CN106600606A - Ship painting profile detection method based on image segmentation - Google Patents
Ship painting profile detection method based on image segmentation Download PDFInfo
- Publication number
- CN106600606A CN106600606A CN201611179450.7A CN201611179450A CN106600606A CN 106600606 A CN106600606 A CN 106600606A CN 201611179450 A CN201611179450 A CN 201611179450A CN 106600606 A CN106600606 A CN 106600606A
- Authority
- CN
- China
- Prior art keywords
- image
- profile
- segmentation
- method based
- painting
- 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
- 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
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30152—Solder
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30156—Vehicle coating
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a ship painting profile detection method based on image segmentation and relates to the technical field of image processing. The method comprises the steps of: image inputting; image preprocessing; image segmentation; extracting a painting binary image; image morphological filtering; extracting a painting binary image profile; screening profiles; outputting a profile point represented by pixel coordinates. Experiments show that the profile detected by the algorithm is clear and consistent, and very satisfactory. Through the method, such as camera calibration and the like, the actual painting profile size and the painting repair coordinates can be obtained accurately, which not only meets the ship painting profile detection requirements, and achieves simple operation and high computational efficiency. As one of prerequisites for the realization of painting repair automation, the method lays a solid foundation for strengthening the research and development of the painting operation, and provides basic technical conditions for realizing the automation of painting repair, improving the productivity and reducing the cost of the shipbuilding industry.
Description
Technical field
The present invention relates to technical field of image processing, refers specifically to a kind of vessel coating contour detecting side based on image segmentation
Method.
Background technology
In shipbuilding industry, the steel plate after hull framework is typically processed by some pieces is welded, in order to prevent steel plate quilt
Corrosion, improves the presentation quality of hull, can carry out spraying operation before welding to steel plate, applies layer number as needed, generally
Have 2-5 layers.In welding process, the application that the high temperature for producing can be damaged around welding region is welded, after the completion of welding, damaged
Bad application will be buffed off and spray again.At present, at home in shipyard, these work are still adopted and are accomplished manually:Workman
First the application for damaging all is polished off, then along the region outside bevel layer by layer of last polishing, directly
To all of coating is shown.
The automatization that application is repaired is realized, while it is to improve shipbuilding industry life to strengthen the unmanned developmental research of painting operation
The effective means of force of labor, reduces cost.Replaced manually in marine surface using non-structural automation equipment (such as climbing robot)
Work, automatic searching painting defect position, auto-mending application, it appears that be a kind of ideal solution, but it is existing
Automation equipment lack the support of some core algorithms, the profile point multiplicity detected by conventional contour detecting algorithm
High, accuracy is poor, fracture, it is impossible to the requirement that satisfaction is accurately positioned, and also cannot just make automation equipment accurately find needs and repair
Coating dress precise boundary, it is impossible to realize application repair automatization with it is unmanned.Further, in vessel coating contour detecting
In, the profile point for detecting must be continuous, and the quantity of profile must be set, and these are to realize that automatization is repaired in application
One of precondition.Profile testing method common at present is generally basede on point of the change of gradation of image to judge in image
No to belong to profile point, the point detected using these methods may be discontinuous, and comprising a large amount of non-profile points, the quantity of profile is also
Can not set.
The content of the invention
It is an object of the invention to overcome the disappearance and deficiency of prior art presence, propose a kind of for many of vessel coating
Threshold values profile testing method, it is contemplated that the regionality of application image substantially, is adapted to image segmentation, and the present invention devises one kind and is based on
The contour detecting algorithm of image segmentation, extracts each layer application region respectively in one application bianry image, then therefrom detects
And profile is screened, finally demonstrate the effect of the method.
A kind of vessel coating profile testing method based on image segmentation, including step：
Image is input into；Image semantic classification；Image segmentation；Extract application bianry image；Morphological image is filtered；Extract application
Bianry image profile；Screening profile；The profile point that output is represented with pixel coordinate.
Image semantic classification
Image semantic classification is divided into two steps:Input picture is switched to single pass gray level image first；Then to gray level image
Medium filtering.Median filter carries out processes pixel to image, when each pixel is processed, can consider certain around the pixel
Pixel in region, sorts them according to the size of gray value, chooses the intermediate value of gray value replacing the gray scale of current pixel
Value.Medium filtering can play a part of image smoothing, and largely effective to eliminating the picture noises such as salt-pepper noise.
