CN112435271A - Firing profile segmentation method applied to laser spot quality measurement - Google Patents
Firing profile segmentation method applied to laser spot quality measurement Download PDFInfo
- Publication number
- CN112435271A CN112435271A CN202011391635.0A CN202011391635A CN112435271A CN 112435271 A CN112435271 A CN 112435271A CN 202011391635 A CN202011391635 A CN 202011391635A CN 112435271 A CN112435271 A CN 112435271A
- Authority
- CN
- China
- Prior art keywords
- burning
- firing
- photosensitive paper
- profile
- segmentation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
本发明公开了一种应用于激光光斑质量测量的灼烧轮廓分割方法，包括制作灼烧光敏纸数据集，对灼烧光敏纸数据集进行标注和数据增强，基于DeepLabV3+构建灼烧轮廓分割神经网络后，将其训练为灼烧轮廓分割模型，完成测试后的灼烧轮廓分割模型能直接对测量图片进行精确、可靠的轮廓分割，且具有鲁棒性高、成本低、通用普适性强的优点，极为适用和有利于测量判断激光光束质量的好坏。
The present invention discloses a burn contour segmentation method for laser spot quality measurement, comprising preparing a burn photosensitive paper data set, annotating and data enhancing the burn photosensitive paper data set, constructing a burn contour segmentation neural network based on DeepLabV3+, and then training it into a burn contour segmentation model. The burn contour segmentation model after the test can directly perform accurate and reliable contour segmentation on the measurement picture, and has the advantages of high robustness, low cost, and strong universality, and is extremely applicable and conducive to measuring and judging the quality of the laser beam.
Description
Technical Field
The invention relates to an image contour segmentation method, in particular to a burning contour segmentation method applied to laser spot quality measurement.
Background
The laser spot is one of important characteristics of the quality of the laser beam and the performance of a laser, how to quickly, accurately and simply extract the shape profile characteristics of the laser spot has important significance for evaluating the quality of the laser beam, and M is2The factor is an important parameter for evaluating the quality of the laser beam, but because the measuring method is complex, the instrument is expensive, the requirement on the environment is high, most of the conditions are limited to be measured under laboratory conditions, and the method cannot be conveniently applied to the complex environment in the industry, the laser is incident on the photosensitive paper, the burning profile of the photosensitive paper is analyzed after laser spots are burned out, and the method is also a method for judging the quality of the laser beam.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides the firing profile segmentation method which has good robustness, high accuracy and strong general universality and can be applied to laser spot quality measurement.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a firing profile segmentation method applied to laser spot quality measurement comprises the following steps:
(1) acquiring a firing image of laser-fired photosensitive paper to manufacture a firing photosensitive paper data set, wherein the firing photosensitive paper data set comprises a training set and a testing set;
(2) carrying out pixel level two classification on the burning photosensitive paper data set, marking the part burned by the burning image as a foreground through a marking tool, and marking the part not burned as a background;
(3) performing data enhancement on the two classified burning photosensitive paper data sets;
(4) constructing a burning contour segmentation neural network based on DeepLabV3+, and setting the ratio of the input image resolution to the output resolution of the burning contour segmentation neural network as 4;
(5) training the firing profile segmentation neural network into a firing profile segmentation model through the training set after data enhancement, and then testing and evaluating the firing profile segmentation model through the test set after data enhancement;
(6) and burning the photosensitive paper by laser which actually needs to measure the quality to form a measurement picture, inputting the measurement picture into the burning profile segmentation model, and outputting a segmentation profile which can be used for judging the beam quality by the burning profile segmentation model.
And the data enhancement of the step 3 comprises the turning over and the angle rotation of the burning image and the angle rotation after the turning over.
The labeling tool of the step 2 is Labelme.
The invention has the beneficial effects that: the method comprises the steps of manufacturing a burning photosensitive paper data set, labeling and data enhancing the burning photosensitive paper data set, building a burning contour segmentation neural network based on DeepLabV3+, training the neural network into a burning contour segmentation model, and directly performing accurate and reliable contour segmentation on a measured picture by the burning contour segmentation model after testing.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a flow chart of the method of the present invention.
Detailed Description
Referring to fig. 1, a burning profile segmentation method applied to laser spot quality measurement includes the following steps:
(1) the method comprises the steps of collecting firing images of laser firing photosensitive paper to manufacture a firing photosensitive paper data set, wherein the firing photosensitive paper data set comprises a training set and a testing set, the training set is only used for training a firing profile segmentation neural network, the testing set cannot be used for training and can only be used for evaluating a firing profile segmentation model, in the firing image manufacturing process, multi-angle firing is carried out on the photosensitive paper through different illumination and lasers, firing images with different laser beam qualities are formed, and the firing profile segmentation model after subsequent training and testing is more robust.
(2) Carrying out pixel level two classification on the burning photosensitive paper data set, marking the part burned by the burning image as a foreground through a marking tool Labelme, and marking the part not burned as a background;
(3) performing data enhancement on the two classified burning photosensitive paper data sets; the data enhancement comprises the turning of the burning image, the angle rotation and the turned angle rotation, so that the burning contour segmentation model after the subsequent training test has the advantage of high universality.
(4) The burning contour segmentation neural network is constructed based on DeepLabV3+, and the burning contour segmentation neural network is mainly different from the DeepLabV3+ network in that the ratio of the input image resolution to the output resolution of a DeepLabV3+ network coding stage is 16, the ratio of the input image resolution to the output resolution of the burning contour segmentation neural network is 4, and if the original ratio of the original DeepLabV3+ network is adopted, the burning contour segmentation model has large deviation on the contour segmentation of a measurement picture, so that the burning contour segmentation model cannot have the characteristic of high accuracy, and a more precise boundary segmentation effect cannot be obtained.
(5) The firing profile segmentation neural network is trained into a firing profile segmentation model through the training set after data enhancement, then the firing profile segmentation model is tested and evaluated through the testing set after data enhancement, the loss value of the training set and the precision of the training set are continuously observed in the training process, and when the firing profile segmentation model is converged and the best performance is achieved in the testing process, the training can be stopped.
(6) After the photosensitive paper is burned by the laser which actually needs to measure the quality to form a measurement picture, the measurement picture is input into the burning profile segmentation model, and the burning profile segmentation model outputs a segmentation profile (namely a segmentation result of the laser facula) which can be used for judging the quality of the laser beam, so that the accurate and reliable segmentation profile is provided for measuring the quality of the laser facula, and the method is extremely suitable for and beneficial to measuring and judging the quality of the laser beam.
The above embodiments do not limit the scope of the present invention, and those skilled in the art can make equivalent modifications and variations without departing from the overall concept of the present invention.
Claims (3)
1. A burning profile segmentation method applied to laser spot quality measurement is characterized by comprising the following steps:
(1) acquiring a firing image of laser-fired photosensitive paper to manufacture a firing photosensitive paper data set, wherein the firing photosensitive paper data set comprises a training set and a testing set;
(2) carrying out pixel level two classification on the burning photosensitive paper data set, marking the part burned by the burning image as a foreground through a marking tool, and marking the part not burned as a background;
(3) performing data enhancement on the two classified burning photosensitive paper data sets;
(4) constructing a burning contour segmentation neural network based on DeepLabV3+, and setting the ratio of the input image resolution to the output resolution of the burning contour segmentation neural network as 4;
(5) training the firing profile segmentation neural network into a firing profile segmentation model through the training set after data enhancement, and then testing and evaluating the firing profile segmentation model through the test set after data enhancement;
(6) and burning the photosensitive paper by laser which actually needs to measure the quality to form a measurement picture, inputting the measurement picture into the burning profile segmentation model, and outputting a segmentation profile which can be used for judging the beam quality by the burning profile segmentation model.
2. The method of burning profile segmentation as claimed in claim 1, wherein the data enhancement of step 3 comprises flipping of burning image, angular rotation and angular rotation after flipping.
3. The method of claim 1, wherein the labeling tool in step 2 is Labelme.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011391635.0A CN112435271B (en) | 2020-12-02 | 2020-12-02 | Firing contour segmentation method applied to laser spot quality measurement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011391635.0A CN112435271B (en) | 2020-12-02 | 2020-12-02 | Firing contour segmentation method applied to laser spot quality measurement |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112435271A true CN112435271A (en) | 2021-03-02 |
| CN112435271B CN112435271B (en) | 2023-06-20 |
Family
ID=74690762
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011391635.0A Active CN112435271B (en) | 2020-12-02 | 2020-12-02 | Firing contour segmentation method applied to laser spot quality measurement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112435271B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1450304A1 (en) * | 2003-02-21 | 2004-08-25 | City University of Hong Kong | Image processing apparatus and method |
| CN101832941A (en) * | 2010-03-19 | 2010-09-15 | 天津大学 | Fruit quality evaluation device based on multispectral image |
| CN108955582A (en) * | 2018-06-21 | 2018-12-07 | 中国人民解放军战略支援部队航天工程大学 | Laser focal beam spot area measuring device |
| CN109483887A (en) * | 2018-11-14 | 2019-03-19 | 华中科技大学 | Shaping layer contour accuracy online test method in the fusion process of selective laser |
| CN110497101A (en) * | 2019-08-26 | 2019-11-26 | 大连理工大学 | A kind of fiber laser cutting blade hole energy regulation method of metal sheet |
| CN110599413A (en) * | 2019-08-15 | 2019-12-20 | 江苏大学 | Laser spot image denoising method based on deep learning convolution neural network |
| CN110781775A (en) * | 2019-10-10 | 2020-02-11 | 武汉大学 | Remote sensing image water body information accurate segmentation method supported by multi-scale features |
| CN110940490A (en) * | 2019-12-13 | 2020-03-31 | 湖南省鹰眼在线电子科技有限公司 | Laser spot scanning precision detection method and device of laser processing equipment |
-
2020
- 2020-12-02 CN CN202011391635.0A patent/CN112435271B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1450304A1 (en) * | 2003-02-21 | 2004-08-25 | City University of Hong Kong | Image processing apparatus and method |
| CN101832941A (en) * | 2010-03-19 | 2010-09-15 | 天津大学 | Fruit quality evaluation device based on multispectral image |
| CN108955582A (en) * | 2018-06-21 | 2018-12-07 | 中国人民解放军战略支援部队航天工程大学 | Laser focal beam spot area measuring device |
| CN109483887A (en) * | 2018-11-14 | 2019-03-19 | 华中科技大学 | Shaping layer contour accuracy online test method in the fusion process of selective laser |
| CN110599413A (en) * | 2019-08-15 | 2019-12-20 | 江苏大学 | Laser spot image denoising method based on deep learning convolution neural network |
| CN110497101A (en) * | 2019-08-26 | 2019-11-26 | 大连理工大学 | A kind of fiber laser cutting blade hole energy regulation method of metal sheet |
| CN110781775A (en) * | 2019-10-10 | 2020-02-11 | 武汉大学 | Remote sensing image water body information accurate segmentation method supported by multi-scale features |
| CN110940490A (en) * | 2019-12-13 | 2020-03-31 | 湖南省鹰眼在线电子科技有限公司 | Laser spot scanning precision detection method and device of laser processing equipment |
Non-Patent Citations (8)
| Title |
|---|
| A. E. SIEGMAN等: "How to (Maybe) Measure Laser Beam Quality", 《DPSS (DIODE PUMPED SOLID STATE) LASERS》 * |
| ANDRII YAROVYI等: "An intelligent system of neural networking recognition of multicolor spot images of laser beam profile", 《PROCEEDINGS OF SPIE》 * |
| BILEL BENNADJI: "USING ENCODER-DECODER ARCHITECTURE FOR MATERIAL SEGMENTATION BASED ON BEAM PROFILE ANALYSIS", 《DEGREE PROGRAMME IN COMPUTER SCIENCE AND ENGINEERING》 * |
| 任凤雷等: "基于DeepLabV3+与超像素优化的语义分割", 《光学精密工程》 * |
| 李晓明等: "基于OTSU算法的激光光斑图像提取研究", 《激光杂志》 * |
| 李晓明等: "基于光斑轮廓特征的激光光束质量简易评价方法", 《中国激光》 * |
| 杨波等: "改进Deeplab_v3+网络的手术器械分割方法", 《计算机工程与应用》 * |
| 陈静等: "基于MATLAB的CO_2激光光斑质量鉴别技术", 《激光与红外》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112435271B (en) | 2023-06-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106949848B (en) | A kind of high-precision laser 3D profile phone structural detection method | |
| CN111751253B (en) | Forming method and quality detection method of concrete aggregate detection model | |
| CN103499303B (en) | A kind of wool fineness method for automatic measurement | |
| CN112883601B (en) | Surface defect detection method, device, equipment and storage medium | |
| CN115493532B (en) | Measuring system, method and medium for measuring area of element to be measured on surface of plate | |
| CN119354981A (en) | A visual inspection and quality assessment method for small aircraft structural parts | |
| CN112927185B (en) | A true stress-true strain curve test calculation method based on digital image correlation method | |
| CN111929310A (en) | Surface defect detection method, device, equipment and storage medium | |
| CN110110758A (en) | A kind of surface roughness classification method based on convolutional neural networks | |
| CN112816187B (en) | A method for judging the quality of laser spot | |
| CN112435271A (en) | Firing profile segmentation method applied to laser spot quality measurement | |
| CN115164753A (en) | A real-time non-contact strain measurement method and system based on standard digital speckle | |
| CN109946636A (en) | An automatic calibration method of digital universal representation value based on deep learning | |
| CN113989513A (en) | A method for identifying the reading of a square pointer meter | |
| CN119470473A (en) | A method and system for quality inspection of perforated workpieces | |
| CN108875124B (en) | Maximum value compensation algorithm for extracting peak value position of confocal axial response curve | |
| CN115451816A (en) | Three-dimensional scanning-based detection method for size of complex casting mold | |
| CN111210409A (en) | A Structural Damage Identification Method Based on Conditional Generative Adversarial Networks | |
| CN107144219A (en) | A kind of confocal laser measuring method | |
| CN101354357A (en) | A method for analyzing micro-density images of tree rings | |
| CN106780489A (en) | A kind of tree ring half-tone information analysis method and device | |
| CN114581390B (en) | Particle flow laser detection method, device and storage medium based on plasma image | |
| CN117036274B (en) | A defect detection method based on photoelectric detection | |
| Shammari et al. | A Study to Evaluate the Calibration of Optical Three-Dimensional Scanner | |
| CN117929633B (en) | On-line detection method and system for uniformity of color paste for photoresist |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20250519 Address after: 528400 Guangdong Province Zhongshan City Zhuangji Development Zone Chengdong Community Yushan Road No. 28 First Floor Area A Patentee after: Zhongshan Lanqi Technology Co., Ltd. Country or region after: China Address before: 528400 Zhongshan, Guangdong Shi Qu District, Xueyuan Road, No. 1 Patentee before: University OF ELECTRONIC SCIENCE AND TECHNOLOGY OF CHINA, ZHONGSHAN INSTITUTE Country or region before: China |
|
| TR01 | Transfer of patent right |