CN106326901A - Water stain image recognition based on edge point self-similarity and TEDS system - Google Patents
Water stain image recognition based on edge point self-similarity and TEDS system Download PDFInfo
- Publication number
- CN106326901A CN106326901A CN201610752438.4A CN201610752438A CN106326901A CN 106326901 A CN106326901 A CN 106326901A CN 201610752438 A CN201610752438 A CN 201610752438A CN 106326901 A CN106326901 A CN 106326901A
- Authority
- CN
- China
- Prior art keywords
- marginal point
- self
- point
- similarity
- edge
- 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.)
- Granted
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
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06V10/443—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 by matching or filtering
Abstract
The invention provides water stain image recognition based on edge point self-similarity, comprising the steps of acquiring edge points of an image to be detected, allocating a reference direction for each edge point and calculating a feature vector thereof, and normalizing the feature vectors; calculating partial and global self-similarity values of each edge point on edge lines, and performing weighted combination on the partial and global self-similarity values to obtain a final self-similarity value of the edge point; acquiring a set of edge points with high self-similarity on the edge lines, and screening and marking irregular edge points with low self-similarity in the set according to a predetermined mode; collecting statistics on directional distribution histograms of the irregular edge points, determining the main direction of each irregular edge point, presetting a downward directional threshold range, removing the irregular edge points of which the main directions are not within the directional threshold range, and marking the irregular edge points of which the main directions are within the directional threshold range as a water stain image. The invention further provides a TEDS system using the water stain image recognition, which effectively recognizes water stains on motor train units and reduces the misjudgment rate.
Description
Technical field
The present invention relates to computer picture detection identification field, particularly relate to a kind of based on marginal point self-similarity water stain
Image recognition and apply the TEDS system of this identification.
Background technology
Water stain image is typically a kind of interferogram picture in field of image recognition, on the impact of the identification of specified features relatively
Little, therefore, at present not the most concrete for water stain knowledge method for distinguishing.
EMUs fault detect is transformed into the analysis to vehicle image from on-the-spot manual detection, utilizes EMUs to run
EMUs driving conditions is monitored by fault dynamic images detecting system (TEDS) in real time, i.e. TEDS installs limit and the flange of rail in-orbit
Position real-time image acquisition to be detected to the EMUs of walking.TEDS detection event EMUs barrier mainly utilizes comparison in difference
Method, a kind of mode of comparison in difference method is the unfaulty conditions during off-duty by storage in the real time imaging gathered and image library
The standard picture of EMUs carry out the contrast of characteristics of image；Another way is to be deposited in image library by the real time imaging of collection
The history image of the EMUs of the recent unfaulty conditions of storage carries out the comparison of characteristics of image, by feature difference in two ways
Significantly place is labeled as failure exception.
Reference picture Precision criterion in above-mentioned standard picture relative method, it is possible to realize accurately sentencing of present image difference
Fixed, but affected by factor water stain etc. on overhaul of train-set maintenance, natural aging, car, the method is easily normal by car body
Change is mistaken for fault, improves False Rate；History image relative method can effectively reduce the fault that car body normal variation is brought
Erroneous judgement problem, is collection in worksite on the spot yet with history image, is affected by Along Railway complex environment, reference picture is inadequate
Accurately, also there is the erroneous judgement of obvious fault.
Application standard picture relative method, if the impact of water stain factor on EMUs locomotive can be overcome, then can reduce fault
Erroneous judgement, improves the accuracy rate of fault detect.
Summary of the invention
The present invention proposes a kind of water stain image recognition based on marginal point self-similarity, it is possible to by the water stain image on motor-car
Identify, solve the problem that in prior art Plays image relative method, fault erroneous judgement is high.
The technical scheme is that and be achieved in that:
A kind of water stain image recognition based on marginal point self-similarity, comprises the following steps:
Step one: input image to be detected in a computer, utilizes canny edge detection algorithm to obtain all limits of this image
Edge point；
Step 2: classify all marginal points, similar marginal point belongs to an initial edge line of an image outline, obtains
Take all of initial edge line of image to be detected, distribute a reference direction to each marginal point, and extract each edge
The characteristic vector of point, and each characteristic vector is normalized；
Step 3: calculate each marginal point on every initial edge line according to the characteristic vector after each marginal point normalized
Local self-similarity value and overall self-similarity value, and by local self-similarity value and the weighted array of overall self-similarity value
As the self-similarity value that this marginal point is final；
Step 4: set a high threshold, obtains the marginal point higher than high threshold of the self-similarity value on every initial edge line
Set, reject the marginal point less than high threshold of self-similarity value on every initial edge line；
Step 5: calculate the self-similarity value of each marginal point marginal point closest with it in above-mentioned set, sets a low threshold
Value, obtains all of self-similarity value in set and, higher than the marginal point of Low threshold, will be less than the marginal point of Low threshold in this set
It is labeled as being formed the broken edge point of irregular image；
Step 6: classifying the marginal point being higher than Low threshold in step 5, same class marginal point forms the school of an image outline
Positive edge line, sets a length threshold to all of calibration edge line, obtains the calibration edge line less than this threshold value, by this school
Point on positive edge line is labeled as being formed the broken edge point of irregular image；
The directional spreding rectangular histogram of the broken edge point in step 7, statistic procedure five and step 6, determine each irregularly
The principal direction of marginal point, presets a downwardly direction threshold range, rejects all principal direction not in the threshold range of direction
Broken edge point, is labeled as water stain image by principal direction broken edge point in the threshold range of direction.
Preferably, in described based on marginal point self-similarity water stain image recognition, according to every in described step 3
Characteristic vector after individual marginal point normalized calculates the final self-similarity value of the marginal point on every initial edge line
Mode is: setWithFor any two marginal point on edge line, its characteristic vector is respectivelyWith, then
Any two marginal point on edge lineWithSimilarity be:, here in vector
Long-pending calculation is multiplied for vector corresponding element and is added, obtains two marginal pointsWithSimilarity；
Marginal pointLocal self-similarity value be: take this marginal pointAnd position adjacent with this marginal point on the edge line at place
Four marginal points in its both sides、、、, then marginal pointLocal self-similarity value be:
；
Marginal pointOverall self-similarity value be: assuming that total n marginal point, then marginal point on this edge lineEntirety
Self-similarity value is:
；
Set the weights of local self-similarity and overall self-similarity as, and,, by local self-similarity value and overall self-similarity value combination, then marginal pointFinal self-similarity
Value is:
；
Characteristic vector after normalizedScope between 0 to 1, then self-similarity value represents similar between 0 to 1
Degree, self-similarity value is to be the most dissimilar state when 0, is to be complete similar state when 1.
Preferably, in described based on marginal point self-similarity water stain image recognition, described step 2 is given each
The mode of individual marginal point one reference direction of distribution is:
For any one marginal point, structure local neighborhood centered by current edge point, all pixels in calculating this neighborhood
Grad and direction, utilize Grad and the direction of all pixels in this neighborhood of statistics with histogram, comprise 0 in rectangular histogram
9 Nogata posts that the direction scope of ~ 180 degree is divided equally, divide equally and are incorporated on 9 Nogata posts for 180 ~ 360 degree；
Calculate each marginal point weight coefficient to adjacent both direction, calculate each edge further according to weight coefficient and Grad
Contribution weights are added to histogrammic each Nogata post at this marginal point place by the some contribution weights to adjacent both direction
On, direction, histogram peak place is the reference direction of this marginal point.
Preferably, in described based on marginal point self-similarity water stain image recognition, in described step 2 any one
The extracting mode of the characteristic vector of marginal point is:
Set any edge point as, the reference direction of this marginal point is, coordinate axes is rotated to reference direction；After rotation
Coordinate system in take distance marginal point respectively along four orientationThe point of predetermined location of pixels、、、, construct with、、、、Centered by 5 local neighborhood, calculate the Grad of each pixel and each pixel to adjacent two
The contribution weights in individual direction；The directional spreding rectangular histogram of 5 local neighborhood of statistics, obtains 5 rectangular histograms；The characteristic vector of this marginal point is:；The finally feature to each marginal point
Vector is normalized.
