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

Deep learning identification method based on X-ray image, system and X-ray machine

A deep learning and recognition method technology, applied in the field of deep learning intelligent recognition based on X-ray images, can solve problems such as high cost and slow detection speed, achieve the effect of improving yield, ensuring recognition speed and accuracy, and improving recognition effect

Pending Publication Date: 2020-02-07
艾偲睿科技(厦门)有限责任公司
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, the existing machines require human-machine collaboration and human eye recognition, and the detection rate is slow. Multiple machines are required to detect at the same time, resulting in high costs. Therefore, it is necessary to improve the intelligent recognition ability and automation of machine detection.

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
  • Deep learning identification method based on X-ray image, system and X-ray machine

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0072] Defects in Existing Defectives include:

[0073] Weld seam: desoldering, virtual welding, incomplete fusion, incomplete penetration, etc.;

[0074] Castings: slag inclusion, porosity, sand hole, shrinkage cavity, slag hole, air hole, crack, cold shut, etc.;

[0075] Chains: empty rings, cracks in connecting rings, base metal cracks, depressions, gaps, duckbills, weld cracks, deformation, dislocation, etc.;

[0076] Composite products: voids, faults, dislocations, foreign matter, impurities, fiber uniformity, wrinkles, cracks, etc.

[0077] A deep learning recognition method based on X-ray images, comprising the steps of:

[0078] Step 1. Train the Faster R-CNN deep learning model according to the X-ray images of defective products, extract the defect information data in the X-ray images of defective products, generate the defect data set by manual calibration and establish a defect database, and Defect information data is classified and stored in the defect database ...

Embodiment 2

[0088] A deep learning recognition method based on X-ray images, comprising the steps of:

[0089] Step 1. Input the X-ray image of the gap or depression of the missing chain into the preset YOLOv3 deep learning model, extract the defect information data in the X-ray image of the defective product, establish a defect database, and store the defect information data according to the different defects. Classified and stored in the defect database;

[0090] Step 2: Obtain the reference value as 0.5mm;

[0091] Step 3. Obtain the X-ray image of the chain, and the X-ray image is synthesized by a plurality of basic X-ray images through the color-difference stereoscopic imaging technology;

[0092] Use the mean filter or high-pass filter algorithm to denoise and enhance the X-ray images, distinguish and arrange according to the types of defects presented in the X-ray images, and select the classification of the defect database corresponding to the defect types,

[0093] Pick out all...

Embodiment 3

[0099] A deep learning recognition method based on X-ray images, comprising the steps of:

[0100] Step 1. Input the X-ray image of the fiber uniformity of the missing composite material product into the preset SSD deep learning model, extract the defect information data in the X-ray image of the defective product, establish a defect database, and store the defect information data according to the defect Different categories are stored in the defect database;

[0101] Step 2: Obtain the benchmark value as 30%;

[0102] Step 3. Obtain the X-ray image of the chain, and the X-ray image is synthesized by multiple basic X-ray images through the color separation and stereoscopic imaging technology;

[0103] Use the mean filter or high-pass filter algorithm to denoise and enhance the X-ray images, distinguish and arrange according to the types of defects presented in the X-ray images, and select the classification of the defect database corresponding to the defect types,

[0104] P...

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 relates to the technical field of detection, and specifically relates to a deep learning identification method based on an X-ray image. The method comprises the following steps: traininga deep learning model according to the X-ray image of a defective product, extracting defect information data in the X-ray image of the defective product, establishing a defect database, and storingthe defect information data in the defect database; acquiring a benchmark value; acquiring the X-ray image of a workpiece, performing feature extraction for the X-ray image, and performing comparisonwith the defect database and calculating whether contrast is less than the benchmark value, if so, judging the workpiece as being qualified; otherwise, judging the workpiece as being unqualified, andstoring the X-ray image of which the contrast is greater than the benchmark value into the defect database. According to the method provided by the invention, by follow-up continuous workpiece detection, the defect information data of the workpieces with different defects can be collected, more defect information data can be obtained, and thus, identification speed and precision of the defective product can be guaranteed, and identification effect can be improved.

Description

technical field [0001] The invention relates to the technical field of detection, in particular to an X-ray image-based deep learning intelligent recognition method. Background technique [0002] In the process of industrial mass production, humans cannot stare at the workpiece for a long time to see if there are defects. Using human eyes to identify and inspect the quality of workpieces is inefficient, has a high error rate, and is extremely prone to eye fatigue. Using the machine detection method can greatly improve the detection accuracy, production efficiency and production automation. [0003] However, the existing machines require human-machine collaboration and human eye recognition, and the detection rate is slow. Multiple machines are required to perform detection simultaneously, resulting in high costs. Therefore, it is necessary to improve the intelligent recognition ability and automation of machine detection. Contents of the invention [0004] The technical p...

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): G01N23/00G06N20/00
CPCG01N23/00G06N20/00
Inventor 李顺仁方正李珣李磊
Owner 艾偲睿科技(厦门)有限责任公司