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

Industrial image anomaly detection method based on deep dual network feature matching

A feature matching and image anomaly technology, applied in image analysis, image data processing, neural learning methods, etc., can solve the problem of inadaptability, industrial image anomaly detection methods cannot operate robustly, and cannot effectively find and identify anomalies or defects. and other problems to achieve the effect of strong image feature expression ability

Pending Publication Date: 2021-11-30
北京中科智眼科技有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is that the existing industrial image anomaly detection methods cannot robustly operate in the industrial production process with no abnormal samples or a small number of labeled samples, and cannot effectively discover and identify various types of anomalies or abnormalities that may occur. The shortcomings of defects can no longer meet the requirements of the increasingly upgraded and optimized industrial production process

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
  • Industrial image anomaly detection method based on deep dual network feature matching
  • Industrial image anomaly detection method based on deep dual network feature matching
  • Industrial image anomaly detection method based on deep dual network feature matching

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0028] Such as Figure 1-2 As shown, the industrial image anomaly detection method based on deep dual network feature matching proposed in the embodiment of the present invention includes the following steps:

[0029] (1) Given an image x to be detected (set its width to 256, height to 256, and number of channels to 3), use a deep convolutional network with a total of 16 convolutional layers pre-trained on the ImageNet image database through classification tasks VGG19 obtains the image depth features in, Represents the output of the oth layer of the network, and its width is set to w o , the height is h o , the number of channels is c o ; (indicated by the mathematical symbol φ).

[0030] (2) Input the image x to be detected and the convolutional neural network with dual characteristics of the relative network φ Get the depth features of the image in, Indicates the o-th layer output of the deep neural network, whose width is w o , the height is h o , the number ...

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 industrial image anomaly detection method based on deep dual network feature matching, and the method comprises the following specific steps: (1) giving a to-be-detected image x, setting the width of the to-be-detected image x as w, the height of the to-be-detected image x as h, and the number of channels of the to-be-detected image x as c, (2) inputting an image x to be detected into a convolutional neural network having dual characteristics with the relative network phi to obtain the depth feature of the image; (3) according to the depth representations obtained in the steps (1) and (2), and (4) according to the step (4), calculating the abnormal score of the image x; and (5) binarizing the abnormal score Ai, j obtained in the step (4) according to a segmentation threshold T given by a user, namely representing the position where the abnormal score is greater than T by 1. According to the method, a pre-trained deep convolutional network and a deep network which is structurally dual with the pre-trained deep convolutional network are included, and the feature matching relation between the two deep networks is directly modeled and evaluated.

Description

technical field [0001] The invention relates to the field of machine vision, in particular to an industrial image anomaly detection method based on deep dual network feature matching. Background technique [0002] In recent years, with the rapid development of artificial intelligence technology, the traditional manufacturing industry is gradually moving towards intelligent manufacturing. Intelligent quality inspection technology based on machine vision anomaly detection can automatically identify abnormalities or defects of industrial products and evaluate product quality, which can effectively improve the production quality and efficiency of manufacturing industries such as industrial parts manufacturing and consumer electronics production. It has a very wide range of applications. Application scenarios and prospects. [0003] Existing industrial image anomaly detection methods are mainly anomaly detection and recognition based on statistical learning models and image chan...

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): G06T7/00G06K9/62G06N3/04G06N3/08
CPCG06T7/0004G06N3/08G06N3/045
Inventor 齐志泉杨洁
Owner 北京中科智眼科技有限公司
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