Preferably, in described based on marginal point self-similarity water stain image recognition, each broken edge point
The determination mode of principal direction is:
For any one broken edge point, the reference direction of this marginal point is, coordinate axes is rotated to reference direction, structure
Make with current edge pointCentered by local neighborhood, the direction of all pixels in calculating this neighborhood；Build one by 0 ~ 180
The direction scope of degree is divided into the bar diagram of 18 Nogata posts, and 180 ~ 360 degree are uniformly distributed and are merged on 18 Nogata posts, system
Count the pixel number in the range of each Nogata post direction in 18 Nogata posts；Finally add up the bar diagram peak value place side obtained
To the principal direction for current edge point.
A kind of TEDS system, including the water stain image recognition based on marginal point self-similarity of any of the above-described item, by labelling
Water stain image is defaulted as the normal condition of EMUs, the not labelling when EMUs fault detect.
The invention have the benefit that in the present invention, first consider every initial edge line, calculate every according to characteristic vector
The self-similarity of the marginal point on edge line, is obtained from similar bigger marginal point set, rejects the marginal point that self-similarity is little；
Recalculate the self-similarity of each marginal point again according to predetermined mode in all marginal point set that self-similarity is bigger,
It is obtained from the similarity marginal point set higher than Low threshold, increases the seriality of marginal point；To set is higher than Low threshold
Marginal point reclassifies, and obtains calibration edge line；Correction edge line length is low less than being less than in the point of length threshold and set
The point of threshold value is labeled as broken edge point；According to the water stain characteristic having, water stain line orientations substantially downwardly, thus is never advised
Image then identifies water stain image.Apply the method in TEDS system, can effectively identify the water stain image on car, reduce
EMUs fault False Rate, improves the accuracy of TEDS system detection.
Accompanying drawing explanation
For the technical scheme being illustrated more clearly that in the embodiment of the present invention, in embodiment being described below required for make
Accompanying drawing be briefly described, it should be apparent that, below describe in accompanying drawing be only some embodiments of the present invention, for
From the point of view of those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to obtain other according to embodiment
Accompanying drawing.
Fig. 1 is the image at a certain position of the existing EMUs gathered；
Fig. 2 is for identifying water stain image in FIG.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Describe, it is clear that described embodiment is only a part of embodiment of the present invention rather than whole embodiments wholely.Based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under not making creative work premise
Embodiment, broadly falls into the scope of protection of the invention.
Embodiment: a kind of based on marginal point self-similarity water stain image recognition applied in TEDS system, including with
Lower step:
Step one: input EMUs image to be detected in a computer, as it is shown in figure 1, utilize canny edge detection algorithm to obtain
Take all marginal points of this image；Detailed process is as follows:
1, it is gray level image by EMUs image procossing on computers；
2, gray level image is carried out Gaussian Blur to reduce the interference of picture noise；
3, Grad and the direction of each pixel in the image after noise reduction are calculated；
4, the Grad to each pixel carries out non-maxima suppression, tentatively obtains image border point set；5, dual threashold is used
Value method carries out edge connection, rejects false edge, completion emargintion, it is thus achieved that more accurate marginal point set.
Step 2: classify all marginal points, similar marginal point belongs to an initial edge of an image outline
Line, obtains all of initial edge line of image to be detected, distributes a reference direction to each marginal point, and extracts every
The characteristic vector of individual marginal point, and the characteristic vector of each marginal point is normalized；
The mode of each marginal point one reference direction of distribution is:
1, for any one marginal point, structure 8*8 neighborhood centered by current edge point, all pixels in calculating this neighborhood
Grad and direction, utilize Grad and the direction of all pixels in this neighborhood of statistics with histogram, comprise 0 in rectangular histogram
9 Nogata posts that the direction scope of ~ 180 degree is divided equally, 20 degree of each Nogata post, divide equally and be incorporated in 9 Nogata posts for 180 ~ 360 degree
On；
2, calculate each marginal point weight coefficient to adjacent both direction, calculate each limit further according to weight coefficient and Grad
Contribution weights are added to histogrammic each Nogata post at this marginal point place by the edge point contribution weights to adjacent both direction
On, direction, histogram peak place is the reference direction of this marginal point.
The extracting mode of the characteristic vector of each marginal point is:
1, set any edge point as, the reference direction of this marginal point is, coordinate axes being rotated to reference direction, coordinate becomes
It is changed to；Coordinate system after rotation takes distance marginal point respectively along four orientationIn advance
The point of fixed location of pixels、、、, construct with、、、、Centered by 5 8*8 neighborhoods, calculate each picture
The Grad of vegetarian refreshments, calculate each pixel to adjacent both direction
Contribution weights；
2, add up the directional spreding rectangular histogram of 5 8*8 neighborhoods, obtain 5 rectangular histograms
；The characteristic vector of this marginal point is:；Finally
The characteristic vector of each marginal point is normalized.
Step 3: calculate each limit on every initial edge line according to the characteristic vector after each marginal point normalized
The local self-similarity value of edge point and overall self-similarity value, and by local self-similarity value and the weighting of overall self-similarity value
Combine the self-similarity value final as marginal point；
The calculation of the final self-similarity value of the marginal point on every initial edge line is: setWithFor edge
Any two marginal point on line, its characteristic vector is respectivelyWith, then
Any two marginal point on edge lineWithSimilarity be:；Here in vector
Long-pending calculation is multiplied for vector corresponding element and is added, obtains two marginal pointsWithSimilarity；
Marginal pointLocal self-similarity value be: take this marginal pointAnd position adjacent with this marginal point on the edge line at place
Four marginal points in its both sides、、、, then marginal pointLocal self-similarity value be:
；
Marginal pointOverall self-similarity value be: assuming that total n marginal point, then marginal point on this edge lineEntirety
Self-similarity value is:
；
Set the weights of local self-similarity and overall self-similarity as, choose,, by local self-similarity value and overall self-similarity value combination, then marginal pointFinal self-similarity
Value is:
；
Characteristic vector after normalizedScope between 0 to 1, then self-similarity value represents similar between 0 to 1
Degree, self-similarity value is to be the most dissimilar state when 0, is to be complete similar state when 1.
Step 4: high threshold is set to 0.7, obtains the edge higher than 0.7 of the self-similarity value on every initial edge line
The set of point, rejects self-similarity value on every initial edge line and is less than the marginal point of high threshold.
Step 5: the self-similarity of any two marginal point in the set of calculation procedure four, is set to 0.2 by Low threshold,
Obtain all of self-similarity value marginal point higher than 0.2 in set, be labeled as being formed by the marginal point being less than 0.2 in this set
The broken edge point of irregular image.
Step 6: classifying the marginal point being higher than Low threshold in step 5, same class marginal point forms an image outline
Calibration edge line, to all of calibration edge line preseting length threshold value 10, obtain the edge line length calibration edge less than 10
Line, is labeled as being formed the broken edge point of irregular image by the point on this calibration edge line.
The directional spreding rectangular histogram of the broken edge point in step 7, statistic procedure five and step 6, determine each not
The principal direction of rule marginal point, presets downwardly direction threshold range 70 ~ 110 degree, rejects all principal direction not at 70 ~ 110 degree
Interior broken edge point, is labeled as water stain image by principal direction broken edge point in 70 ~ 110 degree, this principal direction
Broken edge point is as shown in Figure 2；
The determination mode of the principal direction of each broken edge point is: for any one broken edge point, this marginal point
Reference direction is, coordinate axes is rotated to reference direction, structure is with current edge pointCentered by 8*8 neighborhood, calculate should
The direction of all pixels in neighborhood；Build a bar diagram that the direction scope of 0 ~ 180 degree is divided into 18 Nogata posts,
180 ~ 360 degree are uniformly distributed and are merged on 18 Nogata posts, and each Nogata post is 10 degree, add up in 18 Nogata posts each directly
Pixel number in the range of square column direction；Finally add up the master that direction, peak value place is current edge point of the bar diagram obtained
Direction.
Step 8, the reference picture of the EMUs to be detected extracted in image library, application standard picture method is in TEDS system
Image to be detected and reference picture are compared by system, the water stain image identified in step 7 is defaulted as external interference
Factor, is not the malfunction of EMUs, and when EMUs fault detect, not labelling, reduction fault erroneous judgement, improves the event obtained
The degree of accuracy of barrier detection figure.
The neighborhood of above-mentioned appearance selects as the case may be, it is possible to elect other neighborhoods such as 8*16 as.Above-mentioned high threshold, low
Threshold value, length threshold and direction threshold range can be chosen according to the type of actually detected image.
The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all essences in the present invention
Within god and principle, any modification, equivalent substitution and improvement etc. made, should be included within the scope of the present invention.
Claims (6)
1. a water stain image recognition based on marginal point self-similarity, it is characterised in that comprise the following steps:
Step one: input image to be detected in a computer, utilizes canny edge detection algorithm to obtain all limits of this image
Edge point；
Step 2: classify all marginal points, similar marginal point belongs to an initial edge line of an image outline, obtains
Take all of initial edge line of image to be detected, distribute a reference direction to each marginal point, and extract each edge
The characteristic vector of point, and each characteristic vector is normalized；
Step 3: calculate each marginal point on every initial edge line according to the characteristic vector after each marginal point normalized
Local self-similarity value and overall self-similarity value, and by local self-similarity value and the weighted array of overall self-similarity value
As the self-similarity value that this marginal point is final；
Step 4: set a high threshold, obtains the marginal point higher than high threshold of the self-similarity value on every initial edge line
Set, reject the marginal point less than high threshold of self-similarity value on every initial edge line；
Step 5: calculate the self-similarity value of each marginal point marginal point closest with it in above-mentioned set, sets a low threshold
Value, obtains all of self-similarity value in set and, higher than the marginal point of Low threshold, will be less than the marginal point of Low threshold in this set
It is labeled as being formed the broken edge point of irregular image；
Step 6: classifying the marginal point being higher than Low threshold in step 5, same class marginal point forms the school of an image outline
Positive edge line, sets a length threshold to all of calibration edge line, obtains the calibration edge line less than this threshold value, by this school
Point on positive edge line is labeled as being formed the broken edge point of irregular image；
The directional spreding rectangular histogram of the broken edge point in step 7, statistic procedure five and step 6, determine each irregularly
The principal direction of marginal point, presets a downwardly direction threshold range, rejects all principal direction not in the threshold range of direction
Broken edge point, is labeled as water stain image by principal direction broken edge point in the threshold range of direction.
Water stain image recognition based on marginal point self-similarity the most according to claim 1, it is characterised in that described step
In three according to the characteristic vector after each marginal point normalized calculate the marginal point on every initial edge line final from
The mode of similarity is: setWithFor any two marginal point on edge line, its characteristic vector is respectivelyWith, then
Any two marginal point on edge lineWithSimilarity be:, here in vector
Long-pending calculation is multiplied for vector corresponding element and is added, obtains two marginal pointsWithSimilarity；
Marginal pointLocal self-similarity value be: take this marginal pointOn the edge line at place adjacent with this marginal point and be positioned at
Four marginal points of its both sides、、、, then marginal pointLocal self-similarity value be:
；
Marginal pointOverall self-similarity value be: assuming that total n marginal point, then marginal point on this edge lineEntirety from
Similarity is:
；
Set the weights of local self-similarity and overall self-similarity as, and,, by local self-similarity value and overall self-similarity value combination, then marginal pointFinal self-similarity
Value is:
；
Characteristic vector after normalizedScope between 0 to 1, then self-similarity value represents similar journey between 0 to 1
Degree, self-similarity value is to be the most dissimilar state when 0, is to be complete similar state when 1.
Water stain image recognition based on marginal point self-similarity the most according to claim 1, it is characterised in that described step
The mode distributing a reference direction to each marginal point in two is:
For any one marginal point, structure local neighborhood centered by current edge point, all pixels in calculating this neighborhood
Grad and direction, utilize Grad and the direction of all pixels in this neighborhood of statistics with histogram, comprise 0 in rectangular histogram
9 Nogata posts that the direction scope of ~ 180 degree is divided equally, divide equally and are incorporated on 9 Nogata posts for 180 ~ 360 degree；
Calculate each marginal point weight coefficient to adjacent both direction, calculate each edge further according to weight coefficient and Grad
Contribution weights are added to histogrammic each Nogata post at this marginal point place by the some contribution weights to adjacent both direction
On, direction, histogram peak place is the reference direction of this marginal point.
Water stain image recognition based on marginal point self-similarity the most according to claim 1, it is characterised in that described step
In two, the extracting mode of the characteristic vector of any one marginal point is:
Set any edge point as, the reference direction of this marginal point is, coordinate axes is rotated to reference direction；After rotation
Coordinate system in take distance marginal point respectively along four orientationThe point of predetermined location of pixels、、、, construct with、、、、Centered by 5 local neighborhood, calculate the Grad of each pixel and each pixel to adjacent two
The contribution weights in individual direction；The directional spreding rectangular histogram of 5 local neighborhood of statistics, obtains 5 rectangular histograms；The characteristic vector of this marginal point is:；The finally feature to each marginal point
Vector is normalized.
Water stain image recognition based on marginal point self-similarity the most according to claim 1, it is characterised in that each do not advise
Then the determination mode of the principal direction of marginal point is:
For any one broken edge point, the reference direction of this marginal point is, coordinate axes is rotated to reference direction, structure
Make with current edge pointCentered by local neighborhood, the direction of all pixels in calculating this neighborhood；Build one by 0 ~ 180
The direction scope of degree is divided into the bar diagram of 18 Nogata posts, and 180 ~ 360 degree are uniformly distributed and are merged on 18 Nogata posts, system
Count the pixel number in the range of each Nogata post direction in 18 Nogata posts；Finally add up the bar diagram peak value place side obtained
To the principal direction for current edge point.
6. a TEDS system, it is characterised in that include that any one described in claim 1 to 5 is based on marginal point self-similarity
Water stain image recognition, the water stain image of labelling is defaulted as the normal condition of EMUs, does not marks when EMUs fault detect
Note.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610752438.4A CN106326901B (en) | 2016-08-30 | 2016-08-30 | Water stain image-recognizing method and TEDS system based on marginal point self-similarity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610752438.4A CN106326901B (en) | 2016-08-30 | 2016-08-30 | Water stain image-recognizing method and TEDS system based on marginal point self-similarity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106326901A true CN106326901A (en) | 2017-01-11 |
| CN106326901B CN106326901B (en) | 2019-06-14 |
Family
ID=57789936
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610752438.4A Active CN106326901B (en) | 2016-08-30 | 2016-08-30 | Water stain image-recognizing method and TEDS system based on marginal point self-similarity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106326901B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109063764A (en) * | 2018-07-26 | 2018-12-21 | 福建和盛高科技产业有限公司 | A kind of judgment method of disconnecting switch closing operation in place based on machine vision |
| CN110827220A (en) * | 2019-11-01 | 2020-02-21 | 杭州当虹科技股份有限公司 | Anti-aliasing method based on image edge analysis |
| US11023770B2 (en) | 2019-09-23 | 2021-06-01 | Hong Kong Applied Science And Technology Research Institute Co., Ltd. | Systems and methods for obtaining templates for tessellated images |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1107179A2 (en) * | 1999-11-29 | 2001-06-13 | Eastman Kodak Company | Method for detecting sky in images |
