CN106204631B - Plug-hole detection method based on frequency domain - Google Patents
Plug-hole detection method based on frequency domain Download PDFInfo
- Publication number
- CN106204631B CN106204631B CN201610704539.4A CN201610704539A CN106204631B CN 106204631 B CN106204631 B CN 106204631B CN 201610704539 A CN201610704539 A CN 201610704539A CN 106204631 B CN106204631 B CN 106204631B
- Authority
- CN
- China
- Prior art keywords
- hole
- plug
- original image
- frequency domain
- region
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20056—Discrete and fast Fourier transform, [DFT, FFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20061—Hough transform
Abstract
The plug-hole detection method based on frequency domain that the invention discloses a kind of, can accurately detect plug-hole number.It include: step A, the original image for obtaining body target area to be detected;Step B, after the original image successively to be carried out to Fourier transformation, low-pass filtering, inverse Fourier transform processing, then with original image progress subtraction, the frequency domain for obtaining hole sharpness of border handles picture;Step C, at least one hole is chosen in frequency domain processing picture and template is made according to the profile of selected hole, then do template matching in frequency domain processing picture, the pass blocking of the hole of position is judged according to the gray scale of each position.Frequency domain processing is carried out to the picture of target area, it shows the sharpness of border of each hole, using wherein at least one hole as template, template matching is carried out to the hole in target area, accurately judged by gray scale each perforated be it is logical or stifled, plug-hole number can be accurately calculated.
Description
Technical field
The plug-hole detection method based on frequency domain that the present invention relates to a kind of.
Background technique
Such as porous catalyst porous mass needs to detect plug-hole number, to judge whether product is qualified.With car tail
For the three-way catalyst of gas, three-way catalyst is a kind of cylindrical body with big density honeycomb hole, in the prior art to three
The plug-hole detection method of first catalyst is: take the three-way catalyst of a standard as sample, by camera collect this three
Template image is made using the objective contour of sample top view in the overhead view image of first catalyst sample;In detection three-way catalyst
When, the top view (hereinafter referred to as image to be detected) that each three-way catalyst to be checked is as shown in Figure 1 is acquired, by each detection image
It is matched with template, theoretically, the corresponding coordinate of each hole in template image can be found in image to be detected
Corresponding coordinate, then judges whether hole blocks.However, it fails to match for meeting under some extreme cases for this method, such as go out
The case where existing carrier section inclination or large area block (as shown in Figures 2 and 3), can not accurately detect plug-hole number.
Summary of the invention
In view of the above-mentioned problems, the object of the present invention is to provide a kind of plug-hole detection method based on frequency domain, it can be accurate
Detect plug-hole number.
In order to solve the above technical problems, the technical solution adopted by the present invention are as follows:
A kind of plug-hole detection method based on frequency domain, comprising:
Step A, the original image of body target area to be detected is obtained;
Step B, the original image is successively carried out Fourier transformation, low-pass filtering, inverse Fourier transform processing after, then with institute
It states original image and carries out subtraction, obtain the frequency domain processing picture of hole sharpness of border;
Step C, at least one hole is chosen in frequency domain processing picture and mould is made according to the profile of selected hole
Plate, then does template matching in frequency domain processing picture, judges the logical of the hole of position according to the gray scale of each position
It is stifled.
Preferably, step A includes:
A1, the outer profile for being identified body target area to be detected using Hough transformation in the entire image of body to be detected, are grabbed
The region in outer profile is taken to form the original image, the region in outer profile is the target area.
Preferably, step A includes:
A2, the original image is rotated, arranges hole along x-axis or y-axis.
It is highly preferred that, by the original image binaryzation, retaining the boundary of hole in step a2, identified using Hough transformation straight
Line rotates the original image until the straight line extends along x-axis or extends along x-axis.
Preferably, the plug-hole detection method further include:
Step D, doubtful blocking region is found;
In step C, template matching is done in doubtful blocking region in frequency domain processing picture, according to each in blocking region
The gray scale of position judges the pass blocking of the hole of position.
