CN109523479A - A kind of bridge pier surface gaps visible detection method - Google Patents
A kind of bridge pier surface gaps visible detection method Download PDFInfo
- Publication number
- CN109523479A CN109523479A CN201811334774.2A CN201811334774A CN109523479A CN 109523479 A CN109523479 A CN 109523479A CN 201811334774 A CN201811334774 A CN 201811334774A CN 109523479 A CN109523479 A CN 109523479A
- Authority
- CN
- China
- Prior art keywords
- image
- bridge pier
- region
- area
- detection method
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 32
- 230000009466 transformation Effects 0.000 claims abstract description 33
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000001914 filtration Methods 0.000 claims abstract description 28
- 230000011218 segmentation Effects 0.000 claims abstract description 28
- 230000000877 morphologic effect Effects 0.000 claims abstract description 25
- 238000012545 processing Methods 0.000 claims abstract description 18
- 238000003708 edge detection Methods 0.000 claims abstract description 14
- 239000003550 marker Substances 0.000 claims abstract description 11
- 238000000638 solvent extraction Methods 0.000 claims abstract description 10
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 9
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 9
- 238000009499 grossing Methods 0.000 claims abstract description 7
- 238000011946 reduction process Methods 0.000 claims abstract description 6
- 238000006073 displacement reaction Methods 0.000 claims description 15
- 239000000284 extract Substances 0.000 claims description 8
- 230000002708 enhancing effect Effects 0.000 claims description 6
- 230000001965 increasing effect Effects 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 4
- 238000000926 separation method Methods 0.000 claims description 2
- 238000013316 zoning Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 description 12
- 238000005516 engineering process Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 208000037656 Respiratory Sounds Diseases 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000003709 image segmentation Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000006740 morphological transformation Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000005352 clarification Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- 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
- G06T2207/20064—Wavelet transform [DWT]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of bridge pier surface gaps visible detection methods, obtain bridge pier surface color image by linear array CCD camera;Color image splits' positions are stored, color image are converted to gray level image, and carry out noise reduction process;Using the linear character in bridge pier gap, image enhancement smoothing processing is carried out to gray level image using image masks filtering algorithm;The section of region A and region B are calculated separately using Furthest Neighbor in section, area, distance in area, make that the section of region A and region B, distance ratio reaches maximum in area by changing parameter value, synthesis segmentation threshold of the threshold value result at this time as bridge pier slot image, is detected using slot edge of the wavelet transformation gradient image partitioning algorithm to bridge pier image;The slot edge that bridge pier image is finally extracted using Multiscale Morphological Image Edge-Detection method can be realized the efficient identification detection to bridge pier gap with the gap in the Morphological watersheds algorithm identification bridge pier image of marker character control.
Description
Technical field
The present invention relates to image detection identification technology, in particular to a kind of bridge pier surface gaps visible detection method.
Background technique
With the continuous development of society, nowadays there has also been very big development for the construction of Bridges in Our Country engineering, but due to
Done in the design of bridge, construction period and maintenance processes in the future it is not in place, in the influence plus natural environment, bridge pier clothes
The length of time of labour damages bridge pier structure gradually;The structure gap on bridge pier surface reflects bridge pier with regard to apparent
Hurt degree, if bridge pier gap problem cannot obtain solution substantially, this will come very big to the safety belt of bridge pier engineering
Hidden danger.
In many bridge pier diseases, bridge pier gap is a kind of a kind of damaged shape for jeopardizing bridge pier safety but more difficult measurement
State;Mostly is rested on the manual work stage to the detection of such disease at present, puts crackle usually using the detecting instrument of short distance
Width is detected after big, the length of crackle is obtained by inaccurate measurement or estimation, and this method needs to detect
Personnel carry out by road detection vehicle or built, therefore great work intensity, testing cost are high and require very personnel safety
It is high.
In recent years, mentioning with performances such as the progress of science and technology and computer digital animation ability, speed, capacity
The development of high and digital photography technology, digital image processing techniques are widely used in office automation, industrial robot, geographical number
According to related fieldss such as processing and medicine;Detection means surface deformation and damage, leaf have been successfully applied in terms of Industrial Engineering
The research of piece area and component by complex stress;In view of the deficiency and Digital image technology of domestic current bridge pier detection technique
Advantage, carry out the bridge pier gap automatic identification technology based on image processing technique application practice, to ensure bridge pier operation peace
Entirely, it reduces testing cost, the fast development of China's communication is pushed to be of great significance.
Summary of the invention
A kind of bridge pier surface gaps visible detection method proposed the purpose of the present invention is to solve background technique leads to
It crosses linear array CCD camera and obtains bridge pier surface color image;Color image splits' positions are stored, color image is converted to gray scale
Image, and carry out noise reduction process;Using the linear character in bridge pier gap, gray level image is carried out using image masks filtering algorithm
Image enhancement smoothing processing;The section of region A and region B are calculated separately using Furthest Neighbor in section, area, distance in area, passed through
Change parameter value and make that the section of region A and region B, distance ratio reaches maximum in area, threshold value result at this time is as bridge pier seam
The synthesis segmentation threshold of gap image is examined using slot edge of the wavelet transformation gradient image partitioning algorithm to bridge pier image
It surveys;The slot edge that bridge pier image is finally extracted using Multiscale Morphological Image Edge-Detection method, with marker character control
Morphological watersheds algorithm identifies the gap in bridge pier image, can be realized the efficient identification detection to bridge pier gap, has good
Good application prospect.