Image segmentation and extraction application bianry image
By taking accompanying drawing 2 as an example, this is the typical vessel coating profile gray level image of a width, can be divided the image into according to gray scale
6 regions, middle border circular areas 1 are steel plate, and region 2-6 respectively fuchsin coating, Lycoperdon polymorphum Vitt coating, white coating, redness are applied
Layer, purplish red coating, all coatings all with gray scale mode present, thus can with 5 threshold values image segmentation be 6 regions,
Pixel in each region has identical gray value.Existing many scholars have obtained some multi thresholds images by research at present
The algorithm of segmentation, their basic ideas are all to give an object function f (t1, t2..., tN).Wherein t1, t2..., tNFor threshold
Value, N is number of thresholds, and in interval [0, L-1], (gray levels of the L for image) chooses t1, t2..., tN, make f take maximum or most
Little value.
Otsu methods are a kind of conventional image partition methods, it using the maximum between-cluster variance of image corresponding threshold value as
The optimal threshold of image segmentation.Otsu methods are applied to single threshold image segmentation, but it is also possible to be generalized to multi thresholds, Otsu methods
The maximum between-cluster variance for being used has good effect to splitting this kind of regional significantly image shown in accompanying drawing 2.
The calculating formula of multi thresholds Otsu methods is
In formula, fotsuFor maximum between-cluster variance, the i.e. object function of Otsu methods；K=N+1 is image by N number of threshold value point
The number in the region cut；CkFor the set of gray value in k-th cut zone；piFor the probability that gray value i occurs in the picture；
mGFor gradation of image meansigma methodss.fotsuIt is with segmentation threshold t1, t2..., tNFor the function of independent variable, each function in gray value area
Between [0, L-1] the corresponding threshold value of maximum as optimum segmentation threshold value.
In order to obtain optimum segmentation threshold value, traditional solution procedure travels through all possible t1, t2..., tN, calculate respectively
Their corresponding inter-class variances simultaneously obtain maximum between-cluster variance and its corresponding optimal threshold by comparing.The meter of this method of exhaustion
Calculation complexity is O (LN).With the increase of number of thresholds, computation complexity exponentially increases.Common gray level is L=256,
As N ＞ 5, the calculating time and memory consumption in current main-stream PC has been difficult to receive, so needing to improve computational efficiency.
For Constrained and Unconstrained Optimization of the rapid solving function on interval is specified, by research, many scholars show that some intelligence are excellent
Change algorithm.Wherein particle swarm optimization algorithm as a kind of strong intelligent optimization algorithm of simple, global optimization ability, in multi thresholds figure
As being widely used in segmentation.
Particle swarm optimization algorithm is also known as PSO (particle swarm optimization) algorithm, it is adaptable to solve continuous
Constrained and Unconstrained Optimization of the nonlinear function in search space.PSO algorithms are a kind of intelligent search algorithms based on population, in population
Each individuality can see the particle of a massless and volume in search space as, and the position of each particle represents search space
In a potential solution, flown with certain speed in search space, particle by the study and adjustment to environment, according to individual
Body carrys out dynamic adjustment flight speed with the flying experience of colony, so that colony gradually moves into the more preferable region of search space.Kind
By an adaptive value (fitness value), the quality of each particle in group represents that function that is, to be optimized is in the particle institute
The value in potential solution for representing.Each particle also determines renewal speed and the direction of potential solution with a speed, has one entirely
The maximal rate of office is limiting the speed of each particle.
PSO algorithms are first randomly generated a population, then by iteration updating speed and the position of each particle, search
Seek optimal solution.In each iteration, particle updates the position of oneself by two extreme values:The optimal solution that particle is found in itself
And the optimal solution (gbest) that finds at present of population (pbest).Particle update mathematical expression be：
vid=ω vid+c1r1d(pid-xid)+c2r2d(pnd-xid)； (4)
χid=xid+vid。 (5)
In formula, pid, xidAnd vidRespectively the d of the history optimal solution, current location and present speed of particle i ties up component；
pndFor the d dimension components of globally optimal solution；ω is Inertia Weight；c1And c2For accelerated factor；r1dAnd r2dFor on interval [0,1]
Random number.c1And c2Usually 2.0；ω generally with the increase of iterationses, is linearly changed into 0.4 from 0.9.
Accelerate solution procedure using multi thresholds Otsu method segmentation figure pictures and PSO algorithms, can quickly obtain image segmentation
Optimal threshold combination.
In the case where quality of input image is good, these threshold values every layer of application region segmentation out.Then utilize with
Under method extract bianry image:Optimum segmentation threshold value combination determines some gray value intervals, for each gray value area
Between, definition one is regular, if certain grey scale pixel value in pretreated image is in the interval, the gray scale of the pixel
Value is set as 255.Conversely, being set to 0, store the result in a new image, thus obtained some application binary map
Picture, includes a cut zone in each application bianry image.
The problem that should be noted in image segmentation：Gray scale in vessel coating, between only adjacent coating
It is ensured of what is differed, the gray scale of non-conterminous coating may be identical.If this occurs, segmentation threshold quantity needs setting
For the number of the coating of different gray scales, the coating more than 1 in so having the bianry image of some extractions, is included, this is in profile
Must consider in extracting and screening.
Morphological image is filtered
Noise be might have in the rough application bianry image obtained in previous step, and due to the gray scale of each layer of coating
Distribution value may be uneven, it is possible that some holes or point, directly carrying out contours extract can produce in application bianry image
Substantial amounts of unrelated profile.Morphological image filtering can filter most of interference.
The basic operation of morphological image process, is image to be corroded or is expanded using a structural elements.Corrosion handle
Filter from image less than or equal to the image detail of structural elements.In this example, area little interference filtering, and eliminate
Small protrusion in coating；But corrosion is a kind of contraction or Refinement operation, can also reduce the region of bianry image floating coat.With corruption
Erosion is different, expands the object that can then increase or be roughened in bianry image, in this example, has filled up the hole less than or equal to structural elements
Hole, has repaired the fracture of coating；But this can also increase the region of application bianry image floating coat.
If first corroded to a structural elements image, then image is expanded with identical structural elements again, made
Opening operation., whereas if first being expanded to image with a structural elements, corruption is carried out to image with identical structural elements again then
Erosion, makees closed operation.Corrode and the image detail less than or equal to structural elements is filtered from image.Opening operation typically understands smooth object
Profile, disconnect narrower narrow neck and eliminate thin outthrust；Closed operation equally also can smoothed profile a part, but with open fortune
Calculate conversely, it would generally make narrower interruption and elongated gully, the little hole of elimination, the fracture filled up in contour line up.
In this example, seldom there is fracture in the coating in the bianry image for obtaining, therefore with same structural elements first to figure
Closed operation is carried out again as carrying out opening operation, can effectively eliminate the interference in image.
Application bianry image contours extract and screening
Application bianry image after previous step process has filtered most of interference, then from these application bianry images
Middle extraction profile, corrodes application bianry image using a structural elements, then can be obtained with the image after the corrosion of artwork image subtraction
To continuous profile.This method ensure that the profile for extracting is constituted for continuous pixel.As morphologic filtering can not be protected
The all of interference of card is all filtered out, it is thus possible to detect unrelated profile.Assume that the area of unrelated profile relative to painting
Layer profile is all smaller, according to this it is assumed that comparing the area for detecting profile, removes the little profile of Area comparison, retains face
The maximum front K profile of product, K is the quantity of profile to be detected, is then sat according to the area and mean pixel of the profile for retaining
Mark to judge their position opsition dependent sequence, return using the profile after sequence as the result of algorithm.
Algorithm implementation result
At present common contour detecting algorithm is generally basede on the change of gradation of image to find profile point, accordingly by image wheel
Exterior feature is divided into stepped ramp type, ramp type and roof type three types.Derivative of the gray scale in certain point is to weigh gray scale in the rate of change
Index, therefore derivative be often used as detect profile instrument.In the contour detecting operator based on derivative, comparison basis
There are R oberts operators, Prewitt operators, Sobel operators and Laplace operators.These operators are with one or more templates pair
Based on image filtering, and uncombined picture characteristics and suppression noise.Some more advanced contour detecting operators have Marr-
Hildreth operators and Canny operators, wherein Canny operators are classic.
