CN203083937U - Machine vision-based detection system of U-shaped powder tube - Google Patents
Machine vision-based detection system of U-shaped powder tube Download PDFInfo
- Publication number
- CN203083937U CN203083937U CN 201320021624 CN201320021624U CN203083937U CN 203083937 U CN203083937 U CN 203083937U CN 201320021624 CN201320021624 CN 201320021624 CN 201320021624 U CN201320021624 U CN 201320021624U CN 203083937 U CN203083937 U CN 203083937U
- Authority
- CN
- China
- Prior art keywords
- module
- detection
- detection system
- tube cell
- processing unit
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The utility model discloses a machine vision-based detection system of a U-shaped powder tube. The machine vision-based detection system comprises a detection platform unit, an image acquiring unit and an image processing unit, wherein the image processing unit comprises an ROI (Region Of Interest) extracting module, a characteristic extracting module, a data storing module, a distance calculating module, a quality measuring module and a human-machine interface module. According to the machine-vision-based detection system, high-speed, lossless and qualitative detection on the coating quality of a powder tube of a U-shaped energy-saving fluorescent tube can be realized, and the detection speed is 6 powder tubes per minute; through matching of a photoelectric sensor, by using a reasonably arranged backlight source and a high-speed industrial camera, an image of the powder tube of the fluorescent tube can be effectively acquired; and defect spots in products can be effectively detected through location of key points and detection of a sliding window, and thus the product detection efficiency is increased, and the requirement of lamp tube detection during actual industrial production can be met.
Description
Technical field
The utility model belongs to the detection technique field, is specifically related to a kind of detection method and detection system thereof of the U type tube cell based on machine vision.
Background technology
U type tube cell also is the product in early stage of energy-conservation fluorescent tube as the important component part of U type electricity-saving lamp simultaneously, is that spraying forms through phosphor slurry by U type glass tube.Can the luminous mass of power saving fluorescent lamps depends primarily on the even effect of dusting of tube inner wall three color base powder, thereby realize effective detection of tube cell dusting effect most important.
Tube cell needs to measure the dusting effect rapidly and accurately after coating process machines, and needs to reject rapidly underproof tube cell, thereby avoids entering next procedure, avoids the waste of processing.
In the process of power saving fluorescent lamps, effective detection of the dusting effect of tube cell is to realize the crucial mass parameter of control automatically.Yet because factors such as technologies, the automatic on-line of tube cell detects a technical barrier that becomes in the fluorescent tube production.But the quality detecting system that does not still have the fluorescent tube of robotization at present both at home and abroad, under the existing technology, the general employing of the detection of tube cell is the tube cell dusting effect detection of carrying out backlight by naked eyes with daylight lamp manually, work under bad environment, work uninteresting, inefficiency, and, manually be difficult to carry out qualitative detection for trickle flaw, and detection time is longer, but whether accurate operating experience and the fatigue state that depends on the workman that it estimates; So general accuracy of detection is not high, the reliability of testing result is low, can not satisfy the needs that online in real time detects.
The patent No. be 5408104 U.S. Patent Publication a kind of annulus fluorescent pipe flaw detection method and pick-up unit thereof based on linear CCD, this method is obtained the surface image of annular lamp tube by a plurality of cameras, by the method filtering noise of threshold value is set, utilize the directivity of image slices vegetarian refreshments to come the surface blemish of annular lamp tube is detected.The glass of annular lamp tube outside surface scratches, slight crack but this detection method and pick-up unit only can detect, and can't detect the fluorescent powder jet printing effect of fluorescent tube inwall.
Summary of the invention
At the above-mentioned technological deficiency of existing in prior technology, the utility model provides a kind of detection method and detection system thereof of the U type tube cell based on machine vision, can realize accurately measuring in real time of U type tube cell dusting effect.
A kind of detection method of the U type tube cell based on machine vision comprises the steps:
(1) the tube cell image of collection U type tube cell to be measured correspondence under each visual angle;
(2) from described tube cell image, extract ROI (Region of Interest, area-of-interest);
(3) described ROI is carried out texture feature extraction, obtain LBP (Local BinaryPatterns, the local binary pattern) histogram of ROI, and make up the LBP proper vector according to the LBP histogram;
(4) corresponding LBP standard feature vector carries out Chi-square (card side) distance calculation in the LBP proper vector that makes ROI and the database, obtains the Chi-square distance value;
(5) travel through all tube cell images according to step (2) to (4), obtain the Chi-square distance value of every tube cell image correspondence; According to these Chi-square distance values, judge whether U type tube cell dusting effect to be measured is qualified.
In the described step (2), the ROI method that extracts from the tube cell image is: at first according to the profile of U type tube cell to be measured, adopt some key point that tube cell image boundary curve is estimated; Utilize geometric formula to reconstruct the border of U type tube cell to be measured according to the regular shape of donut and two rectangular tube legs then, and then from the tube cell image, extract ROI.
In the described step (3), the method that makes up the LBP proper vector according to the LBP histogram is: the number of pixels that makes 10 kinds of pattern correspondences in the LBP histogram obtains the LBP proper vector respectively as 10 element values of LBP proper vector thereby make up.
Described LBP standard feature vector is asked for by the following method: obtain the qualified U type tube cell of several dusting effects by manual detection, calculate the pairing LBP proper vector of each visual angle tube cell image of each U type tube cell according to step (1) to (3); For arbitrary visual angle, the LBP proper vector of all tube cell images of belonging to this visual angle to be asked on average, the averaged feature vector that obtains is the LBP standard feature vector of this visual angle tube cell image.
In the described step (4), make the LBP proper vector carry out the Chi-square distance calculation with corresponding LBP standard feature vector according to following formula:
Wherein: D is the LBP proper vector and the Chi-square distance value of corresponding LBP standard feature vector, H
2(i) be i element value in the LBP proper vector, H
1(i) be i element value in the LBP standard feature vector, i is natural number and 1≤i≤10.
In the described step (5), judge whether qualified standard is U type tube cell dusting effect to be measured: each Chi-square distance value is all compared with given distance threshold, if all Chi-square distance values all less than described distance threshold, judge that then U type tube cell dusting effect to be measured is qualified; Otherwise, judge that then U type tube cell dusting effect to be measured is defective.
A kind of detection system of the U type tube cell based on machine vision comprises:
The detection platform unit is used to place U type tube cell to be measured;
Image acquisition units is used to gather the tube cell image of U type tube cell to be measured;
Graphics processing unit is used for extracting ROI from the tube cell image, and ROI is carried out texture feature extraction and quality assessment, and generates evaluation result.
Described graphics processing unit comprises: ROI extraction module, characteristic extracting module, data memory module, distance calculation module and quality assessment module, the ROI extraction module links to each other with image acquisition units, characteristic extracting module links to each other with distance calculation module with the ROI extraction module, data memory module links to each other with distance calculation module, and distance calculation module links to each other with the quality assessment module;
The ROI extraction module is used for extracting ROI from described tube cell image;
Characteristic extracting module is used for ROI is carried out texture feature extraction, obtains the LBP histogram, and then makes up corresponding LBP proper vector;
Data memory module is used to store LBP standard feature vector;
Distance calculation module is used to calculate the Chi-square distance value of the corresponding LBP standard feature with it of LBP proper vector vector;
The quality assessment module is used for according to described Chi-square distance value, judges whether U type tube cell dusting effect to be measured is qualified.
Described graphics processing unit also comprises human-computer interface module, and it is used to show the tube cell image, and reception user's instruction is to carry out the parameter setting to other modules in the graphics processing unit; Human-computer interface module links to each other with the quality assessment module with image acquisition units.
Described image acquisition units adopts video camera, and described graphics processing unit adopts computing machine.
Preferably, described detection platform unit comprises monitor station and light source, and monitor station is provided with the motor rotating disk, is respectively equipped with four photoelectric sensors around the motor rotating disk, motor rotating disk sidepiece is provided with the triggering bar of a sensing photoelectric sensor, and described photoelectric sensor links to each other with image acquisition units; Can be implemented under a plurality of visual angles U type tube cell is carried out image acquisition.
Further preferably, described light source is connected with light source controller, and described light source controller all is connected graphics processing unit with photoelectric sensor; The luminous intensity that graphics processing unit can be regulated light source by light source controller according to the sensor-triggered state, thus best imaging effect obtained.
The utility model can realize that detection speed can reach 6/second to the high speed of the energy-conservation fluorescent tube tube cell of U type dusting quality, harmless, qualitative detection; Its detection system is by under the cooperation of photoelectric sensor, utilize the backlight rationally placed and high-speed industrial mutually function effectively gather fluorescent tube tube cell image; Its detection method utilizes key point location and moving window to detect the flaw point that can effectively detect in the product, thereby has improved the product detection efficiency, and the fluorescent tube that can satisfy in the actual industrial production detects requirement.
Description of drawings
Fig. 1 is the structural representation of the utility model detection system.
Fig. 2 is the synoptic diagram of preceding 9 kinds of patterns in the LBP histogram.
Embodiment
In order more specifically to describe the utility model, the utility model detection system and detection method thereof are elaborated below in conjunction with the drawings and the specific embodiments.
As shown in Figure 1, a kind of detection system of the U type tube cell based on machine vision comprises: detection platform unit, image acquisition units, graphics processing unit and human and machine interface unit;
The detection platform unit comprises monitor station 1 and light source 2, monitor station 1 is provided with motor rotating disk 4, be respectively equipped with four photoelectric sensors 5 around the motor rotating disk 4, motor rotating disk 4 sidepieces are provided with the triggering bar 6 of a sensing photoelectric sensor 5, and photoelectric sensor 5 links to each other with image acquisition units; Light source 2 is connected with light source controller, and light source controller all is connected graphics processing unit with photoelectric sensor 5.
Image acquisition units is used to gather the tube cell image of U type tube cell to be measured, and links to each other with four photoelectric sensors; In the present embodiment, image acquisition units adopts the CG400 black and white 1/3CMOS video camera of company of Daheng, resolution be 768 * 480 and frame per second adjustable, the highest frame per second can reach for 60 frame/seconds, this video camera adopts full frame scan mode line by line, and output interface is the USB mouth, and the camera lens bayonet socket is the C/CS mouth, volume is small and exquisite, is easy to install.Camera lens uses the undistorted camera lens of 12mm high resolving power of Japanese COMPUTAR.
In the present embodiment, light source adopts the blue backlight of the bright 200mm * 200mm of company of latitude, and light source controller uses the latitude RS232 of bright company standard serial ports control adjustable brightness 24V controller, and photoelectric sensor uses the positive-negative-positive photoelectric sensor of the 24V of Omron Corp.
During detection, U type tube cell 3 to be measured is reverse U shape is placed on the motor rotating disk 4, light source 2 is located at monitor station 1 one sides, video camera is located at the opposite side of relative light source 2, by the rotation of control motor, trigger the bar 6 synchronous photoelectric sensors 5 that rotate and will trigger successively on the position, four directions.Sensor-triggered represents that fluorescent tube has rotated to the given photograph position, thereby the photoelectric sensor trigger pip directly triggers the video camera candid photograph, simultaneously photoelectric sensor output level state is delivered to image acquisition units by data collecting card, the luminous intensity that image acquisition units is regulated light source 4 according to the sensor-triggered state by light source controller, thus best imaging effect obtained.
Collect the tube cell image of 4 sides after motor rotates a circle, fluorescent tube is moved to the collection that station carries out the top graph picture; During the top image acquisition, two light sources 2 are 120 ° of angles are arranged on the monitor station 1, upwards U type tube cell 3 to be measured is throwed, video camera is located at directly over the U type tube cell 3 to be measured, collect a top the tube cell image.
Graphics processing unit is used for extracting ROI from the tube cell image, and ROI is carried out texture feature extraction and quality assessment, and generates evaluation result; In the present embodiment, graphics processing unit adopts and grinds magnificent IPC Series Industrial control computer, and this machine adopts Intel dual core processor, dominant frequency 3.0G, 1100M network interface card, 1G internal memory, 160G hard disk; Computing machine is connected with video camera by USB, be connected with photoelectric sensor by the pci bus data collecting card, be connected with light source controller by the RS232 universal serial bus, data collecting card selects for use Taiwan to grind the PCI-1730 integrated circuit board of magnificent company, 16 road digital quantity I/O passages.
Comprise ROI extraction module, characteristic extracting module, data memory module, distance calculation module, quality assessment module and human-computer interface module in the computing machine; The ROI extraction module all links to each other with video camera with human-computer interface module, characteristic extracting module links to each other with distance calculation module with the ROI extraction module, data memory module links to each other with distance calculation module, and distance calculation module links to each other with the quality assessment module, and the quality assessment module links to each other with human-computer interface module; Wherein:
The ROI extraction module is used for extracting ROI from the tube cell image, and specific implementation process is as follows:
At first, the tube cell image of gathering is carried out denoising, obtain background and be white, fluorescent tube zone is the gray level image of grey, placement level, the vertical extent of image is detected also self-adaptation adjustment, with the slight inclination phenomenon that prevents to cause because of mechanical motion; Then, according to the profile of U type tube cell to be measured, adopt some key point that tube cell image boundary curve is estimated; At last, utilize geometric formula to reconstruct the border of U type tube cell to be measured according to the regular shape of donut and two rectangular tube legs, and then from original image, extract ROI.