It is highly preferred that the original image binaryzation is retained hole boundary and blocking region, then to the original image after binaryzation
Morphology opening operation is carried out, normal hole boundary is removed, remaining area is the doubtful blocking region.
Further, in step C, in frequency domain processing picture in each doubtful plug-hole region, according to template matching, one is found
A best match position obtains the perforated of best match position according to the size and shape of template, utilizes the best match
All perforated in doubtful plug-hole region are found in the perforated of position, judge the pass blocking of each hole respectively.
Preferably, this method successively includes:
The foreign steamer of body target area to be detected is identified in step A, in the entire image of body to be detected using Hough transformation
Exterior feature, the region grabbed in outer profile form the original image, and the region in outer profile is the target area;The original image is carried out
Rotation arranges hole along x-axis or y-axis;
Step D, by the original image binaryzation, retain hole boundary and blocking region, then to the original image after binaryzation into
Row morphology opening operation, removes normal hole boundary, and remaining area is the doubtful blocking region;
Step B, the original image is successively carried out Fourier transformation, low-pass filtering, inverse Fourier transform processing after, then with institute
It states original image and carries out subtraction, obtain the frequency domain processing picture of hole sharpness of border;
Step C, a hole is chosen in frequency domain processing picture and template is made according to the profile of selected hole, so
, according to template matching, a best match position is found, according to template in each doubtful plug-hole region in frequency domain processing picture afterwards
Size and shape, obtain the perforated of best match position, found using the perforated of the best match position doubtful
All perforated in plug-hole region, judge the pass blocking of each hole respectively.
Preferably, described hole array is distributed.
Preferably, the body to be detected is the three-way catalyst of vehicle maintenance service.
The present invention by adopting the above technical scheme, has the advantages that the picture progress to target area compared with prior art
Frequency domain processing, shows the sharpness of border of each hole, using wherein at least one hole as template, in target area
Hole carries out template matching, accurately judged by gray scale each perforated be it is logical or stifled, plug-hole number can be accurately calculated.
Detailed description of the invention
Fig. 1 is the top view of the three-way catalyst sample of shooting;
Fig. 2 is the top view shot after three-way catalyst tilts;
Fig. 3 is the top view that shoots after three-way catalyst large area plug-hole;
Fig. 4 is the picture of the three-way catalyst after low-pass filtering treatment
Fig. 5 is that the frequency domain of three-way catalyst handles picture.
Specific embodiment
The preferred embodiments of the present invention will be described in detail with reference to the accompanying drawing, so that advantages and features of the invention energy
It is easier to be understood by the person skilled in the art.
By taking automobile is located at the three-way catalyst of processing as an example, the plug-hole detection method of the invention based on frequency domain is carried out
Explanation.Certainly, method of the invention is not limited to detection three-way catalyst, other similar porous and with periodic feature objects
Body is all suitable for this method, that is, each hole is arranged in array.
Method of the invention in turn includes the following steps: A, detecting target area, image rotation;D, image rotation;D, it seeks
Look for doubtful blocking region;B, frequency-domain analysis processing is carried out to target area;C, images match is done using shape feature.Below to each
A step is described in detail.
The purpose of step A is to obtain the original image of the target area to three-way catalyst, the target area tool of three-way catalyst
Body refers to all perforated in top view as shown in Figure 1.Step A is specifically included:
A1, in the whole picture overhead view image of three-way catalyst using Hough transformation identification circle (three-way catalyst is cylindrical,
Its target area corresponds to circle) outer profile, grab outer profile in region form the original image, the region in outer profile is
The target area;
A2, the original image is rotated, make hole along x-axis or y-axis arrangement (for popular, as smooth vertical and horizontal row
Cloth direction), detailed process is, by the original image binaryzation, to retain the boundary of hole, identifies straight line using Hough transformation, utilizes
The statistical property of Hough transformation can find a longest straight line in target area, this line correspondences is in closest in circle
The straight line in the center of circle, and the inclination angle of this straight line seeks to the angle of rotation, rotate the original image according to the angle until
The straight line along x-axis extend or.When due to detection, the arragement direction of target Hole be it is random, be unfavorable for detecting, thus right
Image is rotated, and the arragement direction of hole is become smooth vertical and horizontal, and hole rule is made to arrange, known to hole distribution.
The purpose of step D is to find doubtful blocking region, since detection target complete region time-consuming is very long, can first be done pre-
Sentence, get rid of apparent throughhole portions, only doubtful blocking region is detected.For example, 4 holes that 2 adjacent rows 2 arrange, at this
There is blocking in a regional center, although seeming that total blocked area has reached the area of single hole, but this practical 4 holes are each
Hole has only blocked 1/4, so each hole is not counting blocking, this region is exactly a kind of doubtful blocking region.Specific practice are as follows: will
The original image binaryzation retains hole boundary and blocking region, then carries out morphology opening operation to the original image after binaryzation, goes
Except normal hole boundary, remaining area is the doubtful blocking region.Doubtful blocking region is in original image and the frequency being described below
There are corresponding relationships in domain processing picture and the picture of each processing stage.
The purpose of step B is to be revealed the porose boundary of institute in target area all, and all holes all have clearly
Boundary separate.Due to there is the hole of regular distributions much more very in target area, very strong periodicity is shown, so,
It is very suitable to frequency domain processing.Grey scale change inside hole is gentle, and most acute variations is all distributed in hole boundary,
So corresponding low frequency part is hole interior zone, and high frequency section corresponds to hole boundary in frequency domain processing image.Tool
Body is that the original image after processing of step A is carried out Fourier transformation, then carries out low-pass filtering, inhibits high fdrequency component, then carry out inverse
Fourier transformation, the very smooth image of an available width, the image can not see boundary at normal hole, and
In large area plug-hole region, the boundary inside hole is high-visible, as shown in Figure 4.Diagram piece and original picture are done into subtraction fortune
It calculates, the width image that the boundary of institute's hole is all separated can be obtained, referred to as frequency domain handles image, as shown in Figure 5.
In step C, in the resulting frequency domain processing picture of step B, a hole is chosen, mould is made according to the shape of hole
Plate.In the frequency domain processing picture that step B is obtained, template matching is done in the doubtful blocking region that all step D are marked off, is sentenced
Break the gray scale of each matched position, can judge that the hole be logical is stifled.Specific step is as follows: in each suspicious region,
First according to template matching, a best match position is found, it, can be with according to the size and shape (such as regular hexagon) of hole
The perforated for knowing best match, since step a2 has rotated image, so the distribution of hole is known, benefit
Perforated all in suspicious region can be found with the hole of best match, judge respectively each perforated be it is logical or
It is stifled, all plug-hole numbers can be accurately calculated.
The above embodiments merely illustrate the technical concept and features of the present invention, is a kind of preferred embodiment, and purpose exists
In those skilled in the art can understand the contents of the present invention and implements accordingly, guarantor of the invention can not be limited with this
Protect range.Equivalent change or modification made by all Spirit Essences according to the present invention, should all cover in protection scope of the present invention
Within.
Claims (10)
1. a kind of plug-hole detection method based on frequency domain characterized by comprising
Step A, the original image of body target area to be detected is obtained, the body to be detected is porous object;
Step B, the original image is successively carried out Fourier transformation, low-pass filtering, inverse Fourier transform processing after, then with the original
Figure carries out subtraction, obtains the frequency domain processing picture of hole sharpness of border;
Step C, at least one hole is chosen in frequency domain processing picture and template is made according to the profile of selected hole, so
Template matching is done in frequency domain processing picture afterwards, the pass blocking of the hole of position is judged according to the gray scale of each position.
2. plug-hole detection method according to claim 1, which is characterized in that step A includes:
A1, the outer profile for identifying body target area to be detected using Hough transformation in the entire image of body to be detected, crawl are outer
Region in profile forms the original image, and the region in outer profile is the target area.
3. plug-hole detection method according to claim 1 or 2, which is characterized in that step A includes:
A2, the original image is rotated, arranges hole along x-axis or y-axis.
4. plug-hole detection method according to claim 3, which is characterized in that in step a2, the original image binaryzation is protected
It boxes out the boundary in hole, identifies straight line using Hough transformation, rotate the original image until the straight line extends along x-axis or prolongs along x-axis
It stretches.
5. plug-hole detection method according to claim 1, which is characterized in that the plug-hole detection method further include:
Step A ', doubtful blocking region is found;
In the step C, template matching is done in doubtful blocking region in frequency domain processing picture, according to each in blocking region
The gray scale of position judges the pass blocking of the hole of position.
6. plug-hole detection method according to claim 5, which is characterized in that by the original image binaryzation, retain hole side
Then boundary and blocking region carry out morphology opening operation to the original image after binaryzation, remove normal hole boundary, remaining area
For the doubtful blocking region.
7. plug-hole detection method according to claim 5 or 6, which is characterized in that in step C, in frequency domain processing picture
In each doubtful plug-hole region, according to template matching, finds a best match position and obtained according to the size and shape of template
The perforated of best match position is found all in doubtful plug-hole region using the perforated of the best match position
Perforated judges the pass blocking of each hole respectively.
8. plug-hole detection method according to claim 1, which is characterized in that successively include:
The outer profile for being identified body target area to be detected in step A, in the entire image of body to be detected using Hough transformation, is grabbed
The region in outer profile is taken to form the original image, the region in outer profile is the target area;The original image is rotated,
Arrange hole along x-axis or y-axis;
Step A ', by the postrotational original image binaryzation, retain hole boundary and blocking region, then to the original after binaryzation
Figure carries out morphology opening operation, removes normal hole boundary, and remaining area is doubtful blocking region;
Step B, after the postrotational original image successively being carried out Fourier transformation, low-pass filtering, inverse Fourier transform processing, then
Subtraction is carried out with the postrotational original image, obtains the frequency domain processing picture of hole sharpness of border;
Step C, a hole is chosen in frequency domain processing picture and template is made according to the profile of selected hole, then exist
In frequency domain processing picture in each doubtful plug-hole region, according to template matching, a best match position is found, according to the big of template
Small and shape obtains the perforated of best match position, finds doubtful plug-hole using the perforated of the best match position
All perforated in region, judge the pass blocking of each hole respectively.
9. plug-hole detection method according to claim 1, which is characterized in that the distribution of described hole array.
10. plug-hole detection method according to claim 1, which is characterized in that the body to be detected is vehicle maintenance service
Three-way catalyst.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610704539.4A CN106204631B (en) | 2016-08-23 | 2016-08-23 | Plug-hole detection method based on frequency domain |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610704539.4A CN106204631B (en) | 2016-08-23 | 2016-08-23 | Plug-hole detection method based on frequency domain |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106204631A CN106204631A (en) | 2016-12-07 |
CN106204631B true CN106204631B (en) | 2019-01-15 |
Family
ID=57523436
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610704539.4A Active CN106204631B (en) | 2016-08-23 | 2016-08-23 | Plug-hole detection method based on frequency domain |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106204631B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116129331B (en) * | 2023-04-12 | 2023-07-04 | 成都航空职业技术学院 | Method and device for counting number of through holes on surface of rotary part |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102147857A (en) * | 2011-03-22 | 2011-08-10 | 黄晓华 | Image processing method for detecting similar round by using improved hough transformation |