| CN101789005A (en) * | 2010-01-22 | 2010-07-28 | 深圳创维数字技术股份有限公司 | Image searching method based on region of interest (ROI) |
| CN102589808A (en) * | 2012-01-16 | 2012-07-18 | 苏州临点三维科技有限公司 | Large-scale tunnel seepage point measuring method |
| CN105551035A (en) * | 2015-12-09 | 2016-05-04 | 深圳市华和瑞智科技有限公司 | Stereoscopic vision matching method based on weak edge and texture classification |
| CN105844278A (en) * | 2016-04-15 | 2016-08-10 | 浙江理工大学 | Multi-feature fused fabric scanning pattern recognition method |
-
2016
- 2016-08-30 CN CN201610752438.4A patent/CN106326901B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1107179A2 (en) * | 1999-11-29 | 2001-06-13 | Eastman Kodak Company | Method for detecting sky in images |
| CN101789005A (en) * | 2010-01-22 | 2010-07-28 | 深圳创维数字技术股份有限公司 | Image searching method based on region of interest (ROI) |
| CN102589808A (en) * | 2012-01-16 | 2012-07-18 | 苏州临点三维科技有限公司 | Large-scale tunnel seepage point measuring method |
| CN105551035A (en) * | 2015-12-09 | 2016-05-04 | 深圳市华和瑞智科技有限公司 | Stereoscopic vision matching method based on weak edge and texture classification |
| CN105844278A (en) * | 2016-04-15 | 2016-08-10 | 浙江理工大学 | Multi-feature fused fabric scanning pattern recognition method |
Non-Patent Citations (1)
| Title |
|---|
| 张正炳 等；: "《图像的局部自相似性用于边缘检测的方法研究》", 《华中理工大学学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109063764A (en) * | 2018-07-26 | 2018-12-21 | 福建和盛高科技产业有限公司 | A kind of judgment method of disconnecting switch closing operation in place based on machine vision |
| US11023770B2 (en) | 2019-09-23 | 2021-06-01 | Hong Kong Applied Science And Technology Research Institute Co., Ltd. | Systems and methods for obtaining templates for tessellated images |
| CN110827220A (en) * | 2019-11-01 | 2020-02-21 | 杭州当虹科技股份有限公司 | Anti-aliasing method based on image edge analysis |
| CN110827220B (en) * | 2019-11-01 | 2023-05-30 | 杭州当虹科技股份有限公司 | Anti-aliasing method based on image edge analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106326901B (en) | 2019-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106373426B (en) | Parking stall based on computer vision and violation road occupation for parking monitoring method | |
| CN106683076B (en) | The method of train wheel tread damage detection based on textural characteristics cluster | |
| CN106682586A (en) | Method for real-time lane line detection based on vision under complex lighting conditions | |
| US10943131B2 (en) | Image based lane marking classification | |
| CN107025432B (en) | A kind of efficient lane detection tracking and system | |
| CN103268471B (en) | A kind of vehicle illegal road occupying detection method and device | |
| CN110047070B (en) | Method and system for identifying rail wear degree | |
| CN107273802A (en) | A kind of detection method and device of railroad train brake shoe drill ring failure | |
| CN106326901A (en) | Water stain image recognition based on edge point self-similarity and TEDS system | |
| CN104933398A (en) | vehicle identification system and method | |
| CN106548147A (en) | A kind of quick noise robustness image foreign matter detection method and TEDS systems | |
| CN109635737A (en) | Automobile navigation localization method is assisted based on pavement marker line visual identity | |
| CN110633671A (en) | Bus passenger flow real-time statistical method based on depth image | |
| CN105354533A (en) | Bag-of-word model based vehicle type identification method for unlicensed vehicle at gate | |
| Premachandra et al. | Image based automatic road surface crack detection for achieving smooth driving on deformed roads | |
| Wang et al. | License plate location algorithm based on edge detection and morphology | |
| CN111753749A (en) | Lane line detection method based on feature matching | |
| CN109558765A (en) | The detection method of automobile and lane line, device | |
| CN114494373A (en) | High-precision rail alignment method and system based on target detection and image registration | |
| CN106327499A (en) | Oil stain image recognition based on edge point self-similarity and TEDS system | |
| KR101205565B1 (en) | Method for Dectecting Front and Rear Vehicle by Using Image | |
| US20230342937A1 (en) | Vehicle image analysis | |
| CN106384358B (en) | The recognition methods of irregular image based on marginal point self-similarity | |
| CN107545216A (en) | A kind of vehicle identification method available for power-line patrolling | |
| CN106651890B (en) | Metal reflective image recognition and TEDS system based on marginal point self-similarity |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |