CN109671043A - A kind of surface cracks rapid detection system and method based on unmanned plane - Google Patents
A kind of surface cracks rapid detection system and method based on unmanned plane Download PDFInfo
- Publication number
- CN109671043A CN109671043A CN201811611646.8A CN201811611646A CN109671043A CN 109671043 A CN109671043 A CN 109671043A CN 201811611646 A CN201811611646 A CN 201811611646A CN 109671043 A CN109671043 A CN 109671043A
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- surface cracks
- snowfield
- subprogram
- 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
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/182—Network patterns, e.g. roads or rivers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration by the use of histogram techniques
-
- 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
- G06T7/136—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
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
Abstract
The invention discloses a kind of surface cracks rapid detection system and method based on unmanned plane.First the unmanned plane in unloaded subsystem carries out surface cracks primary dcreening operation to the present invention, has the picture transfer of surface cracks to ground primary dcreening operation discovery, is further calculated by ground system.Geological disaster hidden danger picture is identified by ground system, reminds artificial treatment.
Description
Technical field
The present invention relates to air vehicle technique fields, and in particular to a kind of surface cracks rapid detection system based on unmanned plane
And method.
Background technique
Surface cracks are the characteristic features of geological disaster hidden danger, and surface cracks are commented in fast quick checking, for the investigation of geological disaster
There is key.Currently, unmanned plane will obtain a large amount of pictures when taking photo by plane geological disaster image.Due to wirelessly communicating number
According to capacity, these pictures can not be real-time transmitted to ground, can only first store unmanned plane, just can be carried out place after unmanned plane landing
Reason.By manually carrying out Assessment of Geological Hazard after landing.This method cannot find that geological disaster hidden danger, artificial treatment are a large amount of in time
Picture needs take a significant amount of time.
Summary of the invention
For above-mentioned deficiency in the prior art, a kind of surface cracks based on unmanned plane provided by the invention are quickly detected
System and method solves the problems, such as that surface cracks can not be identified quickly.
In order to achieve the above object of the invention, a kind of the technical solution adopted by the present invention are as follows: surface cracks based on unmanned plane
Rapid detection system, including the ground subsystem communicated by wireless transmission method and unloaded subsystem；
Acquired image information data is carried out initial treatment for acquiring image information data by unloaded subsystem, and
Send the data to ground subsystem；
Ground subsystem, the image data sent for receiving unloaded subsystem, carries out final process for image data, and
Storing data；
The ground subsystem includes computer interconnected and the first wireless transmission submodule；The zero load subsystem
Including unmanned plane and image processing module, described image processing module is arranged in uav bottom；Described image processing module packet
Include image sensing submodule, programmable gate array, the second wireless transmission submodule and Video compression chip, the figure
As sensing submodule, Video compression chip and the second wireless transmission submodule are connect with programmable gate array；Institute
Stating image sensing submodule includes mounting box, camera lens and sensing circuit, is equipped with sensing circuit, the mounting box in the mounting box
Side be fixed on uav bottom, the other side of the mounting box is connect by circular wire mouth with camera lens.
Further: pixel samples program and surface cracks primary dcreening operation program, institute are equipped in the Programmadle logic gate array
Stating surface cracks primary dcreening operation program includes that snowfield judges that subprogram, snowfield surface cracks hidden danger primary dcreening operation subprogram, meadow judge sub- journey
Sequence, meadow surface cracks hidden danger primary dcreening operation subprogram, waters primary dcreening operation subprogram, surface cracks hidden danger pixel judge subprogram and earth's surface
Crack hidden danger pixel counts subprogram, the pixel samples program is for being sampled image pixel, at the beginning of the surface cracks
For sieve program for calculating surface cracks hidden danger pixel quantity, the snowfield judges that subprogram is used to obtain intermediate parameters Fx, described
For snowfield surface cracks hidden danger primary dcreening operation subprogram for obtaining the bianry image containing crack of snowfield image, the meadow judges subprogram
Intermediate parameters Fc is obtained, the meadow surface cracks hidden danger primary dcreening operation subprogram is for obtaining the binary map containing crack of meadow image
Picture, for the waters primary dcreening operation subprogram for obtaining river region image, the surface cracks hidden danger pixel judges that subprogram is used for
Binaryzation surface cracks hidden danger image is obtained, the surface cracks hidden danger pixel counts subprogram is for obtaining surface cracks hidden danger
Pixel quantity；Surface cracks primary dcreening operation program also is provided in the computer.
A kind of surface cracks rapid detection method based on unmanned plane, comprising:
S1, the initial parameter that programmable gate array in unloaded subsystem is set；
S2, control signal is issued to sensing circuit by programmable gate array；
S3, the pixel intensity signal that sensing circuit is received by programmable gate array, and to the pixel of sensing circuit
Luminance signal carries out Bayer image processing and white balance processing, obtains white balance treated the rgb signal of each pixel；
S4, the rgb signal for calling pixel samples program processing white balance treated each pixel, the figure after being sampled
As pixel；
S5, the image pixel after the processing sampling of surface cracks primary dcreening operation program is called by programmable gate array, obtain
Surface cracks hidden danger pixel quantity；
S6, judge whether surface cracks hidden danger pixel quantity is greater than threshold value, if so, entering step S7, otherwise, return to step
Rapid S2；
S7, the rgb format picture signal of all pixels is sent to Video compression chip, obtains compressed image
Data；
S8, compressed image data is received by programmable gate array, and compressed image data is sent
To the second wireless transmission submodule, the second wireless transmission submodule will press transmission of data to be sent to the first wireless transmission submodule；
S9, compressed image data is sent to by computer by the first wireless transmission submodule in ground subsystem,
And compressed image data is decompressed by computer, the image data after being decompressed；
S10, the image data after surface cracks primary dcreening operation program processing decompression is called by computer, obtain surface cracks inspection
Survey result.
Further: the specific steps of pixel samples program in the step S4 are as follows:
A1, the danger signal in the rgb signal of image sensing circuit original pixels is set as Ro (i1,j1), green Bo
(i1,j1), blue signal is Go (i1,j1)；
i1=1,2,3 ... x, j1=1,2,3 ... y, x are horizontal pixel number, and y is longitudinal pixel number；
A2, set sampling after image pixel rgb signal in danger signal as Ry (i2,j2), green is By (i2,
j2), blue signal is Gy (i2,j2)；
i2=1,2,3 ... m, j2=1,2,3 ... n, m=[x/k], n=[y/k], [] are rounding operation, and k is sampling rate,
And k is integer, m is sampling horizontal pixel number, and n is to sample longitudinal pixel number；
A3, initial parameter i=1, j=1 are enabled；
A4, initial parameter a=i × k, b=j × k are enabled；
A5, Ry (i, j)=Ro (a, b), Gy (i, j)=Go (a, b), By (i, j)=Bo (a, b) are enabled；
A6, it enables the value of i add 1, as i > m, enters step A7, otherwise return step A4；
A7, i=1 is enabled, the value of j adds 1, as j > n, enters step A8, otherwise return step A4；
A8, sampled after image pixel and back to programmable gate array.
Further: the specific steps of surface cracks primary dcreening operation program in the step S5 and S10 are as follows:
B1, it calls snowfield to judge subprogram, obtains the value of intermediate parameters Fx；
B2, as Fx=1, enter step B3, otherwise enter step B4；
B3, snowfield surface cracks hidden danger primary dcreening operation subprogram is called to handle snowfield image, obtain snowfield image contains crack two
It is worth image, enters step B7；
B4, it calls meadow to judge subprogram, obtains the value of intermediate parameters Fc；
B5, as Fc=1, enter step B6, otherwise determine image be not belonging to calculate scope, terminate this program；
B6, meadow surface cracks hidden danger primary dcreening operation subprogram treating meadow image is called, obtain meadow image contains crack two
It is worth image；
B7, it calls primary dcreening operation subprogram in waters to handle river image, obtains river region image；
B8, surface cracks hidden danger pixel is called to judge subprogram to the two-value containing crack of river region image, snowfield image
The bianry image containing crack of image or meadow image obtains binaryzation surface cracks hidden danger image；
B9, to binaryzation surface cracks hidden danger image call surface cracks hidden danger pixel counts subprogram, obtain earth's surface and split
Hidden danger pixel quantity is stitched, this program is terminated.
Further: snowfield judges the specific steps of subprogram in the step B1 are as follows:
C1, snowfield RGB color pixel is converted to gray-scale pixels, forms gray level image；
C2, two-dimensional Fourier transform is carried out to gray level image, forms the two-dimentional Fourier data of gray level image；
The image of C3, the two-dimentional Fourier data positive axis part for enabling gray level image and minus half shaft portion are respectively about respective
Central symmetry, obtain central symmetry Fourier data；
C4, logarithm operation is carried out to the frequency spectrum of central symmetry Fourier data, obtains log-magnitude spectrum, log-magnitude spectrum picture
Element is A (i, j), and i=1,2,3 ... x, x are horizontal pixel number or sampling horizontal pixel number, and j=1,2,3 ... y, y are longitudinal
Pixel number samples longitudinal pixel number；
C5, longitudinal differential signal, pixel C are calculated by log-magnitude spectrum pixel A (i, j)z(i, j), i=1,2,
3 ... x, j=1,2,3 ... y, the calculation formula of longitudinal differential signal are as follows:
If i=1 or x, Cz(i, j)=A (i, j)；
If j=1 or y, Cz(i, j)=A (i, j)；
In the case of remaining, then Cz(i, j)=2A (i, j)-A (i, j-1)-A (i, j+1)；
C6, lateral differential signal, pixel C are calculated by log-magnitude spectrum pixel A (i, j)h(i, j), i=1,2,
3 ... x, j=1,2,3 ... y, the calculation formula of lateral differential signal are as follows:
If i=1 or x, Ch(i, j)=A (i, j)；
If j=1 or y, Ch(i, j)=A (i, j)；
In the case of remaining, then Ch(i, j)=2A (i, j)-A (i-1, j)-A (i+1, j)；
C7, the lateral summing signal S for calculating longitudinal differential signalzh(i) and longitudinal summing signal Szz(i), calculation formula
Are as follows:
C8, the lateral summing signal S for calculating lateral differential signalhh(i) and longitudinal summing signal Shz(i), calculation formula
Are as follows:
C9, site Nh, calculation formula in site Nz and transverse direction are calculated in longitudinal direction are as follows:
Nz=[y/2]
Nh=[x/2]
In above formula, [] is rounding operation；
C10, in the range of i=Nb to y-Nb, search the lateral summing signal S of longitudinal differential signalzh(i) maximum in
Value, works as Szh(i) be maximum value when, i=Wzh；
Nb is that edge is ignored a little；
C11, as Nz-Nb < WzhWhen < Nz+Nb, C12 is entered step, C27 is otherwise entered step；
C12, MAX is enabledzh=Szh(Wzh), and calculate the lateral summation snowfield criterion XD of longitudinal differential signalzh, calculation formula
Are as follows:
In above formula, PzhFor the lateral sum average value of longitudinal differential signal, | | expression takes absolute value；
C13, work as XDzhGreater than the lateral summation snowfield thresholding MIN of longitudinal differential signalzhxdWhen, C14 is entered step, otherwise
Enter step C27；
C14, in the range of i=Nb to x-Nb, search longitudinal summing signal S of longitudinal differential signalzz(i) minimum in
Value, works as Szz(i) be minimum value when, i=Wzz；
C15, as Nh-Nb < WzzWhen < Nh+Nb, C16 is entered step, C27 is otherwise entered step；
C16, MIN is enabledzz=Szz(Wzz), and calculate longitudinal summation snowfield criterion XD of longitudinal differential signalzz, calculation formula
Are as follows:
In above formula, PzzFor longitudinal sum average value of longitudinal differential signal；
C17, work as XDzzGreater than longitudinal summation snowfield thresholding MIN of longitudinal differential signalzzxdWhen, C18 is entered step, otherwise
Enter step C27；
C18, in the range of i=Nb to x-Nb, search longitudinal summing signal S of lateral differential signalhz(i) maximum in
Value, works as Shz(i) be maximum value when, i=Whz；
C19, as Nh-Nb < WhzWhen < Nh+Nb, C20 is entered step, C27 is otherwise entered step；
C20, MAX is enabledhz=Shz(Whz), and calculate longitudinal summation snowfield criterion XD of lateral differential signalhz, calculation formula
Are as follows:
In above formula, PhzFor longitudinal sum average value of lateral differential signal；
C21, work as XDhzGreater than longitudinal summation snowfield thresholding MIN of lateral differential signalhzxdWhen, C22 is entered step, otherwise
Enter step C27；
C22, in the range of i=Nb to y-Nb, search the lateral summing signal S of lateral differential signalhh(i) minimum in
Value, works as Shh(i) be minimum value when, i=Whh；
C23, as Nz-Nb < WhhWhen < Nz+Nb, C24 is entered step, C27 is otherwise entered step；
C24, MIN is enabledhh=Shh(Whh), and calculate the lateral summation snowfield criterion XD of lateral differential signalhh, calculation formula
Are as follows:
In above formula, PhhFor the lateral sum average value of lateral differential signal；
C25, work as XDhhThe lateral summation snowfield thresholding MIN of lateral differential signalhhxdWhen, C26 is entered step, is otherwise entered
Step C27；
C26, Fx=1 is enabled, enters step B2；
C27, Fx=0 is enabled, enters step B2.
Further: the specific steps of snowfield surface cracks hidden danger primary dcreening operation subprogram in the step B3 are as follows:
D1, snowfield image is subjected to greyscale transformation, and is converted to 8 shaping gradation datas；
D2,8 shaping gradation data histograms of snowfield are obtained according to 8 shaping gradation datas；
D3, the maximum value searched in 8 shaping gradation data histograms of snowfield less than snowfield histogram boundary XDZF are corresponding
Brightness value XD1；
D4, the maximum value searched in 8 shaping gradation data histograms of snowfield greater than snowfield histogram boundary XDZF are corresponding
Brightness value XD2；
D5, threshold value YZXD, calculation formula are calculated according to brightness value XD1 and XD2 are as follows:
YZXD=0.5 (XD1+XD2)；
D6, binary conversion treatment is done to 8 shaping gradation datas of snowfield according to threshold value YZXD, obtains each pixel after binaryzation
Data CDZH (i, j), and bianry image Icdlf is constituted by pixel data CDZH (i, j)；
D7, the connected domain for calculating bianry image Icdlf；
D8, each connected domain edge is calculated to the maximum value LMAX (i), i=1,2 of centroid distance, 3 ... nn, nn are connection
The number in domain；
The minimum value LMIN (i) of D9, each connected domain edge of calculating to centroid distance；
D10, the long and narrow than XCB (i) of each connected domain is calculated by maximum value LMAX (i) and minimum value LMIN (i), calculate
Formula are as follows:
XCB (i)=LMAX (i)/LMIN (i)；
D11, when XCB (i) be greater than crack thresholding LFMX, and the pixel of connected domain be greater than crack minimum pixel LFZXXS company
Logical domain is crack, and will be assigned a value of 1 for the connected domain in crack, remaining connected domain is assigned a value of 0, obtains bianry image containing crack
Iclf。
Further: meadow judges the specific steps of subprogram in the step B4 are as follows:
E1, meadow enhancing image is calculated, meadow enhances the calculation formula of image are as follows:
In above formula, CZ (i1,j1) it is the brightness that meadow enhances each pixel of image, i1=1,2,3 ... ... x, j1=1,2,
3 ... ... y, x are that horizontal pixel number perhaps samples horizontal pixel number y and is longitudinal pixel number or samples longitudinal pixel number, Ro(i1,
j1) be sensing circuit original pixels image rgb signal in danger signal, Bo(i1,j1) be rgb signal in green,
Go(i1,j1) be rgb signal in blue signal；
E2, enhance image calculating meadow enhancing image histogram according to meadow, and find out the maximum value MAXcd of histogram；
E3, when MAXcd be greater than meadow judge threshold value MINcd when, enable Fc=1, otherwise enable Fc=0.