CN102945368A (en) * | 2012-10-17 | 2013-02-27 | 西安理工大学 | Method for positioning and identifying laser character of beer bottle cap |
CN103400129A (en) * | 2013-07-22 | 2013-11-20 | 中国科学院光电技术研究所 | Target tracking method based on frequency domain saliency |
CN104091170A (en) * | 2014-06-13 | 2014-10-08 | 电子科技大学 | Automatic mould distinguishing method based on machine vision |
-
2016
- 2016-08-23 CN CN201610704539.4A patent/CN106204631B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102147857A (en) * | 2011-03-22 | 2011-08-10 | 黄晓华 | Image processing method for detecting similar round by using improved hough transformation |
CN102945368A (en) * | 2012-10-17 | 2013-02-27 | 西安理工大学 | Method for positioning and identifying laser character of beer bottle cap |
CN103400129A (en) * | 2013-07-22 | 2013-11-20 | 中国科学院光电技术研究所 | Target tracking method based on frequency domain saliency |
CN104091170A (en) * | 2014-06-13 | 2014-10-08 | 电子科技大学 | Automatic mould distinguishing method based on machine vision |
Non-Patent Citations (1)
Title |
---|
基于帘子布疵点检测系统的研究;杨扬;《中国优秀硕士学位论文全文数据库 信息科技辑》;20150415;参见第I页第1-3段,第4页第2段-参见第5页第3段,第12页第1段-第28页第3段 |
Also Published As
Publication number | Publication date |
---|---|
CN106204631A (en) | 2016-12-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Raghavendra et al. | Face presentation attack detection across spectrum using time-frequency descriptors of maximal response in laplacian scale-space | |
CN110163853B (en) | Edge defect detection method | |
EP3384429B1 (en) | Method for identification of candidate points as possible characteristic points of a calibration pattern within an image of the calibration pattern | |
Wei et al. | Counterfeit iris detection based on texture analysis | |
KR950024115A (en) | Pattern Image Processing Apparatus and Image Processing Method | |
CN108052917B (en) | Method for automatically identifying illegal buildings based on new and old time phase change discovery | |
CN109493339B (en) | Endoscope imaging-based method for detecting defects of pores on inner surface of casting | |
CN106485694B (en) | A kind of high iron catenary double-jacket tube connector six-sided nut based on cascade classifier falls off defective mode detection method | |
WO2013109193A1 (en) | Computational methods and apparatus for meiboqraphy | |
CN103996017B (en) | One kind being based on HuThe not Ship Detection of bending moment and support vector machines | |
JP2008278288A5 (en) | ||
CN106204631B (en) | Plug-hole detection method based on frequency domain | |
CN107909593B (en) | Non-directional selective contour detection method based on receptive field region | |
KR20140141027A (en) | A method and apparatus for detecting a repetitive pattern in image | |
CN109858438A (en) | A kind of method for detecting lane lines based on models fitting | |
CN112750107A (en) | Spherical workpiece identification method and system based on contour edge number fitting and convex hull processing | |
CN105608689B (en) | A kind of panoramic mosaic elimination characteristics of image error hiding method and device | |
CN108647706A (en) | Article identification classification based on machine vision and flaw detection method | |
WO2013102797A1 (en) | System and method for detecting targets in maritime surveillance applications | |
CN102663706A (en) | Adaptive weighted mean value filtering method based on diamond template | |
CN107766866B (en) | Direction selective contour detection method based on receptive field region | |
Zhang et al. | Small target detection based on difference accumulation and Gaussian curvature under complex conditions | |
CN111428624A (en) | Optical remote sensing image on-orbit ship target detection method | |
US20160133023A1 (en) | Method for image processing, presence detector and illumination system | |
Amin et al. | An automatic number plate recognition of Bangladeshi vehicles |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB02 | Change of applicant information | ||
CB02 | Change of applicant information |
Address after: 215100 AD District, South Side of 1F-2F Building No. 38 North Guandu Road, Yuexi Street, Wuzhong District, Suzhou City, Jiangsu Province Applicant after: SUZHOU DCCK MEASUREMENT AND CONTROL TECHNOLOGY CO., LTD. Address before: 215123 5th Floor, Block B3, 1258 Jinfeng South Road, Wuzhong District, Suzhou City, Jiangsu Province (Jinqiao Auto Parts Industrial Park) Applicant before: SUZHOU DCCK MEASUREMENT AND CONTROL TECHNOLOGY CO., LTD. |
|
GR01 | Patent grant | ||
GR01 | Patent grant |