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

Infrared thermal image nondestructive testing method based on convolutional neural network

A technology of convolutional neural network and infrared thermal imaging, which is applied in the field of non-destructive testing of infrared thermal imaging, to achieve the effect of large detection range, intuitive detection range and accurate detection

Pending Publication Date: 2020-06-23
HARBIN INST OF TECH
View PDF0 Cites 9 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 non-destructive detection of the existing convolutional neural network cannot be applied to infrared imaging, and propose a non-destructive detection method for infrared thermal images based on convolutional neural network

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
  • Infrared thermal image nondestructive testing method based on convolutional neural network
  • Infrared thermal image nondestructive testing method based on convolutional neural network
  • Infrared thermal image nondestructive testing method based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0035] A nondestructive detection method for infrared thermal images based on a convolutional neural network in this embodiment, such as figure 1 As shown, the method is realized through the following steps:

[0036] Step 1. Arrange the infrared image data collection scene, and collect the defective infrared image of the object to be tested;

[0037] Step 2, performing enhancement and noise reduction processing on the collected defective infrared image of the object to be measured, and completing the preprocessing process;

[0038] Step 3, using the preprocessed infrared images with defects of the object to be tested to augment and construct the data set;

[0039] Step 4, merging the models of VGG16 and DenseNet169 networks, and using the data set in step 3 to train and test the fused model;

[0040] Step 5, using the fused network model to identify and detect defects in the defective infrared image of the object to be tested.

specific Embodiment approach 2

[0042] The difference from the first specific embodiment is that in this embodiment, a non-destructive detection method for infrared thermal images based on convolutional neural network, in the second step, the collected infrared images with defects of the object under test are enhanced and reduced. Noise processing, complete the preprocessing process, specifically:

[0043] First, calculate the multi-contrast joint image of absolute contrast, change contrast, normalized contrast, standard contrast and differential absolute contrast for the collected infrared image, so as to enhance the defect part in the infrared image;

[0044] In addition, the wavelet transform threshold is used to remove noise, and the non-uniform heating, noise and distortion existing in the image are removed. Since the infrared imaging system is susceptible to interference, and there will be uneven heating in the image, compared with the visible light image, the noise is greater. Infrared detection itsel...

specific Embodiment approach 3

[0047] The difference from the second embodiment is that in this embodiment, a non-destructive detection method for infrared thermal images based on convolutional neural network, the use of wavelet transform threshold to remove noise, remove uneven heating, noise and distortion existing in the image The process is:

[0048] First, decompose the red signal;

[0049] Then, according to the energy concentration of the signal after the wavelet transform, that is, the characteristic that it usually concentrates on the larger wavelet coefficients, if you want to obtain the signal, you must separately process the wavelet coefficients at high and low resolutions, and retain the large wavelet coefficients. The wavelet coefficients of low-resolution scales; the wavelet coefficients corresponding to high-resolution in other scales cannot be directly determined to be retained. It is necessary to set a threshold, compare the amplitude of these wavelet coefficients with the threshold, and r...

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 infrared thermal image nondestructive testing method based on a convolutional neural network, and belongs to the field of image recognition. The problem that nondestructivetesting of an existing convolutional neural network cannot be applied to infrared imaging is solved. An infrared thermal image nondestructive testing method based on a convolutional neural network comprises the following steps: arranging an infrared image data acquisition scene, and acquiring an infrared image with defects of a to-be-tested object; performing enhancement and noise reduction processing on the acquired defective infrared image of the to-be-detected object to complete a preprocessing process; carrying out data set augmentation and construction by utilizing the preprocessed defective infrared image of the to-be-detected object; fusing the models of the VGG16 network and the DenseNet169 network, and training, testing and identifying the fused model by using the data set; and identifying and detecting defects in the defective infrared image of the to-be-detected object by using the fused network model. The identification precision of the detection method provided by the invention reaches 98.5%.

Description

technical field [0001] The invention relates to a method for non-destructive detection of infrared thermal images based on a convolutional neural network. Background technique [0002] In recent years, deep learning algorithms have performed well in areas such as image classification, pattern recognition, and object detection. By applying the convolutional neural network, a cutting-edge algorithm based on computer vision, to non-destructive testing technology, it can not only verify its feasibility and applicability, but also provide new ideas for visual inspection and other non-destructive testing technologies. development is of great significance. [0003] The combination of convolutional neural network and non-destructive testing can improve the detection performance, and can perform fast, real-time and accurate detection of defects. However, the current non-destructive testing of convolutional neural networks is mainly used in visible light images, electromagnetic ultr...

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
IPC IPC(8): G06T7/00
CPCG06T7/0002G06T2207/10048G06T2207/20081G06T2207/20084
Inventor 迟永钢范嘉麟夏岳隆逄博
Owner HARBIN INST OF 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