Further: the specific steps of meadow surface cracks hidden danger primary dcreening operation subprogram in the step B6 are as follows:
F1, meadow enhancing image is converted into 8 shape datas；
F2, enhancing 8 shape data histograms of image in meadow are obtained according to 8 shape datas；
Brightness value corresponding to maximum value in F3, lookup meadow enhancing 8 shape data histograms of image less than 200
LD1；
Brightness value corresponding to minimum value in F4, lookup meadow enhancing 8 shape data histograms of image greater than 200
LD2；
F5, threshold value YZld, calculation formula are calculated by brightness value L D1 and brightness value L D2 are as follows:
YZld=0.5 (LD1+LD2)；
F6, according to threshold value YZld, binary conversion treatment is done to meadow enhancing 8 shape datas of image, obtains binary conversion treatment
Each pixel data CDHZ (i, j) afterwards, and bianry image Icdlf1 is constituted by pixel data CDHZ (i, j), calculate binary map
As the connected domain of Icdlf1；
F7, each connected domain edge is calculated to the maximum value MMAX (i), i=1,2 of centroid distance, 3 ... nn, nn are connection
The number in domain；
The minimum M MIN (i) of F8, each connected domain edge of calculating to centroid distance；
F9, the long and narrow than MXCB (i) of each connected domain is calculated by maximum value MMAX (i) and minimum M MIN (i), calculate
Formula are as follows:
MXCB (i)=MMAX (i)/MMIN (i)；
F10, it is greater than crack thresholding LFMX as MXCB (i), and the pixel of connected domain is greater than crack minimum pixel LFZXXS
Connected domain is crack, and will be assigned a value of 1 for the connected domain in crack, remaining connected domain is assigned a value of 0, obtains bianry image containing crack
Iclf1。
Further: the specific steps of waters primary dcreening operation subprogram in the step B7 are as follows:
G1, color image is converted to 8 gray level images；
G2, binarization threshold Xotsu is calculated based on maximum variance between clusters；
G3, binary conversion treatment is carried out to 8 gray level images according to threshold X otsu, the image after enabling binary conversion treatment is I1；
G4, image I2 is obtained to image I1 progress erosion operation using waters corrosion structure member SYfs；
G5, image I3 is obtained to image I2 progress closed operation using waters closed operation structural elements SYbi；
G6, the connected domain for calculating image I3, it is 0 by the pixel intensity in connected domain, even that taking connected domain maximum, which is river,
Leading to overseas pixel intensity is 1, the river region image Ih1 that obtains that treated；
Surface cracks hidden danger pixel judges the specific steps of subprogram in the step B8 are as follows:
H1, river region image Ih1, the Iclf of bianry image containing crack or the Iclf1 of bianry image containing crack are carried out and transported
It calculates, obtains image Iand；
H2, closed operation is carried out to image Iand using crack closed operation structural elements LFbi, it is hidden obtains binaryzation surface cracks
Suffer from image Izhyh；
The specific steps of surface cracks hidden danger pixel counts subprogram in the step B9 are as follows:
The quantity for being 1 using binaryzation crack hidden danger image Izhyh intermediate value is as surface cracks hidden danger pixel quantity.
The invention has the benefit that first the unmanned plane in unloaded subsystem carries out surface cracks primary dcreening operation to the present invention, it will
Primary dcreening operation discovery has the picture transfer of surface cracks to ground, is further calculated by ground system.Ground system is by geology calamity
Evil hidden danger picture is identified, and reminds artificial treatment, and beneficial effects of the present invention are as follows:
1, surface cracks are commented in fast quick checking, are conducive to quickly find surface cracks；
2, using artificial intelligence technology, manual operation is reduced；
3, vacant lot one, cooperated computing, raising, which is looked into, comments efficiency；
4, the picture for having surface cracks hidden danger is only transmitted in wireless transmission, greatly reduces wireless transmission capacity；
5, the surface cracks on Fast Identification meadow and snowfield.
Detailed description of the invention
Fig. 1 is present system block diagram；
Fig. 2 is the structure chart of image sensing submodule in the present invention；
Fig. 3 is flow chart of the present invention.
Specific embodiment
A specific embodiment of the invention is described below, in order to facilitate understanding by those skilled in the art this hair
It is bright, it should be apparent that the present invention is not limited to the ranges of specific embodiment, for those skilled in the art,
As long as various change is in the spirit and scope of the present invention that the attached claims limit and determine, these variations are aobvious and easy
See, all are using the innovation and creation of present inventive concept in the column of protection.
As shown in Figure 1, a kind of surface cracks rapid detection system based on unmanned plane, including by wireless transmission method into
The ground subsystem of row communication and unloaded subsystem；
Acquired image information data is carried out initial treatment for acquiring image information data by unloaded subsystem, and
Send the data to ground subsystem；
Ground subsystem, the image data sent for receiving unloaded subsystem, carries out final process for image data, and
Storing data.Unloaded subsystem includes unmanned plane and image processing module, and described image processing module is arranged in uav bottom.
Image processing module includes image sensing submodule, programmable gate array, the second wireless transmission submodule and video compress
Chip is handled, image sensing submodule, Video compression chip and the second wireless transmission submodule are and programmable gate
Array connection.As shown in Fig. 2, image sensing submodule includes mounting box, camera lens and sensing circuit, sensing electricity is equipped in mounting box
Road, the side of mounting box are fixed on uav bottom, and the other side of mounting box is connect by circular wire mouth with camera lens.
In embodiments of the present invention, the pixel of image sensing circuit inductance is laterally x pixel, and longitudinal is y pixel (x >=y),
The length and width of sensing pixels are all w；The focal length of camera lens is z.Then the width in image sensing circuit inductance face is xw, is highly yw.
In embodiments of the present invention, image sensing circuit (imaging sensor) uses the ICX205 of Sony Corporation, the sensing
The pixel of device is laterally 1360 pixels, and longitudinal is 1024 pixels, the length of sensing pixels and it is wide be all 4.65 μm, camera lens is using Japan
Totatsu Corp.'s industrial lens, model TAMRON 17H, focal length 16mm.
In embodiments of the present invention, programmable gate array uses the X3S1200 of Xilnx company.
In embodiments of the present invention, video compress chip uses the TE3310 of TOKYO company.
In embodiments of the present invention, wireless transmission submodule is using Shanghai Shun Zhou intelligence Science and Technology Co., Ltd.
The zigbee module of SZ05-L-PRO-3.
Programmable gate array and peripheral circuit are used to control the operating mode of image sensing circuit, receive image sensing electricity
Road signal handles image sensing circuit signal, and by treated, image sensing circuit data gives Video compression
Chip and its peripheral circuit receive Video compression chip and the compressed image data of its peripheral circuit, will compress
Image data afterwards is given the second wireless transmission submodule and is wirelessly transferred.
Pixel samples program and surface cracks primary dcreening operation program, surface cracks primary dcreening operation program are equipped in Programmadle logic gate array
Judge that subprogram, snowfield surface cracks hidden danger primary dcreening operation subprogram, meadow judge subprogram, meadow surface cracks hidden danger including snowfield
Primary dcreening operation subprogram, waters primary dcreening operation subprogram, surface cracks hidden danger pixel judge subprogram and surface cracks hidden danger pixel counts
Program, pixel samples program is for being sampled image pixel, and surface cracks primary dcreening operation program is for calculating surface cracks hidden danger
Pixel quantity, snowfield judge subprogram for obtaining intermediate parameters Fx, and snowfield surface cracks hidden danger primary dcreening operation subprogram is for obtaining
The bianry image containing crack of snowfield image, meadow judge subroutine call to intermediate parameters Fc, the first sieve of meadow surface cracks hidden danger
Program is for obtaining the bianry image containing crack of meadow image, and waters primary dcreening operation subprogram is for obtaining river region image, earth's surface
Crack hidden danger pixel judges subprogram for obtaining binaryzation surface cracks hidden danger image, the sub- journey of surface cracks hidden danger pixel counts
Sequence is for obtaining surface cracks hidden danger pixel quantity；Surface cracks primary dcreening operation program also is provided in computer.
