CN108376403A - Grid Colony hybridization dividing method based on hough-circle transform - Google Patents
Grid Colony hybridization dividing method based on hough-circle transform Download PDFInfo
- Publication number
- CN108376403A CN108376403A CN201810086989.0A CN201810086989A CN108376403A CN 108376403 A CN108376403 A CN 108376403A CN 201810086989 A CN201810086989 A CN 201810086989A CN 108376403 A CN108376403 A CN 108376403A
- Authority
- CN
- China
- Prior art keywords
- bacterium colony
- grid
- image
- target
- gray level
- 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
- 238000009396 hybridization Methods 0.000 title claims abstract description 53
- 238000000034 method Methods 0.000 title claims abstract description 24
- 241000894006 Bacteria Species 0.000 claims abstract description 256
- 230000011218 segmentation Effects 0.000 claims abstract description 17
- 238000001514 detection method Methods 0.000 claims abstract description 13
- 239000003550 marker Substances 0.000 claims abstract description 12
- 101100269850 Caenorhabditis elegans mask-1 gene Proteins 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 6
- 230000007717 exclusion Effects 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 2
- 239000000284 extract Substances 0.000 claims description 2
- 238000000638 solvent extraction Methods 0.000 claims description 2
- 238000002474 experimental method Methods 0.000 description 8
- 230000000875 corresponding effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000002609 medium Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 239000001963 growth medium Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 238000012372 quality testing Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G06T5/70—
-
- 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
- G06T7/168—Segmentation; Edge detection involving transform domain methods
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20061—Hough transform
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20152—Watershed segmentation
Abstract
The present invention proposes a kind of grid Colony hybridization dividing method based on hough-circle transform, for solve it is existing in the prior art can not achieve the technical issues of being split to grid Colony hybridization, realize that step is:Grid Colony hybridization is pre-processed;Obtain the binaryzation boundary image of bacterium colony target in the grid bacterium colony gray level image after denoising;The mean radius r and average gray value g of bacterium colony target in the grid bacterium colony gray level image after denoising are obtained using hough-circle transform;The candidate bacterium colony target marker image in the grid bacterium colony gray level image after denoising is obtained using hough-circle transform;Obtain the segmentation result of input grid Colony hybridization.The accurate Ground Split grid Colony hybridization of energy of the invention, can be used for detection, counting to bacterium colony target in grid Colony hybridization and classification.
Description
Technical field
The invention belongs to technical field of image processing, are related to a kind of dividing method of grid Colony hybridization, and in particular to one
Grid Colony hybridization dividing method of the kind based on hough-circle transform can be used for detection, the meter of bacterium colony target in grid Colony hybridization
Number and classification.
Background technology
Bacterium colony refers to that the naked eyes that media surface is grown are visible single by microbionation after solid culture primary surface culture
Bacterium group.The colony counts generated after destination sample are counted, are carried out in agricultural, food, medical and health analysis
One basic and important work of quality testing.Colony hybridization refers to after bacterium colony is cultivated on culture medium, with industrial camera pair
Bacterium colony shoots the image to be formed, in order to be counted to bacterium colony target information.In the past for the processing of Colony hybridization and analysis
Work is completed by artificial observation, and in order to improve the analysis to bacterium colony target and counting precision, personnel are usually in grid background for analysis
In bacterium is cultivated, so as to form grid Colony hybridization.However, due to artificial observation have process is complicated, time-consuming,
The shortcomings of efficiency is low, and carries subjectivity, and error is big, and reproducibility is bad, therefore the bacterium colony target under grid background is believed
Breath acquisition difficulty is larger, and precision is relatively low.It can be by operating personnel from this heavy work using the method for image processing and analysis
It frees in work, and greatly improves the precision of counting and analysis.The segmentation of grid Colony hybridization is exactly such technology, it passes through
Analysis to target image detaches bacterium colony target from original image, convenient for obtaining the information of bacterium colony target from original image, with right
Further analysis and the processing of bacterium colony target.
Hough-circle transform belongs to a kind of special case of Hough transformation, is mainly used to detect circular target.Hough-circle transform utilizes
The segment boundary point of circular target finds coordinate where the center of circle, to restore entire circular boundary.Hough-circle transform it is basic
Thinking is to think that each non-zero pixels point is likely to be a bit on a potential circle on image, passes through ballot, generates
Coordinate plane is accumulated, an accumulation weight is set and carrys out setting circle.Standard hough-circle transform is by cartesian coordinate system and three-dimensional coordinate
Tying is closed, and it is exactly a three-dimensional song to be mapped in three-dimensional system of coordinate based on all circles by certain point in cartesian coordinate system
The principle of line by judging whether the quantity of the intersection of every bit in three-dimensional system of coordinate is more than certain threshold value, and determines the three-dimensional
Whether the circle in the corresponding two-dimensional coordinate system of point retains, as final circle fitting result.For improve computational efficiency, it is improved suddenly
Husband's circle transformation is directly handled under two-dimensional coordinate system, such as:By all boundary points as the center of circle, with the most smaller part in input parameter
Size within the scope of diameter and maximum radius is that radius draws circle, and all obtained circles will will produce many intersection points, record hypograph
The number of intersection point at middle corresponding pixel points, the minimum pixel for requiring point number that will be greater than in input parameter are denoted as fitting circle
Central point, using meet at this central point it is most same size circle radius as the radius of fitting circle, to fit round target.
Due to the generally circular in cross section target of bacterium colony target, part bacterium colony target can be found using hough-circle transform, to obtain bacterium colony
The correlated characteristic of target.
Related Colony hybridization cutting techniques are split to non-grid Colony hybridization, and grid bacterium colony figure cannot be handled
Picture, there is presently no the records of the document and technology that find to be split specifically for grid Colony hybridization.
Invention content
It is an object of the invention to for more special grid Colony hybridization, propose a kind of net based on hough-circle transform
Lattice Colony hybridization dividing method existing in the prior art can not achieve the skill being split to grid Colony hybridization for solving
Art problem.
To achieve the above object, the technical solution that the present invention takes includes the following steps:
(1) grid Colony hybridization is pre-processed:
(1.1) the grid bacterium colony coloured image that the size of input is m × n-pixel is converted into grid bacterium colony gray level image
I1, m >=300 and n >=300;
(1.2) grid bacterium colony gray level image I is recalculated using statistics of histogram method1Background value mb, and utilize
Calculated background value is to grid bacterium colony gray level image I1Background be configured, obtain resetting the grid bacterium colony ash of background
Spend image I2;
(1.3) to grid bacterium colony gray level image I2Medium filtering is carried out, the grid bacterium colony gray level image I after denoising is obtained3;
(2) the grid bacterium colony gray level image I after denoising is obtained3The binaryzation boundary image I of middle bacterium colony target5:
(2.1) to the grid bacterium colony gray level image I after denoising3Carry out border detection:
Using border detection algorithm to the grid bacterium colony gray level image I after denoising3Border detection is carried out, grid bacterium colony is obtained
Binaryzation boundary image I4;
(2.2) removal grid bacterium colony binaryzation boundary image I4Grid Edge boundary line:
Calculate grid bacterium colony binaryzation boundary image I4In each length on boundary and the length of minimum enclosed rectangle and width
Degree, and exclude I4Middle length is more than preset parameter a1Or the length-width ratio of minimum enclosed rectangle is more than preset ginseng
Number a2Boundary, obtain the grid bacterium colony gray level image I after denoising3The binaryzation boundary image I of middle bacterium colony target5;
(3) hough-circle transform is used to obtain the grid bacterium colony gray level image I after denoising3The mean radius r of middle bacterium colony target
With average gray value g:
(3.1) to grid bacterium colony gray level image I3The binaryzation boundary image I of middle bacterium colony target5Carry out round fitting:
Setting circle fitting judges parameter b1, and using hough-circle transform to the binaryzation boundary image I of bacterium colony target5It carries out
Circle fitting, obtains the binaryzation boundary image I containing part bacterium colony target6;
(3.2) the binaryzation boundary image I containing part bacterium colony target is utilized6, calculate the grid bacterium colony gray scale after denoising
Image I3The mean radius r and average gray value g of middle bacterium colony target;
(4) hough-circle transform is used to obtain the grid bacterium colony gray level image I after denoising3In candidate bacterium colony target label figure
As mask:
(4.1) to the binaryzation boundary image I of bacterium colony target5Hough-circle transform is carried out, I is obtained3In first kind candidate's mesh
Mark tag image mask1;
(4.2) I is utilized3Middle bacterium colony target average gray value g obtains I3In the second class candidate target tag image mask2;
(4.3) first kind candidate target tag image mask1 and the second class candidate target tag image mask2 are added,
Obtain the grid bacterium colony gray level image I after denoising3In candidate bacterium colony target marker image mask;
(5) segmentation result of input grid Colony hybridization is obtained:
(5.1) by bacterium colony target marker image mask and grid bacterium colony gray level image I1Background value mbCarry out certain number
Student movement is calculated, and obtains excluding the bacterium colony target gray image I after grid lines background12;
(5.2) the grid bacterium colony gray level image I after denoising is obtained3The initial binary image I of middle bacterium colony target13:
To the bacterium colony target gray image I after exclusion grid lines background12Thresholding processing is carried out, the grid after denoising is obtained
Bacterium colony gray level image I3The initial binary image I of middle bacterium colony target13;
(5.3) to initial binary image I13In adhesion target be split:
Using adhesion partitioning algorithm, to initial binary image I13In adhesion target handled, after obtaining denoising
Grid bacterium colony gray level image I3The binary image I of middle bacterium colony target14, and by I14Segmentation knot as input grid Colony hybridization
Fruit.
Compared with prior art, the present invention having the following advantages that:
The present invention obtains the grid bacterium colony gray level image I after denoising using hough-circle transform3Average the half of middle bacterium colony target
Diameter r and average gray value g and candidate bacterium colony target marker image, utilize the mean radius and average gray value and time of acquisition
The bacterium colony target marker image is selected to build new gray-scale map through certain mathematical operation, and using Threshold Segmentation Algorithm to newly building
Gray-scale map is divided, and the segmentation to grid Colony hybridization is realized.
Description of the drawings
Fig. 1 is the implementation flow chart of the present invention;
Fig. 2 is analogous diagram of the present invention to the grid Colony hybridization segmentation with a large amount of bacterium colony targets;
Fig. 3 is analogous diagram of the present invention to the grid Colony hybridization segmentation with smeared out boundary bacterium colony target.
Specific implementation method
Below in conjunction with the drawings and specific embodiments, present invention is further described in detail.
Referring to Fig.1, a kind of grid Colony hybridization dividing method based on hough-circle transform, includes the following steps:
Step 1) pre-processes grid Colony hybridization:
The grid bacterium colony coloured image that the size of input is m × n-pixel is converted to grid bacterium colony gray-scale map by step 1.1)
As I1, m >=300 and n >=300;In the present embodiment, m=1280, n=960;
Step 1.2) recalculates grid bacterium colony gray level image I using statistics of histogram method1Background value mb, and
Using calculated background value to grid bacterium colony gray level image I1Background be configured, obtain the grid bacterium for resetting background
Fall gray level image I2, realize that steps are as follows:
Since culture dish is generally circular in cross section or rectangle, bacterium colony target generally concentrate on image middle section circle or
Rectangular area, however culture dish edge and its external context color change are more random, are influenced on the operation of bacterium colony Target Segmentation very big.
In order to exclude the interference at culture dish edge, to grid bacterium colony gray level image I1In do not include culture dish boundary any position delimit
Effective operating area of arbitrary shape, the shape usually delimited are round or rectangle, and the shape that the present embodiment delimited is circle, is had
Effect operating area refers to comprising bacterium colony mesh target area, then counts the grey level histogram of effective operating area, then by intensity histogram
The corresponding gray value of peak value is as grid bacterium colony gray level image I in figure1Background value mb, then reset grid bacterium colony gray scale
Image I1The gray value of all pixels outside middle operating area is background value mb, obtain the grid bacterium colony ash for resetting background
Spend image I2;
Step 1.3) is to grid bacterium colony gray level image I2Medium filtering is carried out, the grid bacterium colony gray level image after denoising is obtained
I3:
In order to exclude the interference of tiny Gaussian noise, it is filtered herein using 8 neighborhood median filter methods.
Step 2) obtains the grid bacterium colony gray level image I after denoising3The binaryzation boundary image I of middle bacterium colony target5:
Step 2.1) is to the grid bacterium colony gray level image I after denoising3Carry out border detection:
The present embodiment uses CannyBorder detection algorithm is to grid bacterium colony gray level image I3Border detection is carried out, grid is obtained
Bacterium colony binaryzation boundary image I4, including bacterium colony target and the boundary of grid lines;
Step 2.2) removes grid bacterium colony binaryzation boundary image I4Grid Edge boundary line:
Grid bacterium colony binaryzation boundary image I4In result include net boundary, bacterium colony object boundary, in order to improve bacterium
The accuracy for falling target signature acquisition, using grid lines is long and narrow and feature that bacterium colony object boundary class is round excludes the dry of grid lines
It disturbs.I is calculated thus4In each length on boundary and the length and width of minimum enclosed rectangle, and exclude I4Middle length is more than advance
The parameter a of setting1Or the length-width ratio of minimum enclosed rectangle is more than preset parameter a2Boundary, obtain grid bacterium colony ash
Spend image I3The binaryzation boundary image I of middle bacterium colony target5.Wherein, a1Depending on the width w of single grid in image, generally take
a1=w/2, in actual grid Colony hybridization, the width of single grid and the size of grid Colony hybridization are directly proportional, and the present invention is false
If mesh width w is bigAnd it is arrangeda2=3, in the present embodiment, a1=55.
Step 3) obtains the grid bacterium colony gray level image I after denoising using hough-circle transform3Average the half of middle bacterium colony target
Diameter r and average gray value g:
Using the circular feature of bacterium colony target class, to the binaryzation boundary image I of bacterium colony target5Become using hough-circle transform
Swap-in row justifies fit operation, and higher round fitting condition is arranged, that is, is just carried out when having more boundary point to belong to same round
Circle fitting obtains the various features of bacterium colony target accurately to determine the position of part bacterium colony target.Bacterium colony target signature includes shape
State, size, gray value, texture etc. should extract appropriate characteristic information for different bacterium colony targets, and the present invention only calculates bacterium colony
The average gray value g and mean radius size r of target.The specific steps are:
Step 3.1) is to grid bacterium colony gray level image I3The binaryzation boundary image I of middle bacterium colony target5Carry out round fitting:
In order to improve fitting accuracy, the circle fitting of setting judges parameter b1Answer larger, b in the present embodiment1=10, through imitative
Very the results show that the hough-circle transform under this parameter can accurately obtain part bacterium colony target, and using hough-circle transform to bacterium
Fall the binaryzation boundary image I of target5Round fitting is carried out, the binaryzation boundary image I containing part bacterium colony target is obtained6;
Step 3.2) utilizes the binaryzation boundary image I containing part bacterium colony target6, calculate the grid bacterium colony ash after denoising
Spend image I3The mean radius r and average gray value g of middle bacterium colony target, are as follows:
Step 3.2.1) to the binaryzation boundary image I containing part bacterium colony target6Binaryzation Boundary filling is carried out, is obtained
Binary image I containing part bacterium colony target7;
Step 3.2.2) extraction I7In each non-zero connected region coordinate, and count I3In respective coordinates at bacterium colony mesh
The size of bacterium colony target and the average value of gray value is calculated in target size and gray value, obtains the grid bacterium colony after denoising
Gray level image I3The mean radius r and average gray value g of middle bacterium colony target.
Step 4) obtains the grid bacterium colony gray level image I after denoising using hough-circle transform3In candidate bacterium colony target mark
Remember image mask:
Binaryzation boundary image I of the step 4.1) to bacterium colony target5Hough-circle transform is carried out, I is obtained3In the first kind wait
Target marker image mask1 is selected, is as follows:
Step 4.1.1) leak detection to circular colonies target in order to prevent, lower circle fitting is set and judges parameter b2,
B in the present embodiment2=5, through a large amount of simulation results shows, the hough-circle transform under this parameter can obtain most similar round bacterium
Target is fallen, ensures that the bacterium colony target omitted is less, then uses binaryzation boundary image I of the hough-circle transform to bacterium colony