CN211825761U - Rolled glass defect detection device - Google Patents
Rolled glass defect detection device Download PDFInfo
- Publication number
- CN211825761U CN211825761U CN201921987121.4U CN201921987121U CN211825761U CN 211825761 U CN211825761 U CN 211825761U CN 201921987121 U CN201921987121 U CN 201921987121U CN 211825761 U CN211825761 U CN 211825761U
- Authority
- CN
- China
- Prior art keywords
- light source
- source group
- rolled glass
- assembly
- processor
- 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.)
- Active
Links
Images
Abstract
The utility model relates to a calendering glass defect detecting device. A rolled glass defect detection device comprises a light source assembly, an imaging assembly, a processor and a data terminal; the light source assembly and the imaging assembly are respectively connected with the processor; the processor is connected with the data terminal; the light source component comprises a first light source group and a second light source group, and the first light source group and the second light source group are oppositely arranged and are respectively used for irradiating one of two surfaces of the rolled glass; the imaging assembly is arranged on one side of the light source assembly and used for collecting image data and transmitting the image data to the processor. The device uses a multidimensional space light source technology, so that the problem of photovoltaic rolled glass rolling is solved optically, the software workload is greatly reduced, the problem of easy false detection of a software algorithm is solved, and the efficiency is greatly improved.
Description
Technical Field
The utility model relates to a glass defect field especially relates to a calendering glass defect detecting device.
Background
The rolled glass comprises embossed glass, corrugated glass, wired glass, polished glass blanks, groove-shaped glass and the like, is applied to solar power generation, can absorb the radiant heat of solar energy to the maximum extent and greatly improve the photoelectric conversion efficiency of a solar cell. The glass can resist the radiation of solar ultraviolet rays, the light transmittance is not reduced, the service life of the cover plate glass can be effectively prolonged, and the alkali resistance, the mildew resistance and the ageing resistance of the cover plate glass in various environments are enhanced compared with those of common flat super-white glass, so that the cover plate glass of the solar cell module is more favorable for protecting a silicon crystal plate below the cover plate glass from being damaged by the outside with the strength of the cover plate glass and is durable. It has the advantages of beautiful appearance, controllable light transmission, energy saving and power generation, no need of fuel, no waste gas, no waste heat, no waste residue and no noise pollution. 76% of China's homeland illumination is abundant, and the light energy resource distribution is relatively uniform; compared with hydroelectric power, wind power, nuclear power and the like, the solar power generation has no emission and noise, and has mature application technology, safety and reliability. The large-scale start of use of solar technology, photovoltaic rolled glass has held a significant position.
Defects can be identified through a large amount of manpower observation in the previous production process. In order to save labor and improve efficiency, the detection equipment is used for detecting defects generated by materials and identifying defective products. However, the special feature of carrying patterns on the transparent rolled glass makes the current detection equipment not ideal in effect.
The existing detection method mainly adopts a software algorithm as a main part, images the photovoltaic through a common optical system to form an embossed image, analyzes the embossed image to find some characteristics of embossing, and can filter and emboss from the image through complex algorithm processing to realize the detection of defects.
Patent document No. 201520028343.9 discloses a solar photovoltaic glass defect detection device, which uses a common light source and a glass support, and then uses a camera to take pictures, although some defects can be detected, but many false detections and missed detections exist when the device faces the embossing which is laminated, and the detection rate and the missed detection rate are unsatisfactory. Patent document No. 201711481653.6 discloses a method and apparatus for classifying defects of photovoltaic glass based on convolutional neural network and patent document No. 201710982581.7 discloses a method and apparatus for detecting defects of photovoltaic glass, both of which mainly use software algorithm to solve the effect of embossing, and although some defects can be detected, there is also a problem of false detection due to embossing. And because the algorithm content is too much, only depending on the data terminal to calculate, there will be a large amount of time consumption, will make the detection speed slow, because the long-term high load operation of the data terminal, stability is greatly discounted at the same time.