Characteristic extracting module is used for ROI is carried out texture feature extraction, obtains the LBP histogram, and then makes up corresponding LBP proper vector.Owing to reasons such as coating process, the fluorescent tube image appearance is that overall intensity is inhomogeneous, has the piecemeal phenomenon, and the variation of gray-scale value is continuous gradual change, thereby can not handle according to the intensity profile rule of integral body, must handle according to the gray-scale value of regional area.Thereby the regional area gray-scale value that present embodiment adopts the method for moving window to extract different parts on the fluorescent tube carries out the homogeneity detection.
The LBP histogram is used for statistical picture local grain feature, has gray scale and rotational invariance.For the arbitrary pixel in ROI not, in the neighborhood 8 pixels are arranged all around it, their gray-scale value is made as g successively
1g
2g
3g
4g
5g
6g
7g
8, if g
iGray-scale value than center pixel is big, is labeled as 1, on the contrary mark 0, with g
1To g
8String together and form one eight place value G (as 10110001).In fact, for the transition times of G value the inside 0 and 1 smaller or equal to 2 (0 or 2, saltus step can not appear 1 time) situation, can be divided into 9 kinds of pattern (solid expressions 0 as shown in Figure 2, hollow expression 1), these 9 kinds of patterns are called unified pattern, and to be classified as the 10th kind of pattern be non-unified pattern (as 10110001, saltus step four times) and 0 and 1 saltus step surpasses twice G value; At last, obtain the distribution of each pixel of ROI on these 10 kinds of patterns with statistics with histogram.
The number of pixels that makes 10 kinds of pattern correspondences in the LBP histogram obtains corresponding LBP proper vector respectively as 10 element values of LBP proper vector thereby make up.
Data memory module is used to store LBP standard feature vector, LBP standard feature vector is asked for by the following method: obtain the qualified U type tube cell of several dusting effects by manual detection, calculate each each visual angle of U type tube cell (four sides and an end face) pairing LBP proper vector of tube cell image by image acquisition, ROI extraction, texture feature extraction then; For arbitrary visual angle, the LBP proper vector of all tube cell images of belonging to this visual angle to be asked on average, the averaged feature vector that obtains is the LBP standard feature vector of this visual angle tube cell image.
Distance calculation module is used for calculating according to following formula the Chi-square distance value of the corresponding LBP standard feature with it of LBP proper vector vector;
Wherein: D is the LBP proper vector and the Chi-square distance value of corresponding LBP standard feature vector, H
2(i) be i element value in the LBP proper vector, H
1(i) be i element value in the LBP standard feature vector, i is natural number and 1≤i≤10.
The quality assessment module is used for the distance value according to Chi-square, judges whether U type tube cell dusting effect to be measured is qualified; Its concrete judgment criteria is: the Chi-square distance value that makes each visual angle tube cell image correspondence of U type tube cell to be measured all compares with given distance threshold, if all Chi-square distance values all less than described distance threshold, judge that then U type tube cell dusting effect to be measured is qualified; Otherwise, judge that then U type tube cell dusting effect to be measured is defective.
Human-computer interface module is used to show the tube cell image, and reception user's instruction is to carry out the parameter setting to other modules in the graphics processing unit; It adopts the LCD touching display screen.
Present embodiment can realize 4 secondary side images to single fluorescent tube, the collection and the detection processing of 1 secondary top graph picture in 100 milliseconds, the image that collects has been eliminated the harmful effect that causes such as reflective, crooked, can realize real-time, the quick Defect Detection to U type fluorescent tube, detection speed can reach 6/second.
Claims (4)
1. the detection system based on the U type tube cell of machine vision comprises: detection platform unit, image acquisition units and graphics processing unit; It is characterized in that: described graphics processing unit comprises ROI extraction module, characteristic extracting module, data memory module, distance calculation module and quality assessment module, the ROI extraction module links to each other with image acquisition units, characteristic extracting module links to each other with distance calculation module with the ROI extraction module, data memory module links to each other with distance calculation module, and distance calculation module links to each other with the quality assessment module.