These methods are widely applied in fields such as medical image, defects detections.Accompanying drawing 3 is detected for Canny operators
Profile obtained by Fig. 2 detects the comparison diagram of 2 gained profile of accompanying drawing with inventive algorithm.
It is as we can see from the figure as the contour detecting algorithm based on grey scale change is easily disturbed by noise, conventional
Contour detecting algorithm has obtained many unrelated profile points and has occurred in that the phenomenon of profile fracture and contour convergence, contour fitting
Degree is also not fully up to expectations, therefore does not meet the requirement of vessel coating contour detecting.And the wheel detected by the inventive algorithm of right side
Exterior feature, clear-cut, coherent, effect is ideal, can obtain exactly accurate actual application wheel by methods such as camera calibrations
Coordinate is repaired in wide size and application, is to realize that the automatization that application is repaired lays a solid foundation.
Description of the drawings
Fig. 1 is basic procedure block diagram of the present invention based on the profile testing method of image segmentation；
The typical vessel coating profile gray level images of Fig. 2；
Fig. 3 is one embodiment of the invention Canny algorithm profile (left side) and inventive algorithm profile (right side).
Specific embodiment
Below in conjunction with drawings and Examples, the invention will be further described
Based on the profile testing method of image segmentation, first to input picture pretreatment, then according to the quantity pair of coating
Image carries out multi-threshold segmentation, each layer application region is extracted in an application bianry image respectively, the painting to obtaining respectively
Dress bianry image carries out morphologic filtering, and profile is extracted and screened in application bianry image, what output was represented with pixel coordinate
Profile point.
The basic flow sheet (as shown in Figure 1) of this method.
First, Image semantic classification
1. input picture is converted to into single channel gray level image.If input is Three Channel Color image, using integer
Algorithm is converted into single channel gray level image.
2. noise reduction is carried out to image using median filter.Tested using the method for exhaustion, chosen N*N neighborhood of pixels construction
Intermediate value, i.e., list each pixel N*N neighborhood of pixels value queue in order, obtain queue intermediate value and replace the pixel.
2nd, image segmentation and extraction application bianry image
1. multi thresholds Otsu method segmentation figure pictures are utilized, and accelerates solution procedure using PSO algorithms, it is quick to obtain image point
Cut optimal threshold combination, using threshold value every layer of application region segmentation out.
First illustrate Otsu methods, to image solve grey level histogram, will gradation of image be divided into 0-255 levels, each picture
Element has a grey level, counts the number of pixels of each grey level.P is obtained according to the result of statisticsiWith mG, select and divide
Cut threshold value N (hierarchy number)；Choose one group of gray threshold t1, t2..., tN, (threshold value value size is 0-255, and t1<t2<…<
tN), threshold value 0-t1Determine a cut zone, t1-t2Confirm a cut zone, the like altogether confirm N+1 cut section
Domain；CkFor the set of gray value in k-th cut zone, each gray scale belonged in the cut zone is obtained using formula (2)
The probability that level pixel occurs in all pixels i.e. area grayscale Probability pk, area grayscale meansigma methodss are obtained using formula (3)
mk；The squared difference of area grayscale meansigma methodss and gradation of image meansigma methodss is obtained into the variance of the cut zone, institute is obtained successively
There is the variance in region；Using formula (1), by the variance of each cut zone respectively with corresponding area grayscale probability multiplication, then
All of product addition is into gray threshold t1, t2..., tNThe inter-class variance of (certain N number of occurrence)；Gray scale threshold is converted afterwards
Value t1, t2..., tNValue obtain corresponding inter-class variance；Finally all of inter-class variance is compared, maximum therein
Value is required optimal threshold combination.
It is then used by PSO algorithms and accelerates gray threshold t1, t2..., tNThe process of optimal threshold is solved, for the purpose of the present invention,
The particle of PSO algorithms is N-dimensional particle, and each dimension is respectively t1, t2..., tNVector, position are respectively x1, x2..., xNLatent (vector length
Spend for 0-255 gray values, meaning is as described in Otsu methods), potential solution is the class tried to achieve using Otsu methods at the position
Between variance；Multiple particles are generated at random so as to move in space, the initial position of particle is random value, at current location,
The optimal solution (gbest) that the potential optimal solution (pbest) for solving and being found with particle of particle calculating in itself and population are found at present is entered
Row compares, and replaces optimal solution with potential solution when potential solution is better than optimal solution, updates particle rapidity, profit using formula (4) afterwards
With formula (5) more new particle current location, then carry out new round iteration, repeat when particle more new position above-mentioned calculating and
The optimum solution preocess of comparison；The particle stop motion when maximum iteration time is reached, optimal solution (gbest) now are maximum kind
Between variance, vector position length is optimum segmentation threshold value.
It is found by experiment that, when segmentation threshold is bigger, the advantage of PSO algorithms is more obvious, when threshold value is more than 5, uses
The efficiency of PSO Algorithm for Solving optimal thresholds is more than 100 times solved using the method for exhaustion.
2. optimum segmentation threshold value combination determines some gray value intervals, by gray value interval, by gradation of image pixel
Binarized pixel is converted to, multiple application bianry images comprising each layer application respectively are obtained.
By one group of optimal threshold being obtained in 1, process first gray threshold first, copy image, and will be in this threshold value
All pixels before are set to 0 (black), after this threshold value have pixel to be set to 255 (whites) more, so just
The only application bianry image comprising ground floor is arrived；Recover original image, again copy image, process second threshold value, will be at this
All pixels before individual threshold value are set to 0 (black), and after this threshold value have pixel to be set to 255 (whites) more, this
Sample has just obtained the only application bianry image comprising the second layer；N+1 image is obtained in this way, wherein N is opened with N+1
Picture shape is identical, and color is conversely, filter off N+1 image.
3rd, morphological image filtering
1. the application bianry image obtained by previous step is filtered using morphological image method, using opening operation horizontal sliding wheel
Wide, elimination outthrust, is eliminated cavity, is filled up fracture, so as to eliminate picture noise using closed operation.
4th, contours extract and screening
1. application bianry image profile is extracted using the method for morphologic filtering, using structural elements corrosion copy application two-value
Image, then the profile that the image for corroding is obtained continuous certain coating is deducted with former application bianry image.
2., in order to prevent from obtaining unrelated profile, the unrelated profile for detecting is filtered using the method for sequence, area is removed
Smaller profile, the maximum front K profile of Retention area, and press area and coordinate position sequence.
3. the profile point that each coating is represented with pixel coordinate is finally obtained, the use of using method is returned to as a result
Family.
In sum, the present invention, as one of precondition for realizing application repairing automatization, is to realize what application was repaired
Automatization, it is to improve the shipbuilding industry productivity, reduces cost to provide the foundation skill to strengthen the unmanned developmental research of painting operation
Art condition.It is demonstrated experimentally that the method not only meets the requirement of vessel coating contour detecting, and it is simple to operate, computational efficiency is high.
Claims (10)
1. a kind of vessel coating profile testing method based on image segmentation, it is characterised in that including step：A. image input, B.
Image semantic classification, C. image segmentations, D. extract application bianry image, and application bianry image is extracted in the filtering of E. morphological images, F.
Profile, G. screening profiles, the profile point that H. outputs are represented with pixel coordinate.
2. the vessel coating profile testing method based on image segmentation as claimed in claim 1, it is characterised in that the B. figures
As pretreatment, input picture is switched to into single pass gray level image, medium filtering is made to gray level image then.
3. the vessel coating profile testing method based on image segmentation as claimed in claim 1, it is characterised in that the C. figures
As segmentation, multi-threshold segmentation is carried out to image according to the quantity of coating.
4. the vessel coating profile testing method based on image segmentation as claimed in claim 1, it is characterised in that the D. is carried
Application bianry image is taken, a bianry image is extracted in each layer application region respectively.
5. the vessel coating profile testing method based on image segmentation as claimed in claim 1, it is characterised in that the E. figures
As morphologic filtering, a structural elements image is corroded or expanded；If first corroded to a structural elements image, so
Image is expanded with identical structural elements again afterwards, make opening operation；, whereas if first being carried out to image with a structural elements swollen
It is swollen, then image is corroded with identical structural elements again, make closed operation；Corrode the image less than or equal to structural elements thin
Section is filtered from image.
The E. morphological images filtering, makees opening operation smoothed profile, eliminates prominent by a structural elements image to bianry image
Go out thing；Make closed operation to eliminate cavity, fill up fracture, so as to eliminate picture noise.
6. the vessel coating profile testing method based on image segmentation as claimed in claim 1, it is characterised in that the F. is carried
Application bianry image profile is taken, and is corroded bianry image using a structural elements, profile is extracted from the application bianry image, then
Continuous profile is obtained with the image after the corrosion of artwork image subtraction.
7. the vessel coating profile testing method based on image segmentation as claimed in claim 1, it is characterised in that the G. sieves
Profile is selected, according to the size screening profile of outlines and contour area and mean pixel coordinate.
8. the vessel coating profile testing method based on image segmentation as claimed in claim 2, it is characterised in that the B. figures
As the medium filtering of pretreatment, processes pixel is carried out to image, when each pixel is processed, it is considered to certain area around the pixel
Pixel in domain, sorts them according to the size of gray value, chooses the intermediate value of gray value replacing the gray value of current pixel.
9. the vessel coating profile testing method based on image segmentation as claimed in claim 3, it is characterised in that the C. figures
As segmentation, using multi thresholds Otsu method segmentation figure pictures, and accelerate solution procedure using PSO algorithms, quickly obtain image segmentation
Optimal threshold combination；
The extraction bianry image, optimum segmentation threshold value combination determine some gray value intervals, for each gray value area
Between, definition one is regular, if certain grey scale pixel value in pretreated image is in the interval, the gray scale of the pixel
Value is set as 255；Conversely, being set to 0, store the result in a new image, thus obtained some bianry images, often
A cut zone is included in individual bianry image.
10. the vessel coating profile testing method based on image segmentation as claimed in claim 4, it is characterised in that the G.
Contours extract and screening, it is assumed that the area of unrelated profile is all smaller relative to coating profile, according to this it is assumed that comparing detection
To the area of profile, the little profile of removal Area comparison, the maximum front K profile of Retention area, K is the number of profile to be detected
Amount, then according to retain profile area and mean pixel coordinate come judge their position and opsition dependent sequence, sort
Profile afterwards is returned as the result of algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611179450.7A CN106600606A (en) | 2016-12-19 | 2016-12-19 | Ship painting profile detection method based on image segmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611179450.7A CN106600606A (en) | 2016-12-19 | 2016-12-19 | Ship painting profile detection method based on image segmentation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106600606A true CN106600606A (en) | 2017-04-26 |
Family
ID=58601718
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611179450.7A Pending CN106600606A (en) | 2016-12-19 | 2016-12-19 | Ship painting profile detection method based on image segmentation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106600606A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108230388A (en) * | 2018-02-06 | 2018-06-29 | 广西艾盛创制科技有限公司 | A kind of recognition positioning method of white body weld point image |
| CN108346140A (en) * | 2018-01-10 | 2018-07-31 | 哈尔滨理工大学 | Based on the Otsu lung images dividing methods for improving PSO |
| CN110544009A (en) * | 2019-07-26 | 2019-12-06 | 中国人民解放军海军航空大学青岛校区 | Aviation organic coating aging damage quantitative evaluation method based on digital image processing |
| CN114155314A (en) * | 2021-11-25 | 2022-03-08 | 航天科工深圳（集团）有限公司 | Intelligent wall painting method, system, equipment and storage medium based on image recognition and character recognition |
| CN114862082A (en) * | 2021-02-04 | 2022-08-05 | 敬鹏工业股份有限公司 | Intelligent logistics data acquisition system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2399629A (en) * | 2003-01-22 | 2004-09-22 | Univ Brunel | Automatic thresholding algorithm method and apparatus |