In summary, in the visual inspection of the defects of the rolled glass, the obtained image is mainly processed by using a software algorithm to eliminate the influence of the rolled image on the defect inspection, so that the efficiency is low, the accuracy is poor, and the defect inspection of the rolled glass has great defects, and research, development and innovation on the technical problem are urgently needed.
SUMMERY OF THE UTILITY MODEL
In view of above-mentioned prior art's weak point, the utility model aims to provide a calendering glass defect detection device mainly solves the problem that calendering glass itself calendering and defect are difficult to distinguish in visual detection equipment for the calendering decorative pattern weakens and even disappears in the work piece image of collection, and the detection range covers the calendering glass and other calendering materials of different models and types, improves detection speed and detection precision, reduces the false retrieval.
In order to achieve the purpose, the utility model adopts the following technical proposal:
a rolled glass defect detection device comprises a light source assembly, an imaging assembly, a processor and a data terminal;
the light source assembly and the imaging assembly are respectively connected with the processor; the processor is connected with the data terminal;
the light source component comprises a first light source group and a second light source group, and the first light source group and the second light source group are oppositely arranged and are respectively used for irradiating one of two surfaces of the rolled glass;
the imaging assembly is arranged on one side of the light source assembly and used for collecting image data and transmitting the image data to the processor.
Preferably, the first light source group and the second light source group are respectively provided with a plurality of illuminating lamps which are arranged in a semicircular shape in bilateral symmetry.
In the rolled glass defect detection device, the number of the illuminating lamps in the first light source group is 10-20.
In the rolled glass defect detecting apparatus, the number of the illuminating lamps in the second light source group is preferably the single number of the illuminating lamps in the range of 11 to 19.
Preferably, the rolled glass defect detection device further comprises a cooling device for cooling the light source assembly.
Preferably, the rolled glass defect detection device is an industrial water cooler.
Preferably, the rolled glass defect detection device comprises an imaging assembly and a control assembly, wherein the imaging assembly comprises a linear array camera and a high-resolution low-distortion lens; the scanning position of the lens on the rolled glass is a detection position, and the detection position is in the range of 0-30 degrees of left-right inclination of the lens.
Compared with the prior art, the utility model provides a calendering glass defect detection device, this device uses multidimensional space light source technique for solve photovoltaic calendering glass calendering optically, has greatly reduced software work load, has solved the problem that software algorithm is easy to be misexamined again, has very big promotion in efficiency; meanwhile, the optimally designed program-controlled multi-dimensional multi-light-field light source is matched, so that the image acquisition of the same defect in different illumination environments is realized, the acquisition quantity of the information of the defect is greatly improved, and the classification and identification effects of the defect are obviously improved; moreover, the calibration method can be suitable for all different types of rolled photovoltaic glass and other rolled materials, and the detection range is wider.
Drawings
FIG. 1 is a block diagram of the rolled glass defect detecting apparatus provided by the present invention;
fig. 2 is a schematic view of a light source module and an imaging module according to the present invention;
fig. 3 is a flowchart of the rolled glass defect detection method provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and effects of the present invention clearer and clearer, the following description of the present invention will refer to the accompanying drawings and illustrate embodiments. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the invention.
Referring to fig. 1-3, wherein the arrow in fig. 2 is the moving direction of the detecting element, the present invention provides a rolled glass defect detecting device, which is characterized in that it comprises a light source assembly 1, an imaging assembly 2, a processor 3 and a data terminal 4;
the light source assembly 1 and the imaging assembly 2 are respectively connected with the processor 3; the processor 3 is connected with the data terminal 4;
the light source assembly 1 comprises a first light source group 11 and a second light source group 12, wherein the first light source group 11 and the second light source group 12 are oppositely arranged and are respectively used for irradiating one of two surfaces of rolled glass;
the imaging assembly 2 is arranged at one side of the light source assembly 1 and used for collecting image data and transmitting the image data to the processor 3.