2. detection system according to claim 1 is characterized in that: described graphics processing unit also comprises human-computer interface module, and described human-computer interface module links to each other with the quality assessment module with image acquisition units.
3. detection system according to claim 1, it is characterized in that: described detection platform unit comprises monitor station and light source, monitor station is provided with the motor rotating disk, be respectively equipped with four photoelectric sensors around the motor rotating disk, motor rotating disk sidepiece is provided with the triggering bar of a sensing photoelectric sensor, and described photoelectric sensor links to each other with image acquisition units.
4. detection system according to claim 3 is characterized in that: described light source is connected with light source controller, and described light source controller all is connected graphics processing unit with photoelectric sensor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201320021624 CN203083937U (en) | 2013-01-16 | 2013-01-16 | Machine vision-based detection system of U-shaped powder tube |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201320021624 CN203083937U (en) | 2013-01-16 | 2013-01-16 | Machine vision-based detection system of U-shaped powder tube |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN203083937U true CN203083937U (en) | 2013-07-24 |
Family
ID=48829762
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201320021624 Expired - Fee Related CN203083937U (en) | 2013-01-16 | 2013-01-16 | Machine vision-based detection system of U-shaped powder tube |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN203083937U (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106501278A (en) * | 2016-11-08 | 2017-03-15 | 浙江科技学院 | Surface of the light tube defect classification method and system based on invariable rotary textural characteristics |
| CN114155203A (en) * | 2021-11-11 | 2022-03-08 | 深圳职业技术学院 | Image processing method and device suitable for mold monitor |
-
2013
- 2013-01-16 CN CN 201320021624 patent/CN203083937U/en not_active Expired - Fee Related
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106501278A (en) * | 2016-11-08 | 2017-03-15 | 浙江科技学院 | Surface of the light tube defect classification method and system based on invariable rotary textural characteristics |
| CN106501278B (en) * | 2016-11-08 | 2019-06-07 | 浙江科技学院 | Surface of the light tube defect classification method and system based on invariable rotary textural characteristics |
| CN114155203A (en) * | 2021-11-11 | 2022-03-08 | 深圳职业技术学院 | Image processing method and device suitable for mold monitor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110070524A (en) | A kind of intelligent terminal panel visual fault detection system | |
| CN107664644B (en) | Object appearance automatic detection device and method based on machine vision | |
| CN109916910B (en) | Photovoltaic glass edge defect detection system and corresponding method | |
| CN109900711A (en) | Workpiece, defect detection method based on machine vision | |
| CN110044910B (en) | Automobile cup box part detection system and detection method | |
| CN106248686A (en) | Glass surface defects based on machine vision detection device and method | |
| CN109307675A (en) | A kind of product appearance detection method and system | |
| CN109974582A (en) | A kind of the conductor diameters non-contact vision detection device and method of automotive wire bundle | |
| CN102374996B (en) | Multicast detection device and method for full-depth tooth side face defects of bevel gear | |
| CN106052586A (en) | Stone big board surface contour dimension obtaining system and method based on machine vision | |
| CN102680494B (en) | Based on arcuation face, the polishing metal flaw real-time detection method of machine vision | |
| CN210071686U (en) | Fruit grading plant based on orthogonal binocular machine vision | |
| CN109461156B (en) | Threaded sealing plug assembly detection method based on vision | |
| CN110146516A (en) | Fruit sorter based on orthogonal binocular machine vision | |
| CN103091332B (en) | Detection method and detection system of U-shaped powder pipe based on machine vision | |
| CN103063137A (en) | Medicine bottle measuring system based on machine vision and measuring system thereof | |
| CN108802052A (en) | A kind of detecting system and its detection method about slide fastener defect | |
| CN113894055A (en) | Hardware surface defect detection and classification system and method based on machine vision | |
| CN110567968A (en) | part defect detection method and device | |
| CN112819844B (en) | Image edge detection method and device | |
| CN108520260B (en) | Method for identifying visible foreign matters in bottled oral liquid | |
| CN113000413A (en) | System, method and terminal for detecting surface defects of synchronizer gear sleeve based on machine vision | |
| CN111458345A (en) | Visual detection mechanism for defects of mask | |
| CN114280075A (en) | Online visual inspection system and method for surface defects of pipe parts | |
| CN106093051A (en) | Paper roll tangent plane burr detection method based on machine vision and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130724 Termination date: 20200116 |