CN102608016A - Method for measuring average size of complicated particles based on Canny boundary detection - Google Patents

Method for measuring average size of complicated particles based on Canny boundary detection Download PDF

Info

Publication number
CN102608016A
CN102608016A CN2012101079257A CN201210107925A CN102608016A CN 102608016 A CN102608016 A CN 102608016A CN 2012101079257 A CN2012101079257 A CN 2012101079257A CN 201210107925 A CN201210107925 A CN 201210107925A CN 102608016 A CN102608016 A CN 102608016A
Authority
CN
China
Prior art keywords
image
size
particle
particles
average
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
Application number
CN2012101079257A
Other languages
Chinese (zh)
Inventor
王卫星
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fuzhou University
Original Assignee
Fuzhou University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Fuzhou University filed Critical Fuzhou University
Priority to CN2012101079257A priority Critical patent/CN102608016A/en
Publication of CN102608016A publication Critical patent/CN102608016A/en
Pending legal-status Critical Current

Links

Images

Abstract

The invention relates to the technical field of image processing, in particular to a method for measuring an average size of complicated particles based on Canny boundary detection. The method comprises the following steps of 1, inputting an image of moving particles; 2, performing Gaussian smoothing noise removal; 3, when the image size is greater than an image size threshold value, shrinking the image; 4, performing Canny boundary scanning, and converting the image into a binary image; 5, calculating the particle edge density of the image, and determining an average shape parameter of the particles and a background space parameter of the image; and 6, calculating the average size of the detected particles and the number of detected particles. through adoption of the method, the average size and the number of the particles in the image can be detected quickly and accurately; and the method is very suitable to statistics and measurement of a plurality of complicated particles at real time on line and can be used for pre-partitioning acomplicated particle image.

Description

Complex granule average-size measuring method based on Canny border detections
Technical field
The present invention relates to technical field of image processing, particularly a kind of complex granule average-size measuring method based on Canny border detections, it is adaptable to the segmentation of the video particle image of the dense distribution of motion.
Background technology
One NI Vision Builder for Automated Inspection based on optics and computer technology is often a part for producing control line, and it can improve speed of production and quality, uniform rules and standard.In recent years, carry out industrial detection using computer vision and be applied to many different fields, for example, integrated circuit, steel production, processing of poultry, road construction, catalase etc..
In stone industries, it is very important that the quality of building stones, which is estimated,.Building stones are exactly the mixture of the sillar of nature sillar and explosion and Mechanical Crushing.In order to judge the quality of building stones, the size and dimension parameter progress estimation to building stones particle is necessary.The average-size of building stones is not used for assessing a data of product quality still, but also is the important information for adjusting disintegrating machine, for example：Adjust its aperture etc..Disintegrating machine is generally set to produce the building stones in some relative narrower size range for strictly specifying, such as from 16mm to 30mm.One leading indicator of usual disintegrating machine operation is exactly average-size.In automatic pulverizing control system, include the feedback signal of average building stones size from what real-time system was beamed back, just show the actual development of shattering process on streamline.In actual applications, the crushed particles come out from disintegrating machine are transmitted on a conveyer belt, and a CCD camera is placed above it and is shot downwards, and then the particle in the image of acquisition is measured with image procossing, segmentation and analysis.
In mining industry and mineral processing production, average-size, also referred to as k50 values are to ensure there is the screening size that half sample can pass through.If k50 values are too low, the cost of rock blasting will be improved；On the contrary, if value is too high, the expense of charge of trucks, transport and secondary blasting can all increase.Therefore, the average-size of rock blasting is to make a key factor of being optimal of Mining Market, is the important information of the whole mining production process of control.
The content of the invention
It is an object of the invention to provide a kind of complex granule average-size measuring method based on Canny border detections, this method is conducive to rapidly and accurately detecting the average-size and quantity of particle in image.
The technical solution adopted by the present invention is：A kind of complex granule average-size measuring method based on Canny border detections, is carried out as follows：
Step 1：Input the moving particle image on conveyer belt or in drop fluidf(x,y)；
Step 2：Smothing filtering is carried out, the noise in the particle image is removed, the process is represented by：                                               
Figure 2012101079257100002DEST_PATH_IMAGE002
, whereinhRepresent smoothing filter；
Step 3：A picture size threshold value is set, when picture size is more than picture size threshold value, image down is carried out in proportion； 
Step 4：Canny boundary scans are carried out to the image after processing, and gradient image is converted into bianry image；
Step 5：Grain edges density calculating is carried out to bianry image, and determines particle average shape parameter and the spatial context parameter of image：
Particle average shape parameter：
Figure 2012101079257100002DEST_PATH_IMAGE004
The spatial context parameter of image：
Figure 2012101079257100002DEST_PATH_IMAGE006
                      
Wherein,
Figure 2012101079257100002DEST_PATH_IMAGE008
Represent marginal density,
Figure 2012101079257100002DEST_PATH_IMAGE010
Actually detected grain edges density is represented,
Figure 952130DEST_PATH_IMAGE010
Equal to the number of edge pixel in image
Figure 2012101079257100002DEST_PATH_IMAGE012
Divided by the total pixel number of image
Figure 2012101079257100002DEST_PATH_IMAGE014
, the factor
Figure 2012101079257100002DEST_PATH_IMAGE016
Determined by intergranular spatial context；
Step 6：Detected particle mean size is estimated using boundary density, and further estimates numbers of particles, particle mean size and numbers of particles are calculated as follows：
Particle mean size：
Figure 2012101079257100002DEST_PATH_IMAGE018
Numbers of particles：。
The beneficial effects of the invention are as follows realizing a kind of without image threshold processing to be quickly based on Canny border detection algorithms, particle mean size is then calculated based on grain boundary density in image again.To a certain extent, this algorithm need not carry out fine image segmentation and can rapidly and accurately detect the quantity and average-size of particle in image, detection speed is fast, high precision, it is highly suitable for complicated many particle statistic measurements of real-time online, simultaneously can be used for the pre-segmentation of complex granule image.
Brief description of the drawings
Fig. 1 is the fundamental diagram of the embodiment of the present invention.
Embodiment
Complex granule average-size measuring method of the invention based on Canny border detections, is carried out as follows：
Step 1：The particle image of input motionf(x,y), the particle image or other dynamic particle images on conveyer belt are may generally be, but the background area in particle image is unsuitable excessive, most preferably less than 20%；
Step 2：Smothing filtering is carried out, the noise in the particle image is removed, the process is represented by：, whereinhSmoothing filter is represented, smoothing filter uses Gaussian filter：
Figure 2012101079257100002DEST_PATH_IMAGE022
Step 3：A picture size threshold value is set, when picture size is more than picture size threshold value, image down is carried out in proportion：When image is more than 1024x1024 pixels, one times of downsizing smoothly image, if image is more than 2000x2000 pixels, two times of downsizing smoothly image；
Step 4：Canny boundary scans are carried out to the image after processing, and gradient image is converted into bianry image：
Figure 2012101079257100002DEST_PATH_IMAGE024
It isgGradient vector,
Figure 2012101079257100002DEST_PATH_IMAGE026
,
Figure 2012101079257100002DEST_PATH_IMAGE028
It is smoothed imagegGradient image；
Step 5：Grain edges density calculating is carried out to bianry image, and determines particle average shape parameter and the spatial context parameter of image：
Particle average shape parameter：
Figure 462056DEST_PATH_IMAGE004
The spatial context parameter of image：                      
Wherein,
Figure 671638DEST_PATH_IMAGE008
Represent marginal density,
Figure 431784DEST_PATH_IMAGE010
Actually detected grain edges density is represented,
Figure 698817DEST_PATH_IMAGE010
Equal to the number of edge pixel in image
Figure 897717DEST_PATH_IMAGE012
Divided by the total pixel number of image
Figure 504279DEST_PATH_IMAGE014
, the factor
Figure 497643DEST_PATH_IMAGE016
Determined by intergranular spatial context；
Step 6：Detected particle mean size is estimated using boundary density, and further estimates numbers of particles, particle mean size and numbers of particles are calculated as follows：
Particle mean size：
Figure 251972DEST_PATH_IMAGE018
Numbers of particles：
Figure 988984DEST_PATH_IMAGE020
。
 
Technical scheme is further described with reference to specific embodiment.
It is assumed that particle is compact and approximate ellipsoidal, the average-size of particle is relevant with the difference size of marginal density, average shape factor and size.If the difference of size is not very big, marginal density and average shape factor can just measure particle mean size in image well, and precision is within 5-10%.It is a kind of less method of relative amount of calculation for obtaining average-size to measure average-size based on marginal density.
If the theoretical particles marginal density in particle image is, and
Figure 450052DEST_PATH_IMAGE010
For actually detected grain edges density.Calculate
Figure 879897DEST_PATH_IMAGE010
Method be exactly use image in edge pixel numbern e Divided by the number of total picture element of imagen tot ：
Figure 2012101079257100002DEST_PATH_IMAGE032
                                                                              (1)
If imageGradient image be
Figure 2012101079257100002DEST_PATH_IMAGE036
If defining * and representing conversion, smooth image can be write as：
Figure 59205DEST_PATH_IMAGE002
                                                                            (2)
WhereinhIt is some smoothing filter.We will use Gaussian filter：
Figure 2012101079257100002DEST_PATH_IMAGE038
                                                 (3)
Figure 2012101079257100002DEST_PATH_IMAGE040
It is
Figure 2012101079257100002DEST_PATH_IMAGE042
Gradient vector,
Figure 2012101079257100002DEST_PATH_IMAGE044
。
Figure 2012101079257100002DEST_PATH_IMAGE046
It is smoothed image
Figure 2012101079257100002DEST_PATH_IMAGE048
Gradient image.Smoothing parameter
Figure 2012101079257100002DEST_PATH_IMAGE050
I.e. so-called filter size parameter.Canny images are defined as the maximum on gradient direction.So in discrete grid, the pixel for reaching this maximum is exactly edge pixel.In bianry image, 0 means that edge pixel, and non-zero then to represent non-edge pixel, it is exactly an edge image, is denoted as
Figure 2012101079257100002DEST_PATH_IMAGE052
, or be more accurately expressed as
Figure 2012101079257100002DEST_PATH_IMAGE054
.Marginal density
Figure 2012101079257100002DEST_PATH_IMAGE056
Always according to this
Figure 2012101079257100002DEST_PATH_IMAGE058
Edge image
Figure 2012101079257100002DEST_PATH_IMAGE060
What value was calculated.
Consider the image of the compact particle of sub-elliptical, this approximation is not configured to describe single grain shape, but in order to set up a model from marginal density to average-size.The concept definition of size is as follows：
Ellipse is designated, make secondary axes and the main axis length be respectivelyWith
Figure 2012101079257100002DEST_PATH_IMAGE066
, and,
Figure 2012101079257100002DEST_PATH_IMAGE070
.Below, Wo Menyong
Figure 2012101079257100002DEST_PATH_IMAGE072
Measurement size is represented, is called length.With
Figure 2012101079257100002DEST_PATH_IMAGE074
With
Figure 2012101079257100002DEST_PATH_IMAGE076
Area and girth are represented respectively.Assuming that there is no border in oval inside.Define marginal density
Figure 2012101079257100002DEST_PATH_IMAGE078
It is as follows：
Figure 2012101079257100002DEST_PATH_IMAGE080
                                                             （4）
With
Figure 2012101079257100002DEST_PATH_IMAGE082
Relation it is as follows：
Figure 2012101079257100002DEST_PATH_IMAGE084
                            （5）
Wherein E is complete elliptic integral.
Average value principle of last equation from integration：
Figure 2012101079257100002DEST_PATH_IMAGE086
If replacing discrete function to can be used for summation with continuous function.Order
Figure 2012101079257100002DEST_PATH_IMAGE088
, have again：
           （6）
Figure 2012101079257100002DEST_PATH_IMAGE092
                                            （7）
Wherein
Figure 2012101079257100002DEST_PATH_IMAGE094
It is to meet the 1 of equation to a value between n.
Figure 2012101079257100002DEST_PATH_IMAGE096
It is to use
Figure 2012101079257100002DEST_PATH_IMAGE098
The component of definitionA useful average value,It is another useful average value, but by component
Figure 2012101079257100002DEST_PATH_IMAGE104
Definition.Notice generally
Figure 2012101079257100002DEST_PATH_IMAGE106
, have：
Figure 2012101079257100002DEST_PATH_IMAGE108
                                                 （8）
Wherein
Figure 2012101079257100002DEST_PATH_IMAGE110
It is average length（
Figure 2012101079257100002DEST_PATH_IMAGE112
）,It is L variance, is defined as
Figure 2012101079257100002DEST_PATH_IMAGE116
.WeIt is called form factor,
,
Figure 2012101079257100002DEST_PATH_IMAGE122
                                        （9）
It is called " average shape factor ".When all elliptical shapes are identical i.e.
Figure 2012101079257100002DEST_PATH_IMAGE124
, then for arbitrary, it is clear that have
Figure 2012101079257100002DEST_PATH_IMAGE128
.The visible shape factor and tight ness ratingIt is closely related.Approximation
Figure 2012101079257100002DEST_PATH_IMAGE132
It is very famous from Spiegel (page 1992,7).
There is known average shape factor
Figure 2012101079257100002DEST_PATH_IMAGE134
And average-size, use formula（8）With（9）And formula（5）It can export：
Figure 2012101079257100002DEST_PATH_IMAGE138
                                                          （10）
We are now with average length
Figure 2012101079257100002DEST_PATH_IMAGE140
With a kind of marginal density
Figure 2012101079257100002DEST_PATH_IMAGE142
Relation.The marginal density measured in an experiment
Figure 2012101079257100002DEST_PATH_IMAGE144
With
Figure 2012101079257100002DEST_PATH_IMAGE146
It is relevant：
Figure 2012101079257100002DEST_PATH_IMAGE148
, the wherein factorDetermined by intergranular spatial context.
A value is introduced again now
Figure 2012101079257100002DEST_PATH_IMAGE152
, this is a kind of standard variance.So, by
Figure 2012101079257100002DEST_PATH_IMAGE154
It can export：
Figure 2012101079257100002DEST_PATH_IMAGE156
                                                                            (11)
Certainly, we are it could not be expected that calculate accurate average valueWith
Figure 2012101079257100002DEST_PATH_IMAGE160
.One approximation can be calculated from coarse partition data to be obtained, and by coarse segmentation, the quantity of the non-dark candidate region with uniform characteristics exists
Figure 269082DEST_PATH_IMAGE048
Place is generated.Therefore, region
Figure 2012101079257100002DEST_PATH_IMAGE162
And parameter
Figure 2012101079257100002DEST_PATH_IMAGE164
It is available.Consider：
                                                               
Figure 2012101079257100002DEST_PATH_IMAGE168
Figure 2012101079257100002DEST_PATH_IMAGE170
                                                                 (14)
When solving, similar ellipse is obtained by formula (12) and (13), and is generated including tight ness rating
Figure 2012101079257100002DEST_PATH_IMAGE174
Conditional (14).They can serve as given area
Figure 2012101079257100002DEST_PATH_IMAGE176
With the girth in divided region
Figure 2012101079257100002DEST_PATH_IMAGE178
, calculated in segmentation figure pictureEquation.We can also be by
Figure 2012101079257100002DEST_PATH_IMAGE182
With
Figure 743180DEST_PATH_IMAGE162
Try to achieve
Figure 78347DEST_PATH_IMAGE082
.By
Figure 807268DEST_PATH_IMAGE082
Try to achieveAverage value
Figure 2012101079257100002DEST_PATH_IMAGE186
It can also calculate.We use
Figure 2012101079257100002DEST_PATH_IMAGE188
Instead of formula（9）In
Figure 2012101079257100002DEST_PATH_IMAGE190
, so that it is determined that the average shape factor used in experiment
Figure 2012101079257100002DEST_PATH_IMAGE192
：
Figure DEST_PATH_IMAGE194
                                                                            （15）
In an image sequence, form factorChange over time is possible to influence temporal average certainly.
1. the spatial context detection of compact particle
One feature of compact particle is exactly that intergranular gap is smaller, and we are referred to as spatial context.Spatial context either particle surface part, or particle the second layer.Although generally attempted in commercial Application control lighting condition, in most cases this be it is difficult, especially quarrying production in it is more difficult.Some marginal illumination condition must reach, but we are difficult that excessive regulation is done to light source in order to produce the effect of determination.
We are interested in spatial context to be based on two reasons：First, in order to calculate marginal density, it would be desirable to detect compact intergranular spatial context；2nd, average shape parameter is necessary, and at least certain coarse segmentation is a kind of possible source, and the edge between spatial context and granule boundary is easier this segmentation.
The quantity of spatial context is very important for size detection between particle, and such as formula (11), wherein β is the ratio in total region and non-spatial context.Compact degree depends on the shape of particle, natural building stones（Gravel from natural resources）Shape is mainly ellipse.The difference on stone shape after broken becomes big, and some particles are probably asymmetric, as circle, triangle, circle rectangle or trapezium etc..And we have simply assumed that particle is that oval, oval particle may is that the rational model for setting a good β value before.Although these modes of oval aggregates together have infinite a variety of, β approximation can be obtained with following method.Tangent ellipse is formed with 3,4 or 5 summits（Angle）Spatial context, be whether this 3,4 or 5 ellipses define a spatial context.We are referred to as 3 jiaos, 4 jiaos or 5 jiaos spatial contexts.In our image, 4 jiaos and 3 jiaos of spatial contexts are that comparison is more.When all ellipses are all equally big, 4 jiaos of spatial contexts and easily calculate.The area of rectangle is
Figure DEST_PATH_IMAGE198
, ellipse area
Figure DEST_PATH_IMAGE200
.Therefore, the spatial context summation in this rectangle is the rectangular area
Figure DEST_PATH_IMAGE202
.The spatial context rate of the image of these rectangles.Even if we use rock group oval in the same direction, and every group of size is essentially identical, and the spatial context between these rocks is slightly changed, and above formula is still set up.What the result of 3 jiaos of spatial contexts was also similar to, therefore, according to some empirical datas to spatial context, we generally take
Figure DEST_PATH_IMAGE206
。
2. Grads threshold
Canny edge detectors are in gradient direction
Figure DEST_PATH_IMAGE208
On to compare three local
Figure DEST_PATH_IMAGE210
.If for example, edge edge
Figure DEST_PATH_IMAGE212
Direction, then to determine a pixel
Figure DEST_PATH_IMAGE214
It is edge pixel, it is necessary to meet
Figure DEST_PATH_IMAGE216
And
Figure DEST_PATH_IMAGE218
.Generally, there is a threshold value, use here
Figure DEST_PATH_IMAGE220
Represent, by judging
Figure DEST_PATH_IMAGE222
To exclude the prospective edge pixel of those low contrasts.This Grads threshold may adapt to each image, generallyIt is the maximum of an image
Figure DEST_PATH_IMAGE226
5% to the 10% of value.Accurately
Figure 512182DEST_PATH_IMAGE220
Value is decided by
Figure DEST_PATH_IMAGE228
Whether that position on histogram has a peak.When we calculate marginal density, adapting to image Grads threshold
Figure 10159DEST_PATH_IMAGE220
All it is different in each frame of an image sequence.This threshold value directly affects density measure
Figure DEST_PATH_IMAGE230
.Work as basisDetermine
Figure 420860DEST_PATH_IMAGE220
When,Change it is relevant with internal stability.Therefore,
Figure 463039DEST_PATH_IMAGE220
With the time it is smooth in an image sequence it is seemingly rational.Order
Figure DEST_PATH_IMAGE232
It is
Figure 343270DEST_PATH_IMAGE220
The function changed with discrete time t.Picture frame can mark for
Figure DEST_PATH_IMAGE234
, a kind of possible rule is to use
Figure DEST_PATH_IMAGE236
Substitute what is automatically generated
Figure 718888DEST_PATH_IMAGE232
：
Figure DEST_PATH_IMAGE238
                                                               (16)
Especially, if picture frame overlapping, such as be loaded with particle（Object）Conveyer belt mobile image size a quarter, then stablize
Figure DEST_PATH_IMAGE240
Value be very important.Emphasize marginal density and selected threshold value
Figure 77188DEST_PATH_IMAGE240
Or the relation of threshold rule, have：
Figure DEST_PATH_IMAGE242
                                                                  （17）
3. ignore change in size
Figure DEST_PATH_IMAGE244
When
Figure DEST_PATH_IMAGE246
It is the 1/25 of gradient, as long as 1.5% to 3.4% error is allowed to, the relation of following simplification can be used.
Figure DEST_PATH_IMAGE248
                                                                            （18）
Applied to broken building stones, this is generally satisfactory.Because by broken（Or screening）Size after operation usually exists
Figure DEST_PATH_IMAGE250
Between.If
Figure DEST_PATH_IMAGE252
, then have successivelyWith
Figure DEST_PATH_IMAGE256
.For example, right
Figure 425124DEST_PATH_IMAGE250
An internal Unify legislation is
Figure DEST_PATH_IMAGE258
.Gaussian Profile is shown.Therefore generally take, and class Gaussian Profile takes
Figure DEST_PATH_IMAGE264
, and ignore generation
Figure DEST_PATH_IMAGE266
Error is about 3.5% to 1.5%。
4. the removal of noise margin
If amplifying a particle, the noise of particle surface is likely to also produce edge.In order to calculate particle mean size, simple geo-statistic edge pixel will be malfunctioned with obtaining marginal density.What marginal density now reflected is not the quantity of particle but the quantity of image style.Therefore, it is necessary for removing the installations of particle surface noise during average-size is detected.We adopt in the following method to alleviate this problem.
Order
Figure DEST_PATH_IMAGE270
For artwork, then the image of lower resolution
Figure DEST_PATH_IMAGE272
Calculated by following formula：
Figure DEST_PATH_IMAGE274
                                                                        (19)
Wherein H represents that operator isConvolution.
Obtained by halving twice
Figure DEST_PATH_IMAGE278
The image for the expression Gaussian smoothing being exactly previously mentioned, so as to obtain measuring marginal density
Figure 413940DEST_PATH_IMAGE144
Required
Figure 339171DEST_PATH_IMAGE048
.Therefore, from now on,
Figure 501162DEST_PATH_IMAGE278
By formula（19）Obtain, with formula（3）Compare available
Figure DEST_PATH_IMAGE280
。
Technique according to the invention scheme, what embodiment was provided quickly calculates marginal density based on Canny border detections, and then calculates the overall numbers of particles of the particle mean size of dense distribution in image, specifically follows these steps to carry out：
1）The particle image of input motion
Figure DEST_PATH_IMAGE282
, the particle image or other dynamic particle images on conveyer belt are may generally be, but the background area in particle image is unsuitable excessive, most preferably less than 20%；
2）Then Gaussian smoothing filter is carried out, the noise in image is removed：Image can be write as：, wherein
Figure DEST_PATH_IMAGE286
It is some smoothing filter.We will use Gaussian filter：
Figure 500342DEST_PATH_IMAGE022
.A kind of approach of design Gaussian filter is the calculation template weights directly from discrete Gaussian Profile.It is integer it is generally desirable to filter weights for convenience of calculation.A value is taken in a corner point of template, and one K of selection makes the corner point value be 1.Wave filter integer can be made by this coefficient, because the template weights sum after integer is not equal to 1, in order to which the inhomogeneous intensity region for ensureing image is unaffected, it is necessary to which weights standardization is carried out to Filtering Template.Specific template is as follows：
3x3 templates：
Figure DEST_PATH_IMAGE288
, 5x5 templates：
Figure DEST_PATH_IMAGE290
；
3）Determine whether image reduces according to the size of picture size：When image is more than 1024x1024 pixels, one times of downsizing smoothly image, if image is more than 2000x2000 pixels, two times of downsizing smoothly image；
4）To Gaussian smoothing（And diminution）Image afterwards carries out Canny boundary scans, and gradient image is converted into bianry image：
Figure DEST_PATH_IMAGE292
It is
Figure DEST_PATH_IMAGE294
Gradient vector,
Figure DEST_PATH_IMAGE296
。
Figure 722376DEST_PATH_IMAGE046
It is smoothed image
Figure 134902DEST_PATH_IMAGE048
Gradient image.Canny operator specific steps：Step1:Use Gaussian filter smoothing image；Step2:Amplitude and the direction of gradient are calculated with the finite difference of single order local derviation；Step3:Non-maxima suppression is carried out to gradient magnitude；Step4:Detected with dual threashold value-based algorithm and connect edge (refer to relevant textbook).
5）Grain edges density calculating is carried out to bianry image, and determines particle average shape parameter and the spatial context parameter of image：
Figure DEST_PATH_IMAGE298
。
The quantity of spatial context is very important for size detection between particle, and wherein β is the ratio in total region and non-spatial context.Compact degree depends on the shape of particle, natural building stones（Gravel from natural resources）Shape is mainly ellipse.The difference on stone shape after broken becomes big, and some particles are probably asymmetric, as circular trigonometric shape, circle rectangle or trapezium etc..And we have simply assumed that particle is that oval, oval particle may is that the rational model for setting a good β value before.Although these modes of oval aggregates together have infinite a variety of, β approximation can be obtained with following method.Tangent ellipse is formed with 3,4 or 5 summits（Angle）Spatial context, be whether this 3,4 or 5 ellipses define a spatial context.We are referred to as 3 jiaos, 4 jiaos or 5 jiaos spatial contexts.In our image, 4 jiaos and 3 jiaos of spatial contexts are that comparison is more.When all ellipses are all equally big, 4 jiaos of spatial contexts and easily calculate.The area of rectangle is
Figure DEST_PATH_IMAGE300
, ellipse area
Figure DEST_PATH_IMAGE302
.Therefore, the spatial context summation in this rectangle is the rectangular area
Figure DEST_PATH_IMAGE304
.There is the spatial context rate of the image of these rectangles
Figure 100584DEST_PATH_IMAGE204
.Even if we use rock group oval in the same direction, and every group of size is essentially identical, and the spatial context between these rocks is slightly changed, and above formula is still set up.What the result of 3 jiaos of spatial contexts was also similar to, therefore, according to some empirical datas to spatial context, we generally take
Figure 219850DEST_PATH_IMAGE206
；
6）Finally calculate detected particle mean size：
Figure DEST_PATH_IMAGE306
, numbers of particles：。
Above is presently preferred embodiments of the present invention, all changes made according to technical solution of the present invention, produced function without departing from technical solution of the present invention scope when, belong to protection scope of the present invention. 

Claims (2)

1. a kind of complex granule average-size measuring method based on Canny border detections, it is characterised in that：Carry out as follows：
Step 1：Input the moving particle image on conveyer belt or in drop fluidf(x,y)；
Step 2：Smothing filtering is carried out, the noise in the particle image is removed, the process is represented by：                                               , whereinhRepresent smoothing filter；
Step 3：A picture size threshold value is set, when picture size is more than picture size threshold value, image down is carried out in proportion； 
Step 4：Canny boundary scans are carried out to the image after processing, and gradient image is converted into bianry image；
Step 5：Grain edges density calculating is carried out to bianry image, and determines particle average shape parameter and the spatial context parameter of image：
Particle average shape parameter：
Figure 2012101079257100001DEST_PATH_IMAGE004
The spatial context parameter of image：
Figure 2012101079257100001DEST_PATH_IMAGE006
Wherein,r m The breadth length ratio of average grain is represented,
Figure 2012101079257100001DEST_PATH_IMAGE008
Represent marginal density,
Figure 2012101079257100001DEST_PATH_IMAGE010
Actually detected grain edges density is represented,
Figure 957562DEST_PATH_IMAGE010
Equal to the number of edge pixel in image
Figure 2012101079257100001DEST_PATH_IMAGE012
Divided by the total pixel number of image
Figure 2012101079257100001DEST_PATH_IMAGE014
, the factor
Figure 2012101079257100001DEST_PATH_IMAGE016
Determined by intergranular spatial context；
Step 6：Detected particle mean size is estimated using boundary density, and further estimates numbers of particles, particle mean size and numbers of particles are calculated as follows：
Particle mean size：
Figure 2012101079257100001DEST_PATH_IMAGE018
Numbers of particles：
Wherein,xsizeThe size in x-axis direction is represented,ysizeRepresent the size in y-axis direction.
2. the complex granule average-size measuring method according to claim 1 based on Canny border detections, it is characterised in that：In step 2, smoothing filter uses Gaussian filter：
Figure 2012101079257100001DEST_PATH_IMAGE022
。
CN2012101079257A 2012-04-13 2012-04-13 Method for measuring average size of complicated particles based on Canny boundary detection Pending CN102608016A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2012101079257A CN102608016A (en) 2012-04-13 2012-04-13 Method for measuring average size of complicated particles based on Canny boundary detection

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2012101079257A CN102608016A (en) 2012-04-13 2012-04-13 Method for measuring average size of complicated particles based on Canny boundary detection

Publications (1)

Publication Number Publication Date
CN102608016A true CN102608016A (en) 2012-07-25

Family

ID=46525575

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2012101079257A Pending CN102608016A (en) 2012-04-13 2012-04-13 Method for measuring average size of complicated particles based on Canny boundary detection

Country Status (1)

Country Link
CN (1) CN102608016A (en)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2014102428A1 (en) * 2012-12-26 2014-07-03 Universidad De Almeria Method for the automatic interpretation of images for the quantification of nucelar tumour markers
CN104182759A (en) * 2014-08-20 2014-12-03 徐州坤泰电子科技有限公司 Scanning electron microscope based particle morphology identification method
CN106157271A (en) * 2016-03-01 2016-11-23 闽南师范大学 A kind of method that carbon-bearing particulate matter nanostructured based on image processing techniques is analyzed
CN106596357A (en) * 2016-11-28 2017-04-26 江苏大学 Method for characterizing morphologies of particulate matters in diesel
CN106682678A (en) * 2016-06-24 2017-05-17 西安电子科技大学 Image angle point detection and classification method based on support domain
CN107024416A (en) * 2017-04-07 2017-08-08 福建福光股份有限公司 With reference to similitude and the quasi-circular particle mean size detection method of discontinuity
CN107314957A (en) * 2017-06-30 2017-11-03 长安大学 A kind of measuring method of rock fragmentation Size Distribution
CN109716355A (en) * 2016-09-07 2019-05-03 马尔文帕纳科公司 Grain boundaries identification
CN110455825A (en) * 2019-08-27 2019-11-15 无锡领先针测电子有限公司 The method that particle on a kind of pair of glass is counted
CN112985260A (en) * 2021-02-04 2021-06-18 捷普电子(新加坡)公司 Particle detection and removal method and device
CN116228747A (en) * 2023-05-04 2023-06-06 青岛穗禾信达金属制品有限公司 Metal cabinet processing quality monitoring method

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101620060A (en) * 2009-08-13 2010-01-06 上海交通大学 Automatic detection method of particle size distribution

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101620060A (en) * 2009-08-13 2010-01-06 上海交通大学 Automatic detection method of particle size distribution

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
官理等: "基于边缘密度自动检测紧凑颗粒的平均尺寸", 《计算机工程与应用》 *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2014102428A1 (en) * 2012-12-26 2014-07-03 Universidad De Almeria Method for the automatic interpretation of images for the quantification of nucelar tumour markers
CN104182759B (en) * 2014-08-20 2017-09-12 中国矿业大学 Particulate matter form recognition methods based on ESEM
CN104182759A (en) * 2014-08-20 2014-12-03 徐州坤泰电子科技有限公司 Scanning electron microscope based particle morphology identification method
CN106157271A (en) * 2016-03-01 2016-11-23 闽南师范大学 A kind of method that carbon-bearing particulate matter nanostructured based on image processing techniques is analyzed
CN106682678B (en) * 2016-06-24 2020-05-01 西安电子科技大学 Image corner detection and classification method based on support domain
CN106682678A (en) * 2016-06-24 2017-05-17 西安电子科技大学 Image angle point detection and classification method based on support domain
CN109716355A (en) * 2016-09-07 2019-05-03 马尔文帕纳科公司 Grain boundaries identification
CN109716355B (en) * 2016-09-07 2023-12-26 马尔文帕纳科公司 Particle boundary identification
CN106596357A (en) * 2016-11-28 2017-04-26 江苏大学 Method for characterizing morphologies of particulate matters in diesel
CN107024416A (en) * 2017-04-07 2017-08-08 福建福光股份有限公司 With reference to similitude and the quasi-circular particle mean size detection method of discontinuity
CN107314957A (en) * 2017-06-30 2017-11-03 长安大学 A kind of measuring method of rock fragmentation Size Distribution
CN107314957B (en) * 2017-06-30 2020-07-07 长安大学 Method for measuring rock block size distribution
CN110455825A (en) * 2019-08-27 2019-11-15 无锡领先针测电子有限公司 The method that particle on a kind of pair of glass is counted
CN112985260A (en) * 2021-02-04 2021-06-18 捷普电子(新加坡)公司 Particle detection and removal method and device
CN112985260B (en) * 2021-02-04 2022-11-22 捷普电子(新加坡)公司 Particle detection and removal method and device
CN116228747A (en) * 2023-05-04 2023-06-06 青岛穗禾信达金属制品有限公司 Metal cabinet processing quality monitoring method

Similar Documents

Publication Publication Date Title
CN102608016A (en) Method for measuring average size of complicated particles based on Canny boundary detection
CN110286124B (en) Machine vision-based refractory brick measuring system
US10115035B2 (en) Vision system and analytical method for planar surface segmentation
Thurley Automated online measurement of limestone particle size distributions using 3D range data
CN108711149B (en) Mineral rock granularity detection method based on image processing
CN107314957B (en) Method for measuring rock block size distribution
Zhang et al. Multi-scale image segmentation of coal piles on a belt based on the Hessian matrix
Thurley Automated image segmentation and analysis of rock piles in an open-pit mine
Kalantarian et al. Accuracy of surface tension measurement from drop shapes: The role of image analysis
CN114324078A (en) Particle size identification method, device, equipment and medium
CN107784646A (en) A kind of road self-adapting detecting method to gather materials
CN107220946B (en) Real-time removing method for bad block degree images on rock conveyer belt
CN105469384B (en) The integrated evaluating method of license plate image quality
Thurley et al. Identifying, visualizing, and comparing regions in irregularly spaced 3D surface data
CN108629776B (en) Mineral rock granularity detecting system
Othman et al. Road crack detection using adaptive multi resolution thresholding techniques
CN107024416A (en) With reference to similitude and the quasi-circular particle mean size detection method of discontinuity
Yu et al. Optimized self-adapting contrast enhancement algorithm for wafer contour extraction
Anusree et al. Characterization of sand particle morphology: state-of-the-art
Andersson et al. Minimizing profile error when estimating the sieve-size distribution of iron ore pellets using ordinal logistic regression
Lu et al. Evaluation of coal screening performance by spatial image reconstruction method
Lu et al. Fusion-based color and depth image segmentation method for rocks on conveyor belt
CN110298799B (en) PCB image positioning correction method
CN111191534B (en) Road extraction method in fuzzy aviation image
Lee et al. Development of a crack type index

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C12 Rejection of a patent application after its publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20120725