Correspondingly, the utility model provides a be used for detection device's calendering glass defect detecting method, a serial communication port, including the step:
s1, debugging a light source: using a piece of pure standard glass as a calibration sample plate, controlling the light source assembly 1 to debug a light source, enabling calendering lines in an obtained image of the calibration sample plate to disappear, and determining configuration parameters of the light source assembly 1; at this point, the configuration parameters are stored in the processor 3 for use in detection;
s2, detection: the light source assembly 1 executes the standard of the configuration parameters to illuminate the detection piece, and the imaging assembly 2 acquires the image data of the detection piece and sends the image data to the processor 3 for preprocessing;
s3, analyzing defects: and the data terminal 4 receives the preprocessed image data, and performs defect identification and classified output by using a deep learning algorithm.
Specifically, in use, firstly, a standard blank rolled glass in the same batch or model to be detected at this time is used as a calibration sample plate to determine configuration parameters of the first light source group 11 and the second light source group 12 in the light source assembly 1, so that patterns on the rolled glass are irradiated by the first light source and the second light source, the images of the imaging assembly 2 are captured without displaying the patterns, the working configuration parameters of the first light source group 11 and the second light source group 12 are recorded, and in the process of detecting the batch or model of rolled glass, the processor 3 controls the light source assembly 1 to detect the detection element according to the configuration parameters. The imaging assembly 2 collects image data of a detection piece, and the image data is preprocessed by the processor 3 and then sent to the data terminal 4 for defect identification and classification output. If yes, an alarm prompt is sent out when the defect is detected, wherein the alarm prompt comprises a display for displaying specific detection conditions, an indicator light or a sound prompt.
It should be noted that, the first light source group 11 and the second light source group 12 are light source groups, in general, the first light source group 11 and the second light source group 12 can emit light with different brightness and angles according to the control of the processor 3, the first light source group 11 and the second light source group 12 are installed oppositely and respectively irradiate one surface of the rolled glass, the brightness and the angle of the illumination can be adjusted according to the control of the processor 3, and the effect that the pattern of the rolled glass with different types or models can disappear in the image of the area collected by the imaging component 2 can be realized.
The pretreatment specifically comprises the following steps: firstly, Gaussian filtering is carried out on an acquired image, then contrast enhancement is carried out on the image, flat field correction is carried out on the enhanced image, then an image segmentation method based on Kalman filtering is adopted to separate a defect image from a background image, a defect image block is extracted, and the defect image block is analyzed and characterized to form defect image data.
The convolutional neural network is a feed-forward neural network with a deep structure, is a deep learning algorithm, can perform translation invariant classification on input information according to a hierarchical structure, can be used for constructing a hierarchical classifier for the problem of defect classification in an image, and can also be used for extracting the distinguishing characteristics of the image in fine classification and identification so as to be used for learning of other classifiers. For the latter, the feature extraction is that different parts of the image are respectively input into a convolutional neural network, and the convolutional neural network extracts by self through unsupervised learning.
After learning to extract the characteristics of the defects according to different types, in the detection process, the acquired image is input into the convolutional neural network, whether the input image contains the corresponding defects or not is judged, and classification identification is carried out.
Preferably, in this embodiment, the first light source group 11 and the second light source group 12 respectively have a plurality of illuminating lamps, and are respectively installed in a semicircular shape with bilateral symmetry. The illuminating lamps are LED lamps with the same type; the left and right pairs are bilaterally symmetrical by using a semicircular central line, light emitted by all the illuminating lamps is concentrated to one position, and an angle formed by the illuminating directions of two adjacent illuminating lamps is fixed, namely a plurality of illuminating lamps are uniformly arranged on each position of the semicircle; the central line position of the first light source group 11 is not provided with the illuminating lamp, and the imaging assembly 2 is used for acquiring images through the central line position; and an illuminating lamp with the irradiating light vertical to the plane where the rolled glass is located is arranged at the central line position of the second light source group 12.
Preferably, in this embodiment, the number of the illuminating lamps in the first light source group 11 is two of 10 to 20.
Preferably, in this embodiment, the number of the illuminating lamps in the second light source group 12 is a single number of the illuminating lamps 11 to 19.
Specifically, the first light source group 11 and the imaging component 2 are arranged on the same side, preferably 14 illuminating lamps are provided, and the illuminating lamps are divided into two sides and are not measured by 7; the second light source group 12 preferably has 15 illuminating lamps. More preferably, the number of the illuminating lamps of the second light source group 12 is equal to the number of the illuminating lamps of the first light source group 11 plus one.
Preferably, in this embodiment, the light source module further includes a cooling device for cooling the light source module 1.
Preferably, in this embodiment, the cooling device is an industrial water cooler, and is configured to operate continuously for 24 hours to provide cooling for the light source assembly 1.
As a preferred solution, in this embodiment, the imaging assembly 2 includes a line camera and a high-resolution low-distortion lens; the scanning position of the lens on the rolled glass is a detection position, and the detection position is in the range of 0-30 degrees of left-right inclination of the lens. Specifically, the lens is optimally arranged in a vertical mode, namely, the inclination angle is optimally 0 degrees.
Preferably, in this embodiment, in step S1, the specific operation of debugging the light source is:
s11, a piece of flawless standard glass is used as a calibration template, the light source brightness of the light source assembly 1 is initial brightness, the imaging assembly 2 collects image data of the calibration template, the image data is processed by the processor 3 and then transmitted to the data terminal 4 for recognition, if the image data has calendering lines, the step S12 is executed, otherwise, the step S14 is executed;
s12, adjusting configuration parameters of the light source assembly 1, changing the brightness or the incident angle of an illuminating lamp, and acquiring image data of the calibration sample plate by the imaging assembly 2;
s13, the image data is processed by the processor 3 and then transmitted to the data terminal 4 for recognition, if the image data has the calendering texture, the step S12 is executed, otherwise, the step S14 is executed;
s14, the processor 3 records and stores the configuration parameters of the light source assembly 1.
Specifically, in the process of adjusting the light source assembly 1, a part of the illuminating lamps, for example, two illuminating lamps, are turned on by default, and then, step S12 is executed once, so that one illuminating lamp is turned on until the calendering pattern in the calendered glass disappears in the image acquired by the imaging assembly 2; in the foregoing commissioning, the initial configuration parameter may be to turn on the illumination lamp at a certain angle, and then turn on the illumination lamps at other angles each time step S12 is executed.
Preferably, in this embodiment, before the step S3, the method further includes the step of:
s3', deep learning training is performed using the defect images of various rolled glasses as samples.
It should be understood that equivalent alterations and modifications can be made by those skilled in the art according to the technical solution of the present invention and the inventive concept thereof, and all such alterations and modifications should fall within the scope of the appended claims.
Claims (7)
1. The rolled glass defect detection device is characterized by comprising a light source assembly, an imaging assembly, a processor and a data terminal;
the light source assembly and the imaging assembly are respectively connected with the processor; the processor is connected with the data terminal;
the light source component comprises a first light source group and a second light source group, and the first light source group and the second light source group are oppositely arranged and are respectively used for irradiating one of two surfaces of the rolled glass;
the imaging assembly is arranged on one side of the light source assembly and used for collecting image data and transmitting the image data to the processor.
2. The rolled glass defect detecting device according to claim 1, wherein the first light source group and the second light source group have a plurality of illuminating lamps, respectively, which are installed in a semicircular shape symmetrical to each other.
3. The rolled glass defect detecting apparatus according to claim 2, wherein the number of the illuminating lamps in the first light source group is a double number of 10 to 20.
4. The rolled glass defect detecting apparatus as claimed in claim 2, wherein the number of illuminating lamps in the second light source group is a single number of 11 to 19.
5. The rolled glass defect detecting apparatus according to any one of claims 1 to 4, further comprising a cooling device for cooling the light source assembly.
6. The rolled glass defect detection apparatus of claim 5, wherein the cooling apparatus is an industrial water cooler.
7. The rolled glass defect detection apparatus of claim 1, wherein the imaging assembly comprises a line camera and a high resolution low distortion lens; the scanning position of the lens on the rolled glass is a detection position, and the detection position is in the range of 0-30 degrees of left-right inclination of the lens.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201921987121.4U CN211825761U (en) | 2019-11-15 | 2019-11-15 | Rolled glass defect detection device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201921987121.4U CN211825761U (en) | 2019-11-15 | 2019-11-15 | Rolled glass defect detection device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN211825761U true CN211825761U (en) | 2020-10-30 |
Family
ID=73027914
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201921987121.4U Active CN211825761U (en) | 2019-11-15 | 2019-11-15 | Rolled glass defect detection device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN211825761U (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110702699A (en) * | 2019-11-15 | 2020-01-17 | 湖南讯目科技有限公司 | Rolled glass defect detection device and method |
-
2019
- 2019-11-15 CN CN201921987121.4U patent/CN211825761U/en active Active
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110702699A (en) * | 2019-11-15 | 2020-01-17 | 湖南讯目科技有限公司 | Rolled glass defect detection device and method |
| CN110702699B (en) * | 2019-11-15 | 2024-03-08 | 湖南讯目科技有限公司 | Rolled glass defect detection device and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2023134792A2 (en) | Led lamp wick defect detection method | |
| CN109084957B (en) | Defect detection and color sorting method and system for photovoltaic solar crystalline silicon cell | |
| CN110702699B (en) | Rolled glass defect detection device and method | |
| CN108332689B (en) | Optical measurement system and method for detecting surface roughness and surface damage | |
| CN104574389A (en) | Battery piece chromatism selection control method based on color machine vision | |
| CN103645573A (en) | Liquid crystal display (LCD) detection method and system based on machine vision | |
| CN102175692A (en) | System and method for detecting defects of fabric gray cloth quickly | |
| CN106709529B (en) | Visual detection method for photovoltaic cell color difference classification | |
| Yang et al. | An automatic aperture detection system for LED cup based on machine vision | |
| CN104048966B (en) | The detection of a kind of fabric defect based on big law and sorting technique | |
| CN106780464A (en) | A kind of fabric defect detection method based on improvement Threshold segmentation | |
| CN211825761U (en) | Rolled glass defect detection device | |
| CN113034482A (en) | Surface roughness detection method based on machine vision and machine learning | |
| CN115115643A (en) | Method for detecting production defects of photovoltaic cell panel | |
| CN203908956U (en) | System for detecting surface defects of workpiece | |
| CN110873718A (en) | Steel plate surface defect detection system and method based on machine vision | |
| CN103091330A (en) | Copper strip surface defect identifying device for simulating human visual perception mechanism and method thereof | |
| CN106918596A (en) | One kind is based on visible ray or ultrared ore machine device for visual identification and its recognition methods | |
| CN112304876A (en) | Cotton color degree detection device and detection method based on machine vision | |
| CN201345056Y (en) | Rubber production online monitoring device | |
| CN111169861A (en) | Automatic household garbage identification and classification equipment | |
| CN109870460A (en) | A kind of composite material battery case surfaces quality determining method based on machine vision | |
| CN109916923A (en) | A kind of customization plate automatic defect detection method based on machine vision | |
| CN113430679B (en) | Production platform for identifying abnormality of carbon fiber in pre-oxidation furnace | |
| CN110610474A (en) | Solar panel defect real-time detection method based on infrared image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant |