition.

specific Embodiment approach 2

[0031] The difference from Embodiment 1 is that in this embodiment, a deep learning-based remote sensing image recognition method for construction waste, the step of preprocessing the acquired remote sensing image described in Step 1 to obtain a remote sensing image dataset includes:

[0032] Due to the influence of the overall positioning accuracy of the remote sensing information platform and the sensor error rate, the layers of the panchromatic and multispectral images in the satellite remote sensing technology image may not be aligned, etc., requiring more preprocessing operations on the satellite remote sensing image, using the ENVI platform Perform remote sensing image preprocessing operations such as orthorectification and image fusion on remote sensing images, and perform histogram averaging operations on the resulting data; improve the relative position accuracy of remote sensing data, improve image quality, and enhance data features.

specific Embodiment approach 3

[0033] The difference from the specific embodiment 1 or 2 is that, in this embodiment, a construction waste remote sensing image recognition method based on deep learning, the step of expanding the sample of the remote sensing image dataset described in step 2 adopts an improved generative confrontation network method The features of multiple images are fused, and the specific steps include:

[0034] Aiming at the problem of fewer samples in the building remote sensing data set, the sample expansion experiment of image generation is carried out, and urban construction waste is detected from the perspective of semantic segmentation of remote sensing images, so as to provide a reliable data source expansion method for remote sensing monitoring of urban construction waste, and provide a basis for the management of construction waste stockpiles. provide technical support. Improve the generative confrontation network, improve the accuracy of the semantic segmentation network, and m...

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 a construction waste remote sensing image recognition method based on deep learning, and belongs to the field of remote sensing image recognition. The existing remote sensing image recognition method is easily interfered, and the whole information cannot be mined, so that the recognition precision is low. A construction waste remote sensing image recognition method based on deep learning comprises the following steps: preprocessing an obtained remote sensing image to obtain a remote sensing image data set; expanding a remote sensing image data set sample, adding an L2 regularization penalty term to a seventh layer of the neural network, and training the network model added with the L2 regularization penalty term by using the expanded data set to obtain a target recognition model; achieving the improvement of a semantic segmentation algorithm by calculating the mIOU ratio of an intersection set and a union set of a true value set and a predicted value set in a DeepLab semantic segmentation method; and carrying out image recognition by using the improved recognition model and algorithm. The method is high in recognition precision, the processing progress of illegal stacking can be monitored, and the urban environment is dynamically tracked, monitored and purified.

Description

technical field [0001] The invention relates to a deep learning-based remote sensing image recognition method for construction waste. Background technique [0002] Remote sensing image recognition has roughly gone through the following processes: Traditional remote sensing image recognition methods based on pixels, such as maximum likelihood method and K-Means mean method, but image spectral brightness information is easily disturbed, and these cannot mine overall information. It is easy to produce "salt and pepper noise", and now it is only used as a comparison item or a preprocessing method; although the object-oriented remote sensing recognition method takes advantage of the rich attributes of polygonal objects, it is easy to over-segment or under-segment, and the scale of the segmentation is different. It is easy to determine; the same pixel-based image semantic segmentation has become a popular research direction in remote sensing image recognition today. Its self-learn...

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): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/241G06F18/253G06F18/214
Inventor 颜子健董静薇
Owner HARBIN UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com