Due to the influence of various factors, such as Image Acquisition, image transmitting, all kinds of transformation of image etc. usually affect bridge
The quality of pier image brings difficulty to the identification of bridge pier gap;The primary goal of image enhancement is processing image, it is made to compare original graph
As being more suitable for specifically applying;For bridge pier slot image, exactly enhance bridge pier gap information, reduces texture, shade, light
According to all kinds of noises such as uneven.
To achieve the goals above, present invention employs following technical solutions:
Linear array CCD camera and displacement sensor are fixed on same detection by a kind of bridge pier surface gaps visible detection method
Platform, in detection platform moving process, displacement sensor provides displacement signal, and linear array CCD camera is obtained according to the displacement signal
Bridge pier surface color image;
Color image splits' positions are stored, color image are converted to gray level image, and carry out noise reduction process;
Using the linear character in bridge pier gap, it is smooth that image enhancement is carried out to gray level image using image masks filtering algorithm
Processing;
The section of region A and region B are calculated separately using Furthest Neighbor in section, area, distance in area, passes through and change parameter value
So that distance ratio reaches maximum in the section of region A and region B, area, threshold value result at this time is as the comprehensive of bridge pier slot image
Close segmentation threshold;
It is detected using slot edge of the wavelet transformation gradient image partitioning algorithm to bridge pier image;
It is grasped using Algorithm of Multi-Scale Morphological Edge Detection according to the type of structural element, the size of structural element, expansion
The number of work is split bridge pier image and then extracts bridge pier image slot edge.
In gray level image, adjacent pixel gray-value variation is gentle inside a region, and difference is smaller, and statistical variance is small;
On the contrary, the gray-value variation between image pixel is larger, thus statistical variance is with regard to big at the edge in region;The purpose of mask process
While being exactly filtered operation to image, as far as possible should not destroyed area edge details, this method be applied to bridge pier
When image enhancement, the noise in image can be not only removed, but also can preferably protect the details of slot edge;To the place of Image neighborhood
Science and engineering is made to be exactly the subgraph picture for handling the pixel value of the neighborhood image and having the distribution of similar or same pixel to the neighborhood
Element value.
The mechanism of filtering processing is to move the exposure mask of predefined point by point in image to be processed, and filter is each
Response at point (x, y) is calculated according to the formula of predefined, its general response output is covered by exposure mask coefficients with filtering
The product accumulation summation that mould is moved through the corresponding pixel value in region provides.
Can be being divided into multiple regions or subgraph in a width digital picture, when subregion, allows the variation of adjacent area as far as possible
Greatly, the variation and inside the same area is smaller, and inside the same area, and the variation of intermediate pixel is less than edge pixel
Variation.
In order to improve the quality of image, reduce and removal image in noise, can abstract image linear feature,
When realizing specific calculate, exposure mask central point is overlapped with pixel to be treated in image, and by each element value in exposure mask and
Exposure mask covers the corresponding pixel value of image-region and is multiplied, the output response of the gray value exposure mask of pixel where exposure mask central point come
Instead of, that is, the template each element sum of products that previous step calculates.
If the linear filter response value of mask with 3 × 3 cover modules, according to above-mentioned method, at image midpoint (x, y)
R are as follows:
R=w (- 1, -1) f (x-1, y-1)+w (- 1,0) f (x-1, y)+...+w (0,0) f (x, y)+w (1,0) f (x+1, y)
+ w (1,1) f (x+1, y+1)
When realizing exposure mask smothing filtering, when exposure mask moves in the picture, filter center point is close to image outline, such as
When the rectangular smooth exposure mask of one n × n size of fruit, exposure mask center and image border distance (n-2)/2 pixel, image border
It is overlapped at least one exposure mask side;When exposure mask continues mobile to image border, the side of exposure mask will be in except image border;Cause
This in mobile exposure mask tries not that exposure mask central point is allowed to be greater than (n-2)/2 pixel with a distance from image border.
Since bridge pier image slot has linear character, and there is directionality, according to this feature, extends a upper section and beg for
The single mode plate of opinion constructs single exposure mask in 8 directions;In general, a pixel in image has 8 neighborhoods, that is,
Say that a pixel there are 8 directions, as shown in Figure 2.
In 8 directions, the template that 8 sizes are 3 × 3 is constructed, 8 templates are successively handled into the same image window,
In image window, the gray value of its each pixel is successively multiplied with the element value of template corresponding position, then add up summation, knot
Fruit is Ci=0,1 ... 7.
The center pixel output valve of image window is indicated with following formula: g (m, n)=max (Ci)。
When image window is 5 × 5, center (soft dot in window) is (j, k), can be determined in the window
9 kinds of different exposure mask templates, as shown in Figure 3.
Therefore, when carrying out exposure mask filtering to image, the mean and variance of each template is calculated first:
In formula: i indicates the number of each exposure mask template, and i=1,2 ..., 9, q be in corresponding exposure mask template comprising pixel
Number, (m, n) are displacement of the pixel relative to center pixel (j, k) in exposure mask template.
Calculate the variance of 9 templates, and be compared, using gray average corresponding to the template with minimum variance as
The new gray value that exposure mask smoothly exports:
Therefore, as further restriction of the invention, image enhancement smoothing processing specifically:
Firstly, calculating the mean value E of each templateiAnd variance Ωi:
In formula: i indicates the number of each mask plate, i=1,2 ..., 9,
Q is the number comprising pixel in corresponding mask plate, and (m, n) is pixel in mask plate relative to center pixel (j, k)
Displacement;
The variance of 9 templates is calculated again and is compared, and gray average corresponding to the template with minimum variance is made
For the new gray value that exposure mask smoothly exports,
Since image is made of target and background, wherein target seeks to the slot object of identification, needs in this way according to picture
Plain value divides the image into two parts, and it is exactly to select a threshold value that the most common method in gap is partitioned into from background, is greater than the threshold
Value is exactly gap target, other composition backgrounds.
In threshold value, if threshold value is excessively highly fixed, the object point accidentally occurred will be considered as background;If threshold
Be worth it is too low, then can there is a situation where opposite;Therefore, when bridge pier surface image is complex, histogram is difficult to occur apparent
Peak value often joins together, and chooses optimal threshold with histogram and becomes difficult;At this moment can be believed according to the overall situation of entire image
Breath carrys out threshold value, can obtain preferably with the gap in maximum area, in zone distance threshold value criterion extraction bridge pier image
Effect.
Therefore, as further restriction of the invention, comprehensive segmentation threshold is sought in the following manner:
Whole picture road surface breakage image is divided into a series of subgraph;Then the segmentation threshold of each subgraph is calculated separately
Value, and seek the average gray of the average gray of full figure, region A and region B;Area is calculated using Furthest Neighbor distribution in section, area
Distance in the section of domain A and region B, area, finally by changeValue acquire the section of region A and region B, distance ratio in area
Reach maximum, the synthesis segmentation threshold as breakage image:
Wherein,For the zone distance of region A and region B;For in the area of region A and region B away from
From.
In bridge pier image, the gap of some position and background and the gap in another region and background not fully phase
Together, exist must difference, this is because the influence of image capture device and acquisition environment, the i.e. non-gap area in some regions
Gray value is relatively low, and the gap area gray value in some regions can be higher;At this point, utilizing Furthest Neighbor meter in global maximum section, area
The segmentation threshold for calculating entire image, easily causes certain gap areas being mistaken for non-gap area, also can be often by certain non-gaps
Region is mistaken for gap area;Therefore, using between global maximum kind, inter- object distance Threshold Segmentation Algorithm calculate segmentation threshold, no
Damaged area can be split from non-damaged area well.
And using distance in the Furthest Neighbor distribution section of zoning A and region B in section, area, area, finally by changing
BecomeValue acquire that the section of region A and region B, distance ratio reaches maximum in area, as the synthesis segmentation threshold of breakage image,
It can be realized in background area, weighted by the threshold value of global threshold, obtain a new threshold value, i.e., comprehensive segmentation threshold,
Then the synthesis segmentation threshold wants small compared with original global threshold;It is carried out using the new segmentation threshold road pavement breakage image
Segmentation, then substantially can be determined as object pixel for the fritter.
In the automatic identification of image slot, due to that can not learn image in advance, there are the noises in what direction, very
The subgraph of hardly possible selection reconstruct, eliminates random noise and increasing in fact, being only reconstructed to smoothed image and can substantially reach
The purpose of strong slot image;It, at this moment can be using selection but if image has a large amount of texture both horizontally and vertically
The reconstruction strategy of horizontal subgraph and oblique line subgraph.
The reconstruct of above-mentioned horizontal subgraph and oblique line subgraph is needed to realize by wavelet transformation to image reconstruction
Multiresolution analysis;Therefore, as further restriction of the invention, wavelet transformation gradient image partitioning algorithm specifically: logical
Cross wavelet transformation and its inverse transformation for reconstruct, gradient direction and mould maximum value be stored in wavelet coefficient, analysis and
Wavelet coefficient maximum value is converted, noise and the corresponding wavelet coefficient of feature are handled respectively, realize the inspection of the slot edge of image
It surveys and noise separates, effective smooth, Edge contrast to gap in image can be realized.
The purpose of image slot edge enhancing is to reinforce clarification of objective information, while inhibiting the influence of noise;However, band
The bridge pier image enhancement of noise is the difficult point that all enhancing algorithms all suffer from, because of the edge that noise and true picture are changed significantly
Equally, high-frequency sub-band is both corresponded in frequency domain, improves the contrast at edge, improvement when highlighting high frequency section using enhancing algorithm
While picture quality, it will inevitably amplify noise.
Therefore, different processing can be done respectively to noise and the corresponding wavelet coefficient of feature, according on each scale 2j
The size of the two components of wavelet coefficient module maximum value and gradient direction determines position and the attribute at edge;Reducing the same of noise
When, there is enhancing to slot edge;Specific practice is two component W on each scale 2j to wavelet transformation1And W2It does
Transformation, and to the subgraph W that jth layer decomposesj(x, y) is adaptively adjusted, and is done such as down conversion:
Wherein,For threshold value,For gain,It is the edge on scale j.
In order to make algorithm that there is good adaptive adjustment capability, parameterIt is as follows respectively:
Wherein,
W1And W2It is two components of gradient image (x, y) wavelet transformation.
Therefore, as further restriction of the invention, noise and the corresponding wavelet coefficient of feature are handled specifically respectively
Are as follows:
The subgraph W that jth layer is decomposedj(x, y) is adaptively adjusted, and following variation is done:
Wherein,For threshold value,For gain,It is the edge on size j;
In order to make algorithm that there is good adaptive adjustment capability, parameterIt is as follows respectively:
Wherein, W1With
W2It is gradient image WjTwo components of (x, y) wavelet transformation.
A), b) be that above-mentioned algorithm of the invention carries out edge detection in Fig. 5 as a result, image after noise reduction, in Fig. 5 a),
B) noise of image is significantly reduced.
Due to the influence of the various factors such as bridge pier unity and coherence in writing, bridge pier material, photo environment, bridge pier image background noise is more multiple
Miscellaneous, the noise for including is also more, in order to avoid influencing the effect of slot edge detection, first should carry out Mathematical morphology filter to bridge pier image
Wave processing, morphological open calculation process eliminate in bridge pier image that the lesser small gray value of size is thin compared with structural element
Section, and the biggish region of sum of the grayscale values of entire bridge pier image is kept, the fine gap sometimes in image is taken as in this way
Background, it is not easy to identified.
Morphologic closure operation can eliminate in bridge pier image the lesser high-gray level value details of size compared with structural element,
And the biggish image-region of sum of the grayscale values of entire bridge pier image is kept substantially;Therefore, morphology opening operation and closed operation are available
In the filtering processing of bridge pier image.
But in practical application, the determination of structural element will consider the shape of bridge pier image slot component part to be processed
Shape feature and its size, because it cannot handle same shape and object of different sizes;So should consider to deal with objects
Shape consider its size again, the present invention is proposed to be continuously increased the filtering processing of the alternating sequence of structural element scale, according to
The sequencing of opening and closing operation is respectively defined as:
Wherein, B1、B2…BiIt is the structural element of continuous increased in size.
A) it is original bridge pier image, b in Fig. 6) it is filtered as a result, c) to original image a) progress square structure element opening and closing
Be to figure a) with constantly increase scale structural element alternately filter as a result, d) be original image a) median-filtered result, from figure
In 6 it is each treated image comparison can be seen that opening and closing filtering plays image must smoothing effect, list and median filtering
It compares, filter effect is more slightly worse than median filtering, and median filtering is better than opening and closing filtering;Figure is c) to original image constantly increase scale
Structural element carries out alternately filtering to original image a), and effect is better than median filtering, and is better than opened and closed filtering;Therefore, alternately
Opening and closing filtering algorithm can obtain better smooth effect.
In bridge pier image, generally there is lower gray value in gap, shows strip structure, therefore can be according to the gray scale of image
Ingredient detects the slot object with minimum gradation value, therefore can convert that (original image image subtraction opens fortune with improved top cap
Image after calculation is known as top cap transformation) as gap detection, detection has been subjected to the image after morphologic filtering.
It is to carry out opening operation to original bridge pier image with structural element that top cap, which converts algorithmic procedure,It counts again
Calculate the difference of original image and the opening operationIts result is exactly the transformation of gray scale bridge pier image top cap.
Since often there is the brightness of very little in gap in bridge pier image, they appear as the dark-coloured area in bright background
Therefore domain will extract gap characteristic, it is necessary to first negate to image, then recycleCarry out top cap change
It changes.
Correspondingly, in order to avoid negating to image, closed operation first can be carried out to image with structure collection nB, then subtracts original image
As f (x), alternatively referred to as improved bottom cap transformation, expression formula are as follows: fnB(x)-f (x), wherein fnBIt (x) is structural element nB to ash
The opening operation of image f (x) is spent,(n times);The image exported after above-mentioned processing is
Gap can be split the threshold operation of image by one shade of gray image from bridge pier image.
Therefore, as further restriction of the invention, Multiscale Morphological Image Edge-Detection method specifically:
Image is carried out using the structural element constantly increased first to be alternately opened and closed filtering, smooth bridge pier slot image is simultaneously gone
It except noise, then negates to image, recycles multiple dimensioned top cap variation to carry out top cap transformation, examined using Multiscale Morphological edge
It surveys device and extracts bridge pier image slot edge, finally extracted in bridge pier image with the Morphological watersheds algorithm of marker character control
Slot object carries out slot image identification.
The present invention extracts the gap in bridge pier image, watershed segmentation using the Morphological watersheds algorithm that marker character controls
Algorithm is image to be regarded as one secondary " topographic map ", and the stronger local pixel value of brightness is larger, the darker local pixel of brightness ratio
Value is smaller, is split by searching " catchment basin " and " watershed boundary line " to image.
It is directly often and bad using the effect of fractional spins, if in the picture to foreground object and background pair
As being labeled difference, preferable segmentation effect can be obtained by reapplying watershed algorithm;Watershed segmentation based on marking of control
Specific step is as follows for method:
Segmentation function is calculated, darker region is part in the object to be divided, that is, primary Calculation image in image
The position of Minimum Area, the position (referring to gap in experiment herein) of cutting object is marked in it;Above-mentioned steps illustrate with
The incoherent details of our segmentation problems calculates the collection of image Minimum Area by some height threshold values T to eliminate these details
It closes, to preferably mark the position of local minimum area;External label is calculated using watershed range conversion, it is general with dividing water
Ridge crestal line is as external label;After obtaining inside and outside marker character, using forcing minimum technology to modify grayscale image, most so as to part
Zonule only occurs in the position of label;Watershed transform is carried out to the image for having modified marker character, is partitioned into slot object.
A) it is bridge pier slot image after morphological transformation in Fig. 7, b) is to a) carrying out watershed transform in Fig. 7 in Fig. 7
Image;As can be seen that Morphological watersheds transformation has certain effect in the gap segmentation to bridge pier image, also have preferably
Ground Split effect.
The beneficial effects of the present invention are:
1, need of the present invention for distribution character and bridge pier the slot image segmentation of bridge pier slot image grey scale pixel value
Ask, using image masks Preprocessing Algorithm, realize the filtering of bridge pier slot image, and from different low-pass filtering and high-pass filtering method
The filter effect of bridge pier slot image is compared, on the basis of reducing image noise, improving image quality, is effectively protected
Image slot edge has been protected, has been provided the foundation for the identification and segmentation in subsequent gap.
2, the present invention searches for the mould maximum value of the small echo on image by the gradient and mould maximum value of calculating wavelet transformation
Point obtains the crack information for needing to identify, is then handled respectively noise and the corresponding wavelet coefficient of feature, is reducing noise
Meanwhile there is enhancing to bridge pier slot edge.
3, by being improved to conventional Morphology Algorithm, using Algorithm of Multi-Scale Morphological Edge Detection, according to knot
Type, the size of structural element, number of expansive working of constitutive element etc. are split bridge pier image, can obtain preferable
Effect.
Detailed description of the invention
Fig. 1 is the schematic diagram of bridge pier surface gaps visible detection method proposed by the present invention
Fig. 2 is template direction definition figure in bridge pier surface gaps visible detection method proposed by the present invention
Fig. 3 is mask window and 9 kinds of different mask plate schematic diagrames
Fig. 4 is the comparison diagram of a variety of image enchancing methods
Fig. 5 is the result handled using wavelet transformation gradient image partitioning algorithm the slot edge of bridge pier image
Fig. 6 is the result that different filtering modes are used to bridge pier slot image
A) it is bridge pier slot image after morphological transformation, b in Fig. 7) it is to the image for a) carrying out watershed transform
Specific embodiment
It is next combined with specific embodiments below that the present invention is further described.
Linear array CCD camera and displacement sensor are fixed on same detection by a kind of bridge pier surface gaps visible detection method
Platform, in detection platform moving process, displacement sensor provides displacement signal, and linear array CCD camera is obtained according to the displacement signal
Bridge pier surface color image;
Color image splits' positions are stored, color image are converted to gray level image, and carry out noise reduction process;
Using the linear character in bridge pier gap, it is smooth that image enhancement is carried out to gray level image using image masks filtering algorithm
Processing;
The section of region A and region B are calculated separately using Furthest Neighbor in section, area, distance in area, passes through and change parameter value
So that distance ratio reaches maximum in the section of region A and region B, area, threshold value result at this time is as the comprehensive of bridge pier slot image
Close segmentation threshold;
It is handled using slot edge of the wavelet transformation gradient image partitioning algorithm to bridge pier image;
The slot edge of bridge pier image is extracted using Multiscale Morphological Image Edge-Detection method, finally uses marker character control
Gap in the Morphological watersheds algorithm identification bridge pier image of system.
Wherein, image enhancement smoothing processing specifically:
Firstly, calculating the mean value E of each templateiAnd variance Ωi:
In formula: i indicates the number of each mask plate, i=1,2 ..., 9,
Q is the number comprising pixel in corresponding mask plate, and (m, n) is pixel in mask plate relative to center pixel (j, k)
Displacement;
The variance of 9 templates is calculated again and is compared, and gray average corresponding to the template with minimum variance is made
For the new gray value that exposure mask smoothly exports,
Wherein, comprehensive segmentation threshold is sought in the following manner:
Whole picture road surface breakage image is divided into a series of subgraph;Then the segmentation threshold of each subgraph is calculated separately
Value, and seek the average gray of the average gray of full figure, region A and region B;Area is calculated using Furthest Neighbor distribution in section, area
Distance in the section of domain A and region B, area, finally by changeValue acquire the section of region A and region B, distance ratio in area
Reach maximum, the synthesis segmentation threshold as breakage image:
Wherein,For the zone distance of region A and region B;For in the area of region A and region B away from
From.
Wavelet transformation gradient image partitioning algorithm specifically:, will be terraced by wavelet transformation and its for the inverse transformation of reconstruct
Degree direction and mould maximum value are stored in wavelet coefficient, analysis and transformation wavelet coefficient maximum value, corresponding to noise and feature
Wavelet coefficient handle respectively, realize the slot edge of image detection and noise separation.
Noise and the corresponding wavelet coefficient of feature are handled respectively specifically:
The subgraph W that jth layer is decomposedj(x, y) is adaptively adjusted, and following variation is done:
Wherein,For threshold value,For gain,It is the edge on size j;
In order to make algorithm that there is good adaptive adjustment capability, parameterIt is as follows respectively:
Wherein,
W1And W2It is gradient image WjTwo components of (x, y) wavelet transformation.
Wherein, Multiscale Morphological Image Edge-Detection method specifically: first using the structural element constantly increased to figure
As carrying out alternately being opened and closed filtering, smooth bridge pier slot image simultaneously removes noise, then negates to image, recycle multiple dimensioned top cap
Variation carries out top cap transformation, using Algorithm of Multi-Scale Morphological Edge Detection according to the type of structural element, structural element it is big
Small, expansive working number is split bridge pier image and then extracts bridge pier image slot edge, finally uses marker character control
The Morphological watersheds algorithm of system extracts the slot object in bridge pier image, carries out slot image identification.
The present invention obtains bridge pier surface color image by linear array CCD camera;Color image splits' positions are stored, it will be color
Chromatic graph picture is converted to gray level image, and carries out noise reduction process;Using the linear character in bridge pier gap, is filtered and calculated using image masks
Method carries out image enhancement smoothing processing to gray level image;The area of region A and region B are calculated separately using Furthest Neighbor in section, area
Between, distance in area, make that the section of region A and region B, distance ratio reaches maximum in area by changing parameter value, threshold at this time
It is worth synthesis segmentation threshold of the result as bridge pier slot image, using wavelet transformation gradient image partitioning algorithm to bridge pier image
Slot edge is detected;The slot edge of bridge pier image, fortune are finally extracted using Multiscale Morphological Image Edge-Detection method
The gap in bridge pier image is identified with the Morphological watersheds algorithm that marker character controls, and can be realized the efficient knowledge to bridge pier gap
It does not detect, has a good application prospect.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Anyone skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (6)
1. a kind of bridge pier surface gaps visible detection method, it is characterised in that: linear array CCD camera and displacement sensor to be fixed on
Same detection platform, in detection platform moving process, displacement sensor provides displacement signal, and linear array CCD camera is according to the displacement
Signal acquisition bridge pier surface color image;
Color image splits' positions are stored, color image are converted to gray level image, and carry out noise reduction process;
Using the linear character in bridge pier gap, image enhancement is carried out to gray level image using image masks filtering algorithm and is smoothly located
Reason;
The section of region A and region B are calculated separately using Furthest Neighbor in section, area, distance in area, is made by changing parameter value
Distance ratio reaches maximum in the section of region A and region B, area, synthesis point of the threshold value result as bridge pier slot image at this time
Cut threshold value;
It is handled using slot edge of the wavelet transformation gradient image partitioning algorithm to bridge pier image;
The slot edge of bridge pier image is extracted using Multiscale Morphological Image Edge-Detection method, finally with marker character control
Morphological watersheds algorithm identifies the gap in bridge pier image.
2. a kind of bridge pier surface gaps visible detection method according to claim 1, it is characterised in that: described image enhancing
Smoothing processing specifically:
Firstly, calculating the mean value E of each templateiAnd variance Ωi:
In formula: i indicates the number of each mask plate, i=1,2 ..., 9,
Q is the number comprising pixel in corresponding mask plate, and (m, n) is position of the pixel relative to center pixel (j, k) in mask plate
Shifting amount;
The variance of 9 templates is calculated again and is compared, using gray average corresponding to the template with minimum variance as covering
The new gray value that film smoothly exports,
3. a kind of bridge pier surface gaps visible detection method according to claim 2, it is characterised in that: the comprehensive segmentation
Threshold value is sought in the following manner:
Whole picture road surface breakage image is divided into a series of subgraph;Then the segmentation threshold of each subgraph is calculated separately,
And seek the average gray of the average gray of full figure, region A and region B;Zoning A is distributed using Furthest Neighbor in section, area
With distance in the section of region B, area, finally by changeValue acquire that the section of region A and region B, distance ratio reaches in area
Maximum, the synthesis segmentation threshold as breakage image:
Wherein,For the zone distance of region A and region B;For distance in the area of region A and region B.
4. a kind of bridge pier surface gaps visible detection method according to claim 3, it is characterised in that: the wavelet transformation
Gradient image partitioning algorithm specifically: by wavelet transformation and its for the inverse transformation of reconstruct, by gradient direction and mould maximum value
It is stored in wavelet coefficient, analysis and transformation wavelet coefficient maximum value locate noise and the corresponding wavelet coefficient of feature respectively
Reason realizes the detection of the slot edge of image and noise separation.
5. a kind of bridge pier surface gaps visible detection method according to claim 4, it is characterised in that: it is described to noise and
The corresponding wavelet coefficient of feature is handled respectively specifically:
The subgraph W that jth layer is decomposedj(x, y) is adaptively adjusted, and following variation is done:
Wherein,For threshold value,For gain,It is the edge on size j;
In order to make algorithm that there is good adaptive adjustment capability, parameterIt is as follows respectively:
Wherein,W1And W2
It is gradient image WjTwo components of (x, y) wavelet transformation.
6. a kind of bridge pier surface gaps visible detection method according to claim 5, it is characterised in that: the multiple dimensioned shape
State Image Edge-Detection method specifically:
First using the structural element constantly increased to image carry out alternately be opened and closed filtering, smooth bridge pier slot image and remove make an uproar
Then sound negates to image, recycle multiple dimensioned top cap variation to carry out top cap transformation, calculated using Multiscale Morphological edge detection
Method according to the type of structural element, the size of structural element, expansive working number, to bridge pier image be split so that extract
Bridge pier image slot edge finally extracts the gap pair in bridge pier image with the Morphological watersheds algorithm of marker character control
As carrying out slot image identification.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811334774.2A CN109523479A (en) | 2018-11-10 | 2018-11-10 | A kind of bridge pier surface gaps visible detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811334774.2A CN109523479A (en) | 2018-11-10 | 2018-11-10 | A kind of bridge pier surface gaps visible detection method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109523479A true CN109523479A (en) | 2019-03-26 |
Family
ID=65773868
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811334774.2A Pending CN109523479A (en) | 2018-11-10 | 2018-11-10 | A kind of bridge pier surface gaps visible detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109523479A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112033385A (en) * | 2020-11-03 | 2020-12-04 | 湖南联智科技股份有限公司 | Pier pose measuring method based on mass point cloud data |
CN112233111A (en) * | 2020-11-18 | 2021-01-15 | 安徽国钜工程机械科技有限公司 | Tunnel gap detection method based on digital image processing |
CN112508859A (en) * | 2020-11-19 | 2021-03-16 | 聚融医疗科技(杭州)有限公司 | Method and system for automatically measuring thickness of endometrium based on wavelet transformation |
CN113702257A (en) * | 2021-08-09 | 2021-11-26 | 西南石油大学 | Conglomerate pore structure characterization method based on CT three-dimensional data volume |
CN115512306A (en) * | 2022-11-15 | 2022-12-23 | 成都睿瞳科技有限责任公司 | Method for early warning of violence events in elevator based on image processing |
CN116597389A (en) * | 2023-07-18 | 2023-08-15 | 山东省地质测绘院 | Geological disaster monitoring and early warning method based on image processing |
CN116630321A (en) * | 2023-07-24 | 2023-08-22 | 铁正检测科技有限公司 | Intelligent bridge health monitoring system based on artificial intelligence |
CN116630225A (en) * | 2023-03-13 | 2023-08-22 | 中铁大桥局集团有限公司 | Method and device for identifying underwater foundation damage of railway bridge and processing equipment |
CN116958182A (en) * | 2023-09-20 | 2023-10-27 | 广东华宸建设工程质量检测有限公司 | Quick concrete crack detection method based on image data |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101493932A (en) * | 2009-03-05 | 2009-07-29 | 西安电子科技大学 | Watershed texture imaging segmenting method based on morphology Haar small wave texture gradient extraction |
CN107154040A (en) * | 2017-05-08 | 2017-09-12 | 重庆邮电大学 | A kind of tunnel-liner surface image crack detection method |
EP3392612A1 (en) * | 2015-12-14 | 2018-10-24 | Nikon-Trimble Co., Ltd. | Defect detection apparatus and program |
-
2018
- 2018-11-10 CN CN201811334774.2A patent/CN109523479A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101493932A (en) * | 2009-03-05 | 2009-07-29 | 西安电子科技大学 | Watershed texture imaging segmenting method based on morphology Haar small wave texture gradient extraction |
EP3392612A1 (en) * | 2015-12-14 | 2018-10-24 | Nikon-Trimble Co., Ltd. | Defect detection apparatus and program |
CN107154040A (en) * | 2017-05-08 | 2017-09-12 | 重庆邮电大学 | A kind of tunnel-liner surface image crack detection method |
Non-Patent Citations (1)
Title |
---|
孙波成: "基于数字图像处理的沥青路面裂缝识别技术研究", 《中国博士学位论文全文数据库 工程科技II辑》 * |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112033385B (en) * | 2020-11-03 | 2021-02-02 | 湖南联智科技股份有限公司 | Pier pose measuring method based on mass point cloud data |
CN112033385A (en) * | 2020-11-03 | 2020-12-04 | 湖南联智科技股份有限公司 | Pier pose measuring method based on mass point cloud data |
CN112233111A (en) * | 2020-11-18 | 2021-01-15 | 安徽国钜工程机械科技有限公司 | Tunnel gap detection method based on digital image processing |
CN112508859A (en) * | 2020-11-19 | 2021-03-16 | 聚融医疗科技(杭州)有限公司 | Method and system for automatically measuring thickness of endometrium based on wavelet transformation |
CN113702257A (en) * | 2021-08-09 | 2021-11-26 | 西南石油大学 | Conglomerate pore structure characterization method based on CT three-dimensional data volume |
CN115512306A (en) * | 2022-11-15 | 2022-12-23 | 成都睿瞳科技有限责任公司 | Method for early warning of violence events in elevator based on image processing |
CN116630225A (en) * | 2023-03-13 | 2023-08-22 | 中铁大桥局集团有限公司 | Method and device for identifying underwater foundation damage of railway bridge and processing equipment |
CN116630225B (en) * | 2023-03-13 | 2024-05-14 | 中铁大桥局集团有限公司 | Method and device for identifying underwater foundation damage of railway bridge and processing equipment |
CN116597389A (en) * | 2023-07-18 | 2023-08-15 | 山东省地质测绘院 | Geological disaster monitoring and early warning method based on image processing |
CN116597389B (en) * | 2023-07-18 | 2023-09-15 | 山东省地质测绘院 | Geological disaster monitoring and early warning method based on image processing |
CN116630321B (en) * | 2023-07-24 | 2023-10-03 | 铁正检测科技有限公司 | Intelligent bridge health monitoring system based on artificial intelligence |
CN116630321A (en) * | 2023-07-24 | 2023-08-22 | 铁正检测科技有限公司 | Intelligent bridge health monitoring system based on artificial intelligence |
CN116958182A (en) * | 2023-09-20 | 2023-10-27 | 广东华宸建设工程质量检测有限公司 | Quick concrete crack detection method based on image data |
CN116958182B (en) * | 2023-09-20 | 2023-12-08 | 广东华宸建设工程质量检测有限公司 | Quick concrete crack detection method based on image data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109523479A (en) | A kind of bridge pier surface gaps visible detection method | |
CN109800824B (en) | Pipeline defect identification method based on computer vision and machine learning | |
CN110443806B (en) | Water surface transparent floating hazardous chemical substance image segmentation method based on target enhancement processing | |
Chen et al. | A novel color edge detection algorithm in RGB color space | |
CN111753577B (en) | Apple identification and positioning method in automatic picking robot | |
CN109410228A (en) | Internal wave of ocean detection algorithm based on Method Based on Multi-Scale Mathematical Morphology Fusion Features | |
CN110766689A (en) | Method and device for detecting article image defects based on convolutional neural network | |
CN104794502A (en) | Image processing and mode recognition technology-based rice blast spore microscopic image recognition method | |
CN106815819B (en) | More strategy grain worm visible detection methods | |
CN101551853A (en) | Human ear detection method under complex static color background | |
CN110348461A (en) | A kind of Surface Flaw feature extracting method | |
CN111062931A (en) | Detection method of spliced and tampered image | |
CN113592782B (en) | Method and system for extracting X-ray image defects of composite material carbon fiber core rod | |
CN116152115B (en) | Garbage image denoising processing method based on computer vision | |
CN114549492A (en) | Quality evaluation method based on multi-granularity image information content | |
CN105405138A (en) | Water surface target tracking method based on saliency detection | |
CN107154044A (en) | A kind of dividing method of Chinese meal food image | |
CN114764801A (en) | Weak and small ship target fusion detection method and device based on multi-vision significant features | |
CN108242060A (en) | A kind of method for detecting image edge based on Sobel operators | |
CN110047041A (en) | A kind of empty-frequency-domain combined Traffic Surveillance Video rain removing method | |
CN103530636A (en) | Snakes model based method for extracting SAR (synthetic aperture radar) image target profile | |
Chen et al. | A study of image segmentation algorithms combined with different image preprocessing methods for thyroid ultrasound images | |
Bansal | Implementing edge detection for detecting neurons from brain to identify emotions | |
Valliammal et al. | Performance analysis of various leaf boundary edge detection algorithms | |
CN105809187A (en) | Multi-manufacturer partial discharge data result diagnosis analysis method based on image identification |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190326 |