As shown in figure 3, a kind of surface cracks rapid detection method based on unmanned plane, comprising:
S1, the initial parameter that programmable gate array in unloaded subsystem is set.Initial parameter includes sampling rate k, horizontal
To pixel number x, point Nb, the lateral summation snowfield thresholding of longitudinal differential signal are ignored in longitudinal pixel number y, threshold value Nmin, edge
MINzhxd, longitudinal summation snowfield thresholding MIN of longitudinal differential signalzzxd, longitudinal summation snowfield thresholding of lateral differential signal
MINhzxd, the lateral summation snowfield thresholding MIN of lateral differential signalhhxd, snowfield histogram boundary XDZF；Waters corrosion structure member
SYfs, waters closed operation structural elements SYbi, crack closed operation structural elements LFbi, crack thresholding LFMX.
S2, control signal is issued to sensing circuit by programmable gate array.
S3, the pixel intensity signal that sensing circuit is received by programmable gate array, and to the pixel of sensing circuit
Luminance signal carries out Bayer image processing and white balance processing, obtains white balance treated the rgb signal of each pixel.
S4, the rgb signal for calling pixel samples subprogram processing white balance treated each pixel, after being sampled
Image pixel.
The specific steps of pixel samples program are as follows:
A1, the danger signal in the rgb signal of image sensing circuit original pixels is set as Ro (i1,j1), green Bo
(i1,j1), blue signal is Go (i1,j1)；
i1=1,2,3 ... x, j1=1,2,3 ... y, x are horizontal pixel number, and y is longitudinal pixel number；
A2, set sampling after image pixel rgb signal in danger signal as Ry (i2,j2), green is By (i2,
j2), blue signal is Gy (i2,j2)；
i2=1,2,3 ... m, j2=1,2,3 ... n, m=[x/k], n=[y/k], [] indicate that rounding operation, k are sampling
Rate, and k is integer, m is sampling horizontal pixel number, and n is to sample longitudinal pixel number；
A3, initial parameter i=1, j=1 are enabled；
A4, initial parameter a=i × k, b=j × k are enabled；
A5, Ry (i, j)=Ro (a, b), Gy (i, j)=Go (a, b), By (i, j)=Bo (a, b) are enabled；
A6, it enables the value of i add 1, as i > m, enters step A7, otherwise return step A4；
A7, i=1 is enabled, the value of j is enabled to add 1, as j > n, enter step A8, otherwise return step A4；
A8, sampled after image pixel and back to programmable gate array.
S5, the image pixel after the processing sampling of surface cracks primary dcreening operation subprogram is called by programmable gate array, obtain
To surface cracks hidden danger pixel quantity.Surface cracks primary dcreening operation subprogram in the surface cracks primary dcreening operation subprogram and step S10 walks
It is rapid identical, it is only necessary to the danger signal Ro (i, j) in program will to be replaced with Ry (i, j), green Bo (i, j) replaces with By
(i, j), blue signal Bo (i, j) replace with By (i, j)；I=1,2,3 ... ... x replace with i=1,2,3 ... ... m, j=1,2,
3 ... ... y replace with j=1, and 2,3 ... ... n, x replace with m, and y replaces with n.
S6, judge whether surface cracks hidden danger pixel quantity is greater than threshold value, if so, entering step S7, otherwise, return to step
Rapid S2.Threshold value is determined by experiment.
S7, the rgb format picture signal of all pixels is sent to Video compression chip, obtains compressed image
Data.
S8, compressed image data is received by programmable gate array, and compressed image data is sent
To the second wireless transmission submodule, the second wireless transmission submodule will press transmission of data to be sent to the first wireless transmission submodule.
S9, compressed image data is sent to by computer by the first wireless transmission submodule in ground subsystem,
And compressed image data is decompressed by computer, the image data after being decompressed.
S10, the image data after surface cracks primary dcreening operation subprogram processing decompression is called by computer, obtain surface cracks
Testing result.
The specific steps of surface cracks primary dcreening operation program are as follows:
B1, it calls snowfield to judge subprogram, obtains the value of intermediate parameters Fx；
Snowfield judges the specific steps of subprogram are as follows:
C1, snowfield RGB color pixel is converted to gray-scale pixels, forms gray level image；
C2, two-dimensional Fourier transform is carried out to gray level image, forms the two-dimentional Fourier data of gray level image；
The image of C3, the two-dimentional Fourier data positive axis part for enabling gray level image and minus half shaft portion are respectively about respective
Central symmetry, obtain central symmetry Fourier data；
C4, logarithm operation is carried out to the frequency spectrum of central symmetry Fourier data, obtains log-magnitude spectrum, log-magnitude spectrum picture
Element is A (i, j), and i=1,2,3 ... x, j=1,2,3 ... y, x are horizontal pixel number or sampling horizontal pixel number, and y is longitudinal
Pixel number samples longitudinal pixel number；
C5, longitudinal differential signal, pixel C are calculated by log-magnitude spectrum pixel A (i, j)z(i, j), i=1,2,
3 ... x, j=1,2,3 ... y, the calculation formula of longitudinal differential signal are as follows:
If i=1 or x, Cz(i, j)=A (i, j)；
If j=1 or y, Cz(i, j)=A (i, j)；
In the case of remaining, then Cz(i, j)=2A (i, j)-A (i, j-1)-A (i, j+1)；
C6, lateral differential signal, pixel C are calculated by log-magnitude spectrum pixel A (i, j)h(i, j), i=1,2,
3 ... x, j=1,2,3 ... y, the calculation formula of lateral differential signal are as follows:
If i=1 or x, Ch(i, j)=A (i, j)；
If j=1 or y, Ch(i, j)=A (i, j)；
In the case of remaining, then Ch(i, j)=2A (i, j)-A (i-1, j)-A (i+1, j)；
C7, the lateral summing signal S for calculating longitudinal differential signalzh(i) and longitudinal summing signal Szz(i), calculation formula
Are as follows:
C8, the lateral summing signal S for calculating lateral differential signalhh(i) and longitudinal summing signal Shz(i), calculation formula
Are as follows:
C9, site Nh, calculation formula in site Nz and transverse direction are calculated in longitudinal direction are as follows:
Nz=[y/2]
Nh=[x/2]
In above formula, [] is rounding operation；
C10, in the range of i=Nb to y-Nb, search the lateral summing signal S of longitudinal differential signalzh(i) maximum in
Value, works as Szh(i) be maximum value when, i=Wzh；Nb is that edge is ignored a little, is determined according to experiment；
C11, as Nz-Nb < WzhWhen < Nz+Nb, C12 is entered step, C27 is otherwise entered step；
C12, MAX is enabledzh=Szh(Wzh), and calculate the lateral summation snowfield criterion XD of longitudinal differential signalzh,
Calculation formula are as follows:
In above formula, PzhFor the lateral sum average value of longitudinal differential signal, | | expression takes absolute value；
A113, work as XDzhGreater than the lateral summation snowfield thresholding MIN of longitudinal differential signalzhxdWhen, C14 is entered step, otherwise
Enter step C27；
C14, in the range of i=Nb to x-Nb, search longitudinal summing signal S of longitudinal differential signalzz(i) minimum in
Value, works as Szz(i) be minimum value when, i=Wzz；
C15, as Nh-Nb < WzzWhen < Nh+Nb, C16 is entered step, C27 is otherwise entered step；
C16, MIN is enabledzz=Szz(Wzz), and calculate longitudinal summation snowfield criterion XD of longitudinal differential signalzz, calculation formula
Are as follows:
In above formula, PzzFor longitudinal sum average value of longitudinal differential signal；
C17, work as XDzzGreater than longitudinal summation snowfield thresholding MIN of longitudinal differential signalzzxdWhen, C18 is entered step, otherwise
Enter step C27；
C18, in the range of i=Nb to x-Nb, search longitudinal summing signal S of lateral differential signalhz(i) maximum in
Value, works as Shz(i) be maximum value when, i=Whz；
C19, as Nh-Nb < WhzWhen < Nh+Nb, C20 is entered step, C27 is otherwise entered step；
C20, MAX is enabledhz=Shz(Whz), and calculate longitudinal summation snowfield criterion XD of lateral differential signalhz, calculation formula
Are as follows:
In above formula, PhzFor longitudinal sum average value of lateral differential signal；
C21, work as XDhzGreater than longitudinal summation snowfield thresholding MIN of lateral differential signalhzxdWhen, C22 is entered step, otherwise
Enter step C27；
C22, in the range of i=Nb to y-Nb, search the lateral summing signal S of lateral differential signalhh(i) minimum in
Value, works as Shh(i) be minimum value when, i=Whh；
C23, as Nz-Nb < WhhWhen < Nz+Nb, C24 is entered step, C27 is otherwise entered step；
C24, MIN is enabledhh=Shh(Whh), and calculate the lateral summation snowfield criterion XD of lateral differential signalhh, calculation formula
Are as follows:
In above formula, PhhFor the lateral sum average value of lateral differential signal；
C25, work as XDhhThe lateral summation snowfield thresholding MIN of lateral differential signalhhxdWhen, C26 is entered step, is otherwise entered
Step C27；
C26, Fx=1 is enabled, enters step B2；
C7, Fx=0 is enabled, enters step B2.
B2, as Fx=1, enter step B3, otherwise enter step B4；
B3, snowfield surface cracks hidden danger primary dcreening operation subprogram is called to handle snowfield image, obtain snowfield image contains crack two
It is worth image, enters step B7；
The specific steps of snowfield surface cracks hidden danger primary dcreening operation subprogram are as follows:
D1, snowfield image is subjected to greyscale transformation, and is converted to 8 shaping gradation datas；
D2,8 shaping gradation data histograms of snowfield are obtained according to 8 shaping gradation datas；
D3, the maximum value searched in 8 shaping gradation data histograms of snowfield less than snowfield histogram boundary XDZF are corresponding
Brightness value XD1, snowfield histogram demarcate XDZF according to experiment determine；
D4, the maximum value searched in 8 shaping gradation data histograms of snowfield greater than snowfield histogram boundary XDZF are corresponding
Brightness value XD2；
D5, threshold value YZXD, calculation formula are calculated according to brightness value XD1 and XD2 are as follows:
YZXD=0.5 (XD1+XD2)；
D6, binary conversion treatment is done to 8 shaping gradation datas of snowfield according to threshold value YZXD, obtains each pixel after binaryzation
Data CDZH (i, j), and bianry image Icdlf is constituted by pixel data CDZH (i, j)；
D7, the connected domain for calculating bianry image Icdlf；
D8, each connected domain edge is calculated to the maximum value LMAX (i), i=1,2 of centroid distance, 3 ... nn, nn are connection
The number in domain；
The minimum value LMIN (i) of D9, each connected domain edge of calculating to centroid distance；
D10, the long and narrow than XCB (i) of each connected domain is calculated by maximum value LMAX (i) and minimum value LMIN (i), calculate
Formula are as follows:
XCB (i)=LMAX (i)/LMIN (i)；
D11, when XCB (i) be greater than crack thresholding LFMX, and the pixel of connected domain be greater than crack minimum pixel LFZXXS company
Logical domain is crack, and will be assigned a value of 1 for the connected domain in crack, remaining connected domain is assigned a value of 0, obtains bianry image containing crack
Iclf, crack thresholding LFMX and crack minimum pixel LFZXXS are determined all in accordance with experiment.
B4, it calls meadow to judge subprogram, obtains the value of intermediate parameters Fc；
Meadow judges the specific steps of subprogram are as follows:
E1, meadow enhancing image is calculated, meadow enhances the calculation formula of image are as follows:
In above formula, CZ (i1,j1) it is the brightness that meadow enhances each pixel of image, i1=1,2,3 ... ... x, j1=1,2,
3 ... ... y, x are that horizontal pixel number perhaps samples horizontal pixel number y and is longitudinal pixel number or samples longitudinal pixel number, Ro(i1,
j1) be sensing circuit original pixels image rgb signal in danger signal, Bo(i1,j1) be rgb signal in green,
Go(i1,j1) be rgb signal in blue signal；
E2, enhance image calculating meadow enhancing image histogram according to meadow, and find out the maximum value MAXcd of histogram；
E3, when MAXcd be greater than meadow judge threshold value MINcd when, enable Fc=1, otherwise enable Fc=0, meadow judges thresholding
Value MINcd is determined according to experiment.
B5, as Fc=1, enter step B6, otherwise determine image be not belonging to calculate scope, terminate this program；
B6, meadow surface cracks hidden danger primary dcreening operation subprogram treating meadow image is called, obtain meadow image contains crack two
It is worth image；
The specific steps of meadow surface cracks hidden danger primary dcreening operation subprogram are as follows:
F1, meadow enhancing image is converted into 8 shape datas；
F2, enhancing 8 shape data histograms of image in meadow are obtained according to 8 shape datas；
Brightness value corresponding to maximum value in F3, lookup meadow enhancing 8 shape data histograms of image less than 200
LD1；
Brightness value corresponding to minimum value in F4, lookup meadow enhancing 8 shape data histograms of image greater than 200
LD2；
F5, threshold value YZld, calculation formula are calculated by brightness value L D1 and brightness value L D2 are as follows:
YZld=0.5 (LD1+LD2)；
F6, according to threshold value YZld, binary conversion treatment is done to meadow enhancing 8 shape datas of image, obtains binary conversion treatment
Each pixel data CDHZ (i, j) afterwards, and bianry image Icdlf1 is constituted by pixel data CDHZ (i, j), calculate binary map
As the connected domain of Icdlf1；
F7, each connected domain edge is calculated to the maximum value MMAX (i), i=1,2 of centroid distance, 3 ... nn, nn are connection
The number in domain；
The minimum M MIN (i) of F8, each connected domain edge of calculating to centroid distance；
F9, the long and narrow than MXCB (i) of each connected domain is calculated by maximum value MMAX (i) and minimum M MIN (i), calculate
Formula are as follows:
MXCB (i)=MMAX (i)/MMIN (i)；
F10, it is greater than crack thresholding LFMX as MXCB (i), and the pixel of connected domain is greater than crack minimum pixel LFZXXS
Connected domain is crack, and will be assigned a value of 1 for the connected domain in crack, remaining connected domain is assigned a value of 0, obtains bianry image containing crack
Iclf1, crack thresholding LFMX and crack minimum pixel LFZXXS are determined all in accordance with experiment.
B7, it calls primary dcreening operation subprogram in waters to handle river image, obtains river region image；
The specific steps of waters primary dcreening operation subprogram are as follows:
G1, color image is converted to 8 gray level images；
G2, binarization threshold Xotsu is calculated based on maximum variance between clusters；
G3, binary conversion treatment is carried out to 8 gray level images according to threshold X otsu, the image after enabling binary conversion treatment is I1；
G4, image I2, waters corrosion structure are obtained to image I1 progress erosion operation using waters corrosion structure member SYfs
First SYfs is determined according to experiment；
G5, image I3, waters closed operation knot are obtained to image I2 progress closed operation using waters closed operation structural elements SYbi
Constitutive element SYbi is determined according to experiment；
G6, the connected domain for calculating image I3, it is 0 by the pixel intensity in connected domain, even that taking connected domain maximum, which is river,
Leading to overseas pixel intensity is 1, the river region image Ih1 that obtains that treated.
B8, surface cracks hidden danger pixel is called to judge subprogram to the two-value containing crack of river region image, snowfield image
The bianry image containing crack of image or meadow image obtains binaryzation surface cracks hidden danger image；
Surface cracks hidden danger pixel judges the specific steps of subprogram are as follows:
H1, river region image Ih1, the Iclf of bianry image containing crack or the Iclf1 of bianry image containing crack are carried out and transported
It calculates, obtains image Iand；
H2, closed operation is carried out to image Iand using crack closed operation structural elements LFbi, it is hidden obtains binaryzation surface cracks
Suffer from image Izhyh, crack closed operation structural elements LFbi is determined according to experiment.
B9, to binaryzation surface cracks hidden danger image call surface cracks hidden danger pixel counts subprogram, obtain earth's surface and split
Hidden danger pixel quantity is stitched, this program is terminated.
The specific steps of surface cracks hidden danger pixel counts subprogram are as follows:
The quantity for being 1 using binaryzation crack hidden danger image Izhyh intermediate value is as surface cracks hidden danger pixel quantity.
In formula of the present invention, | | indicate the operation that takes absolute value, [] indicates rounding operation, does not illustrate the threshold value or meter of judgement
Structural elements are calculated to be determined by experiment.
Claims (10)
1. a kind of surface cracks rapid detection system based on unmanned plane, which is characterized in that including by wireless transmission method into
The ground subsystem of row communication and unloaded subsystem；
Acquired image information data is carried out initial treatment, and will count for acquiring image information data by unloaded subsystem
According to being sent to ground subsystem；
Ground subsystem, the image data sent for receiving unloaded subsystem, carries out final process for image data, and store
Data；
The ground subsystem includes computer interconnected and the first wireless transmission submodule；It is described zero load subsystem include
Unmanned plane and image processing module, described image processing module are arranged in uav bottom；Described image processing module includes figure
As sensing submodule, programmable gate array, the second wireless transmission submodule and Video compression chip, described image pass
Sense submodule, Video compression chip and the second wireless transmission submodule are connect with programmable gate array；The figure
As sensing submodule includes mounting box, camera lens and sensing circuit, it is equipped with sensing circuit in the mounting box, the one of the mounting box
Side is fixed on uav bottom, and the other side of the mounting box is connect by circular wire mouth with camera lens.
2. the surface cracks rapid detection system according to claim 1 based on unmanned plane, which is characterized in that described to compile
It collects and is equipped with pixel samples program and surface cracks primary dcreening operation program in logic gate array, the surface cracks primary dcreening operation program includes snowfield
Judge that subprogram, snowfield surface cracks hidden danger primary dcreening operation subprogram, meadow judge subprogram, meadow surface cracks hidden danger just sieve journey
Sequence, waters primary dcreening operation subprogram, surface cracks hidden danger pixel judge subprogram and surface cracks hidden danger pixel counts subprogram, described
Pixel samples program is for being sampled image pixel, and the surface cracks primary dcreening operation program is for calculating surface cracks hidden danger picture
Prime number amount, the snowfield judge that subprogram is used for obtaining intermediate parameters Fx, the snowfield surface cracks hidden danger primary dcreening operation subprogram
In obtaining the bianry image containing crack of snowfield image, the meadow judges subroutine call to intermediate parameters Fc, the meadow earth's surface
Crack hidden danger primary dcreening operation subprogram is for obtaining the bianry image containing crack of meadow image, and the waters primary dcreening operation subprogram is for obtaining
River region image, the surface cracks hidden danger pixel judge subprogram for obtaining binaryzation surface cracks hidden danger image, institute
Surface cracks hidden danger pixel counts subprogram is stated for obtaining surface cracks hidden danger pixel quantity；It also is provided with ground in the computer
Table crack primary dcreening operation program.
3. a kind of surface cracks rapid detection method based on unmanned plane characterized by comprising
S1, the initial parameter that programmable gate array in unloaded subsystem is set；
S2, control signal is issued to sensing circuit by programmable gate array；
S3, the pixel intensity signal that sensing circuit is received by programmable gate array, and to the pixel intensity of sensing circuit
Signal carries out Bayer image processing and white balance processing, obtains white balance treated the rgb signal of each pixel；
S4, the rgb signal for calling pixel samples program processing white balance treated each pixel, the image slices after being sampled
Element；
S5, the image pixel after the processing sampling of surface cracks primary dcreening operation program is called by programmable gate array, obtain earth's surface
Crack hidden danger pixel quantity；
S6, judge whether surface cracks hidden danger pixel quantity is greater than threshold value, if so, S7 is entered step, otherwise, return step
S2；
S7, the rgb format picture signal of all pixels is sent to Video compression chip, obtains compressed picture number
According to；
S8, compressed image data is received by programmable gate array, and send for compressed image data
Two wireless transmission submodules, the second wireless transmission submodule will press transmission of data to be sent to the first wireless transmission submodule；
S9, compressed image data is sent to by computer by the first wireless transmission submodule in ground subsystem, and led to
It crosses computer to decompress compressed image data, the image data after being decompressed；
S10, the image data after surface cracks primary dcreening operation program processing decompression is called by computer, obtain surface cracks detection knot
Fruit.
4. the surface cracks rapid detection method according to claim 3 based on unmanned plane, which is characterized in that the step
The specific steps of pixel samples program in S4 are as follows:
A1, the danger signal in the rgb signal of image sensing circuit original pixels is set as Ro (i1,j1), green is Bo (i1,
j1), blue signal is Go (i1,j1)；
i1=1,2,3 ... x, j1=1,2,3 ... y, x are horizontal pixel number, and y is longitudinal pixel number；
A2, set sampling after image pixel rgb signal in danger signal as Ry (i2,j2), green is By (i2,j2), it is blue
Chrominance signal is Gy (i2,j2)；
i2=1,2,3 ... m, j2=1,2,3 ... n, m=[x/k], n=[y/k], [] are rounding operation, and k is sampling rate, and k is
Integer, m are sampling horizontal pixel number, and n is to sample longitudinal pixel number；
A3, initial parameter i=1, j=1 are enabled；
A4, initial parameter a=i × k, b=j × k are enabled；
A5, Ry (i, j)=Ro (a, b), Gy (i, j)=Go (a, b), By (i, j)=Bo (a, b) are enabled；
A6, it enables the value of i add 1, as i > m, enters step A7, otherwise return step A4；
A7, i=1 is enabled, the value of j adds 1, as j > n, enters step A8, otherwise return step A4；
A8, sampled after image pixel and back to programmable gate array.
5. the surface cracks rapid detection method according to claim 3 based on unmanned plane, which is characterized in that the step
The specific steps of surface cracks primary dcreening operation program in S5 and S10 are as follows:
B1, it calls snowfield to judge subprogram, obtains the value of intermediate parameters Fx；
B2, as Fx=1, enter step B3, otherwise enter step B4；
B3, it calls snowfield surface cracks hidden danger primary dcreening operation subprogram to handle snowfield image, obtains the binary map containing crack of snowfield image
Picture enters step B7；
B4, it calls meadow to judge subprogram, obtains the value of intermediate parameters Fc；
B5, as Fc=1, enter step B6, otherwise determine image be not belonging to calculate scope, terminate this program；
B6, meadow surface cracks hidden danger primary dcreening operation subprogram treating meadow image is called, obtains the binary map containing crack of meadow image
Picture；
B7, it calls primary dcreening operation subprogram in waters to handle river image, obtains river region image；
B8, surface cracks hidden danger pixel is called to judge that subprogram handles the binary map containing crack of river region image, snowfield image
The bianry image containing crack of picture or meadow image, obtains binaryzation surface cracks hidden danger image；
B9, to binaryzation surface cracks hidden danger image call surface cracks hidden danger pixel counts subprogram, it is hidden to obtain surface cracks
Suffer from pixel quantity, terminates this program.
6. the surface cracks rapid detection method according to claim 5 based on unmanned plane, which is characterized in that the step
Snowfield judges the specific steps of subprogram in B1 are as follows:
C1, snowfield RGB color pixel is converted to gray-scale pixels, forms gray level image；
C2, two-dimensional Fourier transform is carried out to gray level image, forms the two-dimentional Fourier data of gray level image；
The image of C3, the two-dimentional Fourier data positive axis part for enabling gray level image and minus half shaft portion are respectively about in respective
The heart is symmetrical, obtains central symmetry Fourier data；
C4, logarithm operation is carried out to the frequency spectrum of central symmetry Fourier data, obtains log-magnitude spectrum, log-magnitude spectrum pixel is
A (i, j), i=1,2,3 ... x, x are horizontal pixel number or sampling horizontal pixel number, and j=1,2,3 ... y, y are longitudinal pixel
Number samples longitudinal pixel number；
C5, longitudinal differential signal, pixel C are calculated by log-magnitude spectrum pixel A (i, j)z(i, j), i=1,2,3 ... x, j
=1,2,3 ... y, the calculation formula of longitudinal differential signal are as follows:
If i=1 or x, Cz(i, j)=A (i, j)；
If j=1 or y, Cz(i, j)=A (i, j)；
In the case of remaining, then Cz(i, j)=2A (i, j)-A (i, j-1)-A (i, j+1)；
C6, lateral differential signal, pixel C are calculated by log-magnitude spectrum pixel A (i, j)h(i, j), i=1,2,3 ... x, j
=1,2,3 ... y, the calculation formula of lateral differential signal are as follows:
If i=1 or x, Ch(i, j)=A (i, j)；
If j=1 or y, Ch(i, j)=A (i, j)；
In the case of remaining, then Ch(i, j)=2A (i, j)-A (i-1, j)-A (i+1, j)；
C7, the lateral summing signal S for calculating longitudinal differential signalzh(i) and longitudinal summing signal Szz(i), calculation formula are as follows:
C8, the lateral summing signal S for calculating lateral differential signalhh(i) and longitudinal summing signal Shz(i), calculation formula are as follows:
C9, site Nh, calculation formula in site Nz and transverse direction are calculated in longitudinal direction are as follows:
Nz=[y/2]
Nh=[x/2]
In above formula, [] is rounding operation；
C10, in the range of i=Nb to y-Nb, search the lateral summing signal S of longitudinal differential signalzh(i) maximum value in,
Work as Szh(i) be maximum value when, i=Wzh；
Nb is that edge is ignored a little；
C11, as Nz-Nb < WzhWhen < Nz+Nb, C12 is entered step, C27 is otherwise entered step；
C12, MAX is enabledzh=Szh(Wzh), and calculate the lateral summation snowfield criterion XD of longitudinal differential signalzh, calculation formula are as follows:
In above formula, PzhFor the lateral sum average value of longitudinal differential signal, | | expression takes absolute value；
C13, work as XDzhGreater than the lateral summation snowfield thresholding MIN of longitudinal differential signalzhxdWhen, C14 is entered step, is otherwise entered
Step C27；
C14, in the range of i=Nb to x-Nb, search longitudinal summing signal S of longitudinal differential signalzz(i) minimum value in,
Work as Szz(i) be minimum value when, i=Wzz；
C15, as Nh-Nb < WzzWhen < Nh+Nb, C16 is entered step, C27 is otherwise entered step；
C16, MIN is enabledzz=Szz(Wzz), and calculate longitudinal summation snowfield criterion XD of longitudinal differential signalzz, calculation formula are as follows:
In above formula, PzzFor longitudinal sum average value of longitudinal differential signal；
C17, work as XDzzGreater than longitudinal summation snowfield thresholding MIN of longitudinal differential signalzzxdWhen, C18 is entered step, is otherwise entered
Step C27；
C18, in the range of i=Nb to x-Nb, search longitudinal summing signal S of lateral differential signalhz(i) maximum value in,
Work as Shz(i) be maximum value when, i=Whz；
C19, as Nh-Nb < WhzWhen < Nh+Nb, C20 is entered step, C27 is otherwise entered step；
C20, MAX is enabledhz=Shz(Whz), and calculate longitudinal summation snowfield criterion XD of lateral differential signalhz, calculation formula are as follows:
In above formula, PhzFor longitudinal sum average value of lateral differential signal；
C21, work as XDhzGreater than longitudinal summation snowfield thresholding MIN of lateral differential signalhzxdWhen, C22 is entered step, is otherwise entered
Step C27；
C22, in the range of i=Nb to y-Nb, search the lateral summing signal S of lateral differential signalhh(i) minimum value in,
Work as Shh(i) be minimum value when, i=Whh；
C23, as Nz-Nb < WhhWhen < Nz+Nb, C24 is entered step, C27 is otherwise entered step；
C24, MIN is enabledhh=Shh(Whh), and calculate the lateral summation snowfield criterion XD of lateral differential signalhh, calculation formula are as follows:
In above formula, PhhFor the lateral sum average value of lateral differential signal；
C25, work as XDhhThe lateral summation snowfield thresholding MIN of lateral differential signalhhxdWhen, C26 is entered step, is otherwise entered step
C27；
C26, Fx=1 is enabled, enters step B2；
C27, Fx=0 is enabled, enters step B2.
7. the surface cracks rapid detection method according to claim 5 based on unmanned plane, which is characterized in that the step
The specific steps of snowfield surface cracks hidden danger primary dcreening operation subprogram in B3 are as follows:
D1, snowfield image is subjected to greyscale transformation, and is converted to 8 shaping gradation datas；
D2,8 shaping gradation data histograms of snowfield are obtained according to 8 shaping gradation datas；
D3, the maximum value searched in 8 shaping gradation data histograms of snowfield less than snowfield histogram boundary XDZF are corresponding bright
Angle value XD1；
D4, the maximum value searched in 8 shaping gradation data histograms of snowfield greater than snowfield histogram boundary XDZF are corresponding bright
Angle value XD2；
D5, threshold value YZXD, calculation formula are calculated according to brightness value XD1 and XD2 are as follows:
YZXD=0.5 (XD1+XD2)；
D6, binary conversion treatment is done to 8 shaping gradation datas of snowfield according to threshold value YZXD, obtains each pixel data after binaryzation
CDZH (i, j), and bianry image Icdlf is constituted by pixel data CDZH (i, j)；
D7, the connected domain for calculating bianry image Icdlf；
D8, each connected domain edge is calculated to the maximum value LMAX (i), i=1,2 of centroid distance, 3 ... nn, nn are connected domain
Number；
The minimum value LMIN (i) of D9, each connected domain edge of calculating to centroid distance；
D10, the long and narrow than XCB (i), calculation formula of each connected domain is calculated by maximum value LMAX (i) and minimum value LMIN (i)
Are as follows:
XCB (i)=LMAX (i)/LMIN (i)；
D11, when XCB (i) be greater than crack thresholding LFMX, and the pixel of connected domain be greater than crack minimum pixel LFZXXS connected domain
For crack, and it will be assigned a value of 1 for the connected domain in crack, remaining connected domain is assigned a value of 0, obtains the Iclf of bianry image containing crack.
8. the surface cracks rapid detection method according to claim 5 based on unmanned plane, which is characterized in that the step
Meadow judges the specific steps of subprogram in B4 are as follows:
E1, meadow enhancing image is calculated, meadow enhances the calculation formula of image are as follows:
In above formula, CZ (i1,j1) it is the brightness that meadow enhances each pixel of image, i1=1,2,3 ... ... x, j1=1,2,3 ...
Y, x are that horizontal pixel number perhaps samples horizontal pixel number y and is longitudinal pixel number or samples longitudinal pixel number, Ro(i1,j1) be
Danger signal in the rgb signal of sensing circuit original pixels image, Bo(i1,j1) be rgb signal in green, Go(i1,
j1) be rgb signal in blue signal；
E2, enhance image calculating meadow enhancing image histogram according to meadow, and find out the maximum value MAXcd of histogram；
E3, when MAXcd be greater than meadow judge threshold value MINcd when, enable Fc=1, otherwise enable Fc=0.
9. the surface cracks rapid detection method according to claim 5 based on unmanned plane, which is characterized in that the step
The specific steps of meadow surface cracks hidden danger primary dcreening operation subprogram in B6 are as follows:
F1, meadow enhancing image is converted into 8 shape datas；
F2, enhancing 8 shape data histograms of image in meadow are obtained according to 8 shape datas；
Brightness value L D1 corresponding to maximum value in F3, lookup meadow enhancing 8 shape data histograms of image less than 200；
Brightness value L D2 corresponding to minimum value in F4, lookup meadow enhancing 8 shape data histograms of image greater than 200；
F5, threshold value YZld, calculation formula are calculated by brightness value L D1 and brightness value L D2 are as follows:
YZld=0.5 (LD1+LD2)；
F6, according to threshold value YZld, binary conversion treatment is done to meadow enhancing 8 shape datas of image, is obtained every after binary conversion treatment
A pixel data CDHZ (i, j), and bianry image Icdlf1 is constituted by pixel data CDHZ (i, j), calculate bianry image
The connected domain of Icdlf1；
F7, each connected domain edge is calculated to the maximum value MMAX (i), i=1,2 of centroid distance, 3 ... nn, nn are connected domain
Number；
The minimum M MIN (i) of F8, each connected domain edge of calculating to centroid distance；
F9, the long and narrow than MXCB (i), calculation formula of each connected domain is calculated by maximum value MMAX (i) and minimum M MIN (i)
Are as follows:
MXCB (i)=MMAX (i)/MMIN (i)；
F10, when MXCB (i) be greater than crack thresholding LFMX, and the pixel of connected domain be greater than crack minimum pixel LFZXXS connection
Domain is crack, and will be assigned a value of 1 for the connected domain in crack, remaining connected domain is assigned a value of 0, obtains the Iclf1 of bianry image containing crack.
10. the surface cracks rapid detection method according to claim 5 based on unmanned plane, which is characterized in that the step
The specific steps of waters primary dcreening operation subprogram in rapid B7 are as follows:
G1, color image is converted to 8 gray level images；
G2, binarization threshold Xotsu is calculated based on maximum variance between clusters；
G3, binary conversion treatment is carried out to 8 gray level images according to threshold X otsu, the image after enabling binary conversion treatment is I1；
G4, image I2 is obtained to image I1 progress erosion operation using waters corrosion structure member SYfs；
G5, image I3 is obtained to image I2 progress closed operation using waters closed operation structural elements SYbi；
G6, the connected domain for calculating image I3, it is 0 by the pixel intensity in connected domain that taking connected domain maximum, which is river, connected domain
Outer pixel intensity is 1, the river region image Ih1 that obtains that treated；
Surface cracks hidden danger pixel judges the specific steps of subprogram in the step B8 are as follows:
H1, to river region image Ih1, the Iclf of bianry image containing crack or the Iclf1 of bianry image containing crack carry out and operation, obtain
To image Iand；
H2, closed operation is carried out to image Iand using crack closed operation structural elements LFbi, obtains binaryzation surface cracks hidden danger figure
As Izhyh；
The specific steps of surface cracks hidden danger pixel counts subprogram in the step B9 are as follows:
The quantity for being 1 using binaryzation crack hidden danger image Izhyh intermediate value is as surface cracks hidden danger pixel quantity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811611646.8A CN109671043B (en) | 2018-12-27 | 2018-12-27 | Unmanned aerial vehicle-based surface crack rapid detection system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811611646.8A CN109671043B (en) | 2018-12-27 | 2018-12-27 | Unmanned aerial vehicle-based surface crack rapid detection system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109671043A true CN109671043A (en) | 2019-04-23 |
| CN109671043B CN109671043B (en) | 2020-07-07 |
Family
ID=66147348
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811611646.8A Active CN109671043B (en) | 2018-12-27 | 2018-12-27 | Unmanned aerial vehicle-based surface crack rapid detection system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109671043B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130216089A1 (en) * | 2010-04-22 | 2013-08-22 | The University Of North Carolina At Charlotte | Method and System for Remotely Inspecting Bridges and Other Structures |
| CN105427377A (en) * | 2015-11-03 | 2016-03-23 | 武汉地大信息工程股份有限公司 | Unmanned plane based geological disaster data collection processing method and device |
| CN106960027A (en) * | 2017-03-20 | 2017-07-18 | 武汉大学 | The UAV Video big data multidate association analysis method of spatial information auxiliary |
| CN107205118A (en) * | 2017-06-27 | 2017-09-26 | 中国地质环境监测院 | Seven camera lens unmanned plane panoramic cameras and its image processing method |
| CN107680114A (en) * | 2017-09-22 | 2018-02-09 | 交通运输部天津水运工程科学研究所 | A kind of meadow cover degree measuring method based on Computer Image Processing |
| CN108447123A (en) * | 2018-03-27 | 2018-08-24 | 贵州电网有限责任公司输电运行检修分公司 | A kind of power transmission line corridor geological disaster investigation method and system |
-
2018
- 2018-12-27 CN CN201811611646.8A patent/CN109671043B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130216089A1 (en) * | 2010-04-22 | 2013-08-22 | The University Of North Carolina At Charlotte | Method and System for Remotely Inspecting Bridges and Other Structures |
| CN105427377A (en) * | 2015-11-03 | 2016-03-23 | 武汉地大信息工程股份有限公司 | Unmanned plane based geological disaster data collection processing method and device |
| CN106960027A (en) * | 2017-03-20 | 2017-07-18 | 武汉大学 | The UAV Video big data multidate association analysis method of spatial information auxiliary |
| CN107205118A (en) * | 2017-06-27 | 2017-09-26 | 中国地质环境监测院 | Seven camera lens unmanned plane panoramic cameras and its image processing method |
| CN107680114A (en) * | 2017-09-22 | 2018-02-09 | 交通运输部天津水运工程科学研究所 | A kind of meadow cover degree measuring method based on Computer Image Processing |
| CN108447123A (en) * | 2018-03-27 | 2018-08-24 | 贵州电网有限责任公司输电运行检修分公司 | A kind of power transmission line corridor geological disaster investigation method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109671043B (en) | 2020-07-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103914708B (en) | Food kind detection method based on machine vision and system | |
| US9811746B2 (en) | Method and system for detecting traffic lights | |
| CN106548165A (en) | A kind of face identification method of the convolutional neural networks weighted based on image block | |
| EP1507232A2 (en) | Method for classifying a digital image | |
| JP2573434B2 (en) | Specific color extraction method | |
| CN107818303A (en) | Unmanned plane oil-gas pipeline image automatic comparative analysis method, system and software memory | |
| CN111369529B (en) | Article loss and leave-behind detection method and system | |
| CN109284759A (en) | One kind being based on the magic square color identification method of support vector machines (svm) | |
| CN111783803A (en) | Image processing method and device for realizing privacy protection | |
| CN110766636B (en) | Intelligent image processing method for AI chip | |
| CN109685024A (en) | A kind of vacant lot one artificial intelligence geological disaster hidden danger primary dcreening operation system and method | |
| CN106558044A (en) | The resolution measuring method of image module | |
| CN110507283A (en) | Retina camera and its implementation | |
| CN109671043A (en) | A kind of surface cracks rapid detection system and method based on unmanned plane | |
| US20230005103A1 (en) | Image processor | |
| CN112101260A (en) | Method, device, equipment and storage medium for identifying safety belt of operator | |
| CN116757988A (en) | Infrared and visible light image fusion method based on semantic enrichment and segmentation tasks | |
| US8437545B1 (en) | System and method for digital image signal compression using intrinsic images | |
| CN106408617A (en) | Interactive single image material acquiring system based on YUV color space and method | |
| KR20030091471A (en) | YCrCb color based human face location detection method | |
| KR101521269B1 (en) | Method for detecting snow or rain on video | |
| CN115063605B (en) | Method for identifying color printing package by using electronic equipment | |
| US20150350647A1 (en) | Selective perceptual masking via downsampling in the spatial and temporal domains using intrinsic images for use in data compression | |
| US20240087128A1 (en) | Adaptive auto white balancing | |
| CN115330787B (en) | Photovoltaic module icing area detection method based on image recognition |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |