Unlock instant, AI-driven research and patent intelligence for your innovation.

Defect detection method of continuous casting billet surface image based on depth convolution neural network

A neural network and deep convolution technology is applied in the field of image defect detection on the surface of continuous casting billets to ensure production quality, improve detection efficiency, and reduce manual workload.

Active Publication Date: 2019-01-15
SHANGHAI JINYI INSPECTION TECH +1
View PDF13 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The technical problem to be solved by the present invention is to provide a method for detecting surface image defects of continuous casting slabs based on deep convolutional neural network. Manual workload, convenient storage and traceability of defect data, ensuring the production quality of continuous casting slabs

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
  • Defect detection method of continuous casting billet surface image based on depth convolution neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0013] Implementation example figure 1 As shown, the method for detecting defects in continuous casting slab surface images based on deep convolutional neural network of the present invention includes the following steps:

[0014] Step 1. Preprocess the image of the known continuous casting billet, starting from the image origin, cropping the image area of ​​256×256 pixels in order from top to bottom every 128 pixels and from left to right every 7 pixels as image block data Set, the image block containing complete defects is used as the defect sample set, the normal and non-defective image blocks are used as the normal sample set, three-quarters of the cropped image blocks are randomly selected as the training set, and one-fourth as the verification set and test Set and convert to LMDB format data set;

[0015] Step 2: Use a deep convolutional neural network composed of four layers of convolutional layers + three layers of fully connected layers + Softmax classification layers as ...

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 defect detection method of a continuous casting billet surface image based on a depth convolution neural network. Firstly, the method pretreats a known continuous casting billet image to obtain a defect sample set and a normal sample set, and randomly selects three quarters of the image blocks as a training set and the rest as a verification set and a test set. Deep convolution neural network is used to classify the defect area, and the training set and verification set are used to test the image block classification model of the test set under the set parameters, andthe image block classification model with high accuracy is obtained. To test the image to be inspected, the defect image block is defined as the defect image, and the defect image is classified and predicted by the image block classification model. A certain algorithm is used to judge whether the prediction result is true or false, and the true defect image is obtained. The method effectively improves the detection efficiency, reduces the manual workload, facilitates the storage and tracing of defect data, and ensures the production quality of the continuous casting billet.

Description

Technical field [0001] The invention relates to a continuous casting billet surface image defect detection method based on a deep convolution neural network. Background technique [0002] In the continuous casting slab production, due to the influence of the production process, various defects such as cracks, foreign objects or dents will inevitably occur. Serious defects will have an adverse effect on the next rolling process. Therefore, the continuous casting slab surface image is used for defect detection Very important. The defect detection of the traditional continuous casting slab surface image is done by manual reading. Although it has certain accuracy, the cost is high, and the manual workload is large, the efficiency is low, and it is not conducive to the storage and traceability of defect data. Therefore, the continuous casting slab The surface image realizes automatic defect detection and marking, which is of great significance to the production of continuous casting ...

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): G06T7/00G06T7/13G06T7/60G06T7/62G06K9/62
CPCG06T7/0004G06T7/13G06T7/60G06T7/62G06T2207/20084G06T2207/20081G06T2207/20021G06T2207/30116G06F18/24G06F18/214Y02P90/30
Inventor 刘晗刘志庄新卿王向阳胡嘉成薛松袁楚雄张公俊
Owner SHANGHAI JINYI INSPECTION TECH