CN108171242A - A kind of efficient baffle ring detecting system - Google Patents
A kind of efficient baffle ring detecting system Download PDFInfo
- Publication number
- CN108171242A CN108171242A CN201611117516.XA CN201611117516A CN108171242A CN 108171242 A CN108171242 A CN 108171242A CN 201611117516 A CN201611117516 A CN 201611117516A CN 108171242 A CN108171242 A CN 108171242A
- Authority
- CN
- China
- Prior art keywords
- image
- baffle ring
- detecting system
- module
- detection
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/30—Noise filtering
Abstract
The invention discloses a kind of efficient baffle ring detecting system, which includes：Image capture module and image processing module；Wherein, described image acquisition module is connected with described image processing module by USB interface.The present invention program passes through a series of image acquisition, the operating process of video procession, realize the automatic detection and identification to baffle ring, the shortcomings that assembly center accuracy of detection is low is effectively compensated for, so as to reduce the amount of labour of operator, ensures that assembly work normally completes.
Description
Technical field
The invention belongs to Machine Vision Recognition fields, are related to a kind of efficient baffle ring detecting system.
Background technology
Machine vision is a complex art, including image procossing, mechanical engineering technology, control, electric source lighting, optics
Imaging, sensor, simulation and digital video technology, computer hardware technique.One typical machine vision applications system packet
Include picture catching, light-source system, image digitazation module, Digital Image Processing module, intelligent decision decision-making module and machinery control
Execution module processed.
The characteristics of NI Vision Builder for Automated Inspection is most basic is exactly to improve flexibility and the degree of automation of production.It is unsuitable at some
The dangerous work environment of manual work or artificial vision are difficult to the occasion met the requirements, and machine in normal service vision manually regards to substitute
Feel.Meanwhile in the repeated industrial processes of high-volume, the efficiency of production can be greatly improved with machine vision detection method
And the degree of automation.
With the development of intelligent Production Technology, machine vision is had been widely used in the automated production of process industry,
Production cost is significantly reduced, the qualification rate of product is ensure that, greatly improves enterprises production efficiency.
Invention content
Present invention aims at a kind of efficient baffle ring detecting system of offer, at a series of image acquisition, image
Reason and the operating process of identification realize automatic detection and identification to baffle ring, effectively compensate for assembly center accuracy of detection
The shortcomings that low, so as to reduce the amount of labour of operator, ensures that assembly work normally completes.
In order to solve the above technical problems, the present invention adopts the following technical scheme that：A kind of efficient baffle ring detecting system, should
System includes：Image capture module and image processing module；Wherein, described image acquisition module leads to described image processing module
USB interface is crossed to be connected.
Further, described image acquisition module is made of industrial camera, light source and detection target, to be checked for acquiring
Target image.
Further, described image processing module is made of display and industrial computer, be used to implement image preprocessing,
Edge detection is identified and is matched with image.
The present invention has following advantageous effect compared with prior art：
The present invention program is realized by a series of image acquisition, the operating process of video procession to baffle ring
Automatic detection and identification.
Description of the drawings
Fig. 1 is the structure diagram of efficient baffle ring detecting system.
Fig. 2 is the image procossing schematic diagram of image processing module.
Specific embodiment
Below in conjunction with the accompanying drawings and specific embodiment the present invention is carried out in further detail with complete explanation.It is appreciated that
It is that specific embodiment described herein is only used for explaining the present invention rather than limitation of the invention.
With reference to Fig. 1, a kind of efficient baffle ring detecting system of the invention, which includes：Image capture module and image
Processing module；Wherein, described image acquisition module is connected with described image processing module by USB interface.
Wherein,
(1) described image acquisition module is made of industrial camera, light source and detection target, for acquiring target figure to be checked
Picture.
In order to reach relatively good detection result, Daheng's industrial camera, model DH-HV3151UC, due to selected are selected
CMOS cameras are USB interface, can directly be connected with computer, without buying image pick-up card, reduce cost；Light source selects
It is illuminated with double LED light, the energy consumption of LED light is very small, safety and stability, while meets the lighting requirement of this system.
During Image Acquisition, need to adjust distance between CMOS cameras and baffle ring to be detected, make CMOS cameras
Camera lens face baffle ring to be detected, while the distance of LED light, angle are adjusted, to reach good lighting condition, so as to obtain
The preferable image of quality.
Its course of work is to be taken pictures by CMOS cameras to baffle ring to be detected, acquires digital picture, while pass through USB
Interface will store and process in image transmitting to computer.
(2) described image processing module is made of display and industrial computer, is used to implement image preprocessing, edge inspection
It surveys and identifies and match with image.
Image procossing is very important part in image-detection process.With reference to Fig. 2, due to the shadow of outside environmental elements
It rings, the image to be detected obtained by imaging sensor is often of low quality, it is impossible to meet the requirement of detecting system, this is just needed
It the pretreatments such as to be filtered first to original image, the processing such as binaryzation, edge detection then be carried out to image again, so as to subtract
Few influence of the extraneous factor to image to be detected, is more clear the image of acquisition, visual effect is more preferable, and local feature is brighter
It is aobvious, to carry out image identification and matching treatment.
(2.1) image gray processing
Coloured image is little to workpiece sensing use, while for convenience of calculation, generally before image filtering is carried out, will scheme
As gray processing, generally use the following formula colors image into gray level image：
G=0.299r+0.587g+0.114b
In formula, G is gray value of image, and r, g, b are respectively coloured image red, green, blue component value.
(2.2) image filtering
Image filtering is an indispensable step in image preprocessing, is in the premise for retaining image detail feature as possible
Under, the processing of noise is removed to original image.Image filtering can improve the quality of image, improve visual effect, image
The quality of filtering often influences subsequent image identification and matching treatment.Image filtering is divided into filter in spatial domain and frequency domain filter
Wave, wherein, filter in spatial domain is broadly divided into mean filter and medium filtering, and frequency domain filtering is divided into low pass and high-pass filtering.
Mean filter can effectively eliminate partial noise interference, make pixel grey scale in neighborhood more uniform, smooth, but simultaneously
Also image is made to thicken, edge details are not clear enough；Medium filtering, which overcomes mean filter, makes lacking for image blurring
Point, can be under the premise of edge details not be influenced, to the eradicating efficacy of noise clearly, especially salt-pepper noise.Therefore,
This system is filtered image using medium filtering.
Medium filtering (comes from paper " machine vision automatic measurement technique ", author：Yu Wenyong, stone paint Beijing:Chemical work
Industry publishing house, 2013:94.) belong to nonlinear filtering, be typical a kind of low-pass filtering in statistical filtering.The base of medium filtering
Present principles are to select a sliding window M, the gray value of pixel in window is ranked up and takes intermediate value, then with the intermediate value
Replace the gray value of specified pixel, i.e.,
G (x, y)=med { f (x-i, y-i) }
Med takes median operation for sequence；
i,j∈M；
F (x, y) is each grey scale pixel value in window M.
It should be noted that the pixel in sliding window M takes odd number under normal circumstances, convenient for taking intermediate value；If pixel takes
Even number, intermediate value are the average value of intermediate two grey scale pixel values.
(2.3) image binaryzation
In order to which more preferably target to be detected in image is distinguished with background, it usually needs image is carried out at binaryzation
Reason.Image binaryzation has many methods, and the most commonly used is threshold method, basic principle is by setting binary conversion treatment ash
Threshold value T is spent, pixel f (x, y) of the gray value of image more than threshold value T is replaced with 255, otherwise being replaced with 0, i.e.,：
G (x, y) be binaryzation after image, by above-mentioned formula, we can be clearly seen that, binary conversion treatment it
Afterwards, original image gray value becomes only 0 and 255 bianry image.Gray value in image is 0 to be partially shown as carrying on the back by we
Scape is worth and is partially shown as target to be detected for 255.
The key of threshold method is the selection of threshold value, and the method for selected threshold is broadly divided into Global thresholding, local threshold
Method and dynamic thresholding method.Global thresholding, as the term suggests be in gray level image choose a fixed threshold value come to image into
Row binary conversion treatment.Global thresholding is simple and easy to do, poor anti jamming capability, is commonly available to have in histogram apparent bimodal
Image, and for the image under the conditions of illumination unevenness etc., binaryzation effect is undesirable.Common Global thresholding has Nogata
Map analysis method and maximum variance between clusters (Ostu algorithms) etc..Local thresholding method is by choosing mask, according to object pixel neighbour
The gray value of each pixel determines the threshold value of object pixel in domain.For more complicated image, the binaryzation of local thresholding method
Treatment effect is more preferable, practical；But easily there is the phenomenon that gray scale disconnecting in each borderline region junction.Common part
Threshold method has Niblack algorithms, Sauvola algorithms and Bernsen algorithms etc..The selection of dynamic thresholding method threshold value is compared with local threshold
It is also related with the position of object pixel in the picture for method.Dynamic thresholding method is actually improved local thresholding method, its energy
It enough handles background and target to be detected distinguishes relatively difficult image, its usual operand is bigger.Common dynamic thresholding method
There are Chow algorithms and Kaneko algorithms etc..
Histogram analysis method is a kind of method of fairly simple selection binary-state threshold, and basic principle is to pass through gray scale
The histogram of image finds the minimum point between two wave crests, i.e. trough gray value is used as the threshold value of binaryzation, for this system
For this method effect it is less desirable.
Ostu algorithms are a kind of global Binarization methods of classics, its simple and effective is stablized, in industrialized production very
It is practical.Maximum variance between clusters can be chosen optimal binary-state threshold automatically and (come from paper " digitized map in entire image
As processing ", author：The strong Xi'an in what east:Publishing house of Xian Electronics Science and Technology University, 2008:84-85.).Its basic principle is, false
Surely an arbitrary gray value t (0≤t≤L) is chosen, the gray value of entire image is divided into t two class A0 and A1, A0 gray value exist
In the range of (0, t), A1 gray values are in the range of (t+1, L), it is assumed that entire image has N number of pixel, and gray value is that the pixel of i has ni
Probability p a, then that gray value i occursiFor pi=ni/ N, then the probability that A0 and two class pixels of A1 occur is,
The mean value of two class pixel of A0 and A1 is,
Inter-class variance is,
σ2(t)=ω0ω1(μ0-μ1)2
Make the hypothesis gray value t of inter-class variance maximum, be our optimal binary-state threshold T to be looked for.Two class picture at this time
The gray value difference of element is maximum, so can well distinguish the background of image and target to be detected, convenient for us to mesh
Mark is identified.
(2.4) edge detection
In order to preferably detect the profile of baffle ring, need to carry out edge detection to its bianry image, to extract image
Edge contour feature.Edge detection is the pith in image procossing, it is to imitate the mistake that human visual system identifies target
The edge feature value of target image by algorithm is extracted, judges that target to be detected is further according to these characteristic values by journey
It is no to meet the requirements.The detection algorithm of edge detection has very much, common to have Sobel operators, Laplace operator (Laplacian
Operator) and Canny operator (Canny operators) etc..
Sobel operators are simple and easy to do, simultaneously because the only template of transverse and longitudinal both direction, it can only detection level and Vertical Square
To edge, therefore, for the more complicated image border of detection texture, effect is not highly desirable.
Laplace operator is a kind of second derivative operator, very sensitive to noise, can generate bilateral effect, it is impossible to detect
Go out the direction on side, be usually not directly used for the detection on side, only play a part of assist, for detect a pixel be side it is bright one
Side or dark one side.
Canny operators are a kind of single order edge detection operators of classics, and the method for edge detection is to find image gradient
Local maximum (come from paper " the canny operator edge detections based on edge preserving smooth filter ", author：Chen Hong wishes orchids
State university of communications journal, 2006,25 (1):86-90.).The essence of this method is that image is put down with a quasi-Gaussian function
Sliding operation
fs=f (x, y) × G (x, y)
Then derivative maximum value is positioned with the first order differential operator with direction.Canny operators are not easily susceptible to the interference of noise,
It is able to detect that real weak edge.
(2.5) image identifies
Image identifies, refers to handle image using computer, analyzed and understood, to identify various different modes
Target and the technology to picture.Image identification is also referred to as pattern-recognition, is based on the main feature of image, so as to be carried out to target
Identification and classification.
The key of image identification is extraction and analysis to characteristics of image, and image characteristics extraction is to image target edge
The features such as perimeter, area, circularity measure, the characteristic value measured can be with composition characteristic vector, so as to as special
The foundation of sign extraction and classification.Statistical-simulation spectrometry is a kind of image identification side that is most ripe at present and being most widely used
Method, it is by carrying out big measurement statistical analysis to target image, finding out its characteristic rule, and thus selection can reflect
The feature of image essence is identified and classifies.The present invention classifies to target image using minimum distance classifier,
Its assorting process is as follows：
A. the average vector of present mode is obtained
In formula, NiFor WiThe number of quasi-mode vector, m are the number of pattern class.
B. Euclidean distance is calculated
Di(x)=‖ x-Wi‖ (i=1,2 ..., m)
According to Di(x) value sorts out vector, Di(x) when value is minimum range, vector is grouped into WiClass.
Target image can be classified by this minimum distance classifier, so as to divide baffle ring to be detected
Class, identifies whether it meets the requirements.
The foregoing is merely the preferred embodiment of the present invention, are not intended to restrict the invention, for those skilled in the art
For, the present invention can have various modifications and changes.All any modifications made within spirit and principles of the present invention are equal
Replace, improve etc., it should all be included in the protection scope of the present invention.
Claims (3)
1. a kind of efficient baffle ring detecting system, which is characterized in that the system comprises：Image capture module and image procossing mould
Block；Wherein, described image acquisition module is connected with described image processing module by USB interface.
2. efficient baffle ring detecting system according to claim 1, which is characterized in that described image acquisition module is by industry
Camera, light source and detection target composition, for acquiring target image to be checked.
3. efficient baffle ring detecting system according to claim 1, which is characterized in that described image processing module is by showing
Device is formed with industrial computer, is used to implement image preprocessing, edge detection is identified and matched with image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611117516.XA CN108171242A (en) | 2016-12-07 | 2016-12-07 | A kind of efficient baffle ring detecting system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611117516.XA CN108171242A (en) | 2016-12-07 | 2016-12-07 | A kind of efficient baffle ring detecting system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108171242A true CN108171242A (en) | 2018-06-15 |
Family
ID=62526910
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611117516.XA Pending CN108171242A (en) | 2016-12-07 | 2016-12-07 | A kind of efficient baffle ring detecting system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108171242A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109087286A (en) * | 2018-07-17 | 2018-12-25 | 江西财经大学 | A kind of detection method and application based on Computer Image Processing and pattern-recognition |
-
2016
- 2016-12-07 CN CN201611117516.XA patent/CN108171242A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109087286A (en) * | 2018-07-17 | 2018-12-25 | 江西财经大学 | A kind of detection method and application based on Computer Image Processing and pattern-recognition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107705288B (en) | Infrared video detection method for dangerous gas leakage under strong interference of pseudo-target motion | |
| CN107808161B (en) | Underwater target identification method based on optical vision | |
| CN109872300B (en) | Visual saliency detection method for appearance defects of friction plate | |
| GB2569751A (en) | Static infrared thermal image processing-based underground pipe leakage detection method | |
| CN108491892A (en) | fruit sorting system based on machine vision | |
| CN108288264B (en) | Wide-angle camera module contamination testing method | |
| CN108765402B (en) | Non-woven fabric defect detection and classification method | |
| CN108596880A (en) | Weld defect feature extraction based on image procossing and welding quality analysis method | |
| CN110807355A (en) | Pointer instrument detection and reading identification method based on mobile robot | |
| CN108387516A (en) | Threading defects detection device and detection method based on CCD camera image processing technologies | |
| Feng et al. | A separating method of adjacent apples based on machine vision and chain code information | |
| CN106651893A (en) | Edge detection-based wall body crack identification method | |
| CN103198319A (en) | Method of extraction of corner of blurred image in mine shaft environment | |
| CN110807763A (en) | Method and system for detecting ceramic tile surface bulge | |
| CN108109154A (en) | A kind of new positioning of workpiece and data capture method | |
| CN107818583A (en) | Cross searching detection method and device | |
| CN113971681A (en) | Edge detection method for belt conveyor in complex environment | |
| CN107545565B (en) | Solar screen plate detection method | |
| CN109544535B (en) | Peeping camera detection method and system based on optical filtering characteristics of infrared cut-off filter | |
| CN108171242A (en) | A kind of efficient baffle ring detecting system | |
| CN117036259A (en) | Metal plate surface defect detection method based on deep learning | |
| CN115343313A (en) | Visual identification method based on artificial intelligence | |
| Xu et al. | Computer assistance image processing spores counting | |
| CN105115987B (en) | Taper roller upside-down mounting defect inspection method based on digital filtering | |
| Yang et al. | Cherry recognition based on color channel transform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180615 |