| CN103106408A (en) * | 2013-01-25 | 2013-05-15 | 西安电子科技大学 | Stomach computer tomography (CT) image lymph node automatic auxiliary detecting system without supervision segmentation |
-
2016
- 2016-12-19 CN CN201611179450.7A patent/CN106600606A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2399629A (en) * | 2003-01-22 | 2004-09-22 | Univ Brunel | Automatic thresholding algorithm method and apparatus |
| CN103106408A (en) * | 2013-01-25 | 2013-05-15 | 西安电子科技大学 | Stomach computer tomography (CT) image lymph node automatic auxiliary detecting system without supervision segmentation |
Non-Patent Citations (2)
| Title |
|---|
| ZHENHUA ZHANG 等: ""A novel image segmentation method combined Otsu and improved PSO", 《2012 IEEE FIFTH INTERNATIONAL CONFERENCE ON ADVANCED COMPUTATIONAL INTELLIGENCE》 * |
| 白伟志 等: "基于图像分割的船舶涂装轮廓检测方法", 《机电一体化》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108346140A (en) * | 2018-01-10 | 2018-07-31 | 哈尔滨理工大学 | Based on the Otsu lung images dividing methods for improving PSO |
| CN108230388A (en) * | 2018-02-06 | 2018-06-29 | 广西艾盛创制科技有限公司 | A kind of recognition positioning method of white body weld point image |
| CN108230388B (en) * | 2018-02-06 | 2021-04-23 | 广西艾盛创制科技有限公司 | Recognition and positioning method for welding spot image of white car body |
| CN110544009A (en) * | 2019-07-26 | 2019-12-06 | 中国人民解放军海军航空大学青岛校区 | Aviation organic coating aging damage quantitative evaluation method based on digital image processing |
| CN110544009B (en) * | 2019-07-26 | 2022-12-09 | 中国人民解放军海军航空大学青岛校区 | Aviation organic coating aging damage quantitative evaluation method based on digital image processing |
| CN114862082A (en) * | 2021-02-04 | 2022-08-05 | 敬鹏工业股份有限公司 | Intelligent logistics data acquisition system |
| CN114155314A (en) * | 2021-11-25 | 2022-03-08 | 航天科工深圳（集团）有限公司 | Intelligent wall painting method, system, equipment and storage medium based on image recognition and character recognition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106600606A (en) | Ship painting profile detection method based on image segmentation | |
| CN109711474B (en) | Aluminum product surface defect detection algorithm based on deep learning | |
| CN109191459B (en) | Automatic identification and rating method for continuous casting billet macrostructure center segregation defect | |
| CN110363781B (en) | Molten pool contour detection method based on deep neural network | |
| CN108846844B (en) | Sea surface target detection method based on sea antenna | |
| CN111681249B (en) | Grabcut-based improved segmentation algorithm research of sand particles | |
| CN103871062B (en) | A kind of lunar surface rock detection method described based on super-pixel | |
| CN109461141A (en) | A kind of workpiece starved detection method | |
| CN110660040A (en) | Industrial product irregular defect detection method based on deep learning | |
| CN115100174B (en) | Ship sheet metal part paint surface defect detection method | |
| CN103413314A (en) | Textile defect detecting algorithm based on texture gradients | |
| CN107154044B (en) | Chinese food image segmentation method | |
| CN113506246B (en) | Concrete 3D printing component fine detection method based on machine vision | |
| CN103325118A (en) | Method and device for acquiring characteristic parameters of core hole of carbonatite | |
| CN111582004A (en) | Target area segmentation method and device in ground image | |
| CN107993219A (en) | A kind of deck of boat detection method of surface flaw based on machine vision | |
| US11710158B2 (en) | Vehicle damage estimation | |
| CN114743102A (en) | Furniture board oriented flaw detection method, system and device | |
| Sanghadiya et al. | Surface defect detection in a tile using digital image processing: Analysis and evaluation | |
| CN107871315B (en) | Video image motion detection method and device | |
| CN115147401A (en) | Digital image processing-based intelligent detection method for dam material qualification of earth and rockfill dam | |
| CN107590801A (en) | A kind of image processing method split and graded for beef eye muscle marbling | |
| CN109558877A (en) | Naval target track algorithm based on KCF | |
| CN107967690B (en) | Self-adaptive ferrographic abrasive particle image binarization processing method | |
| Singhka et al. | Surface defect classification of steel strip based on machine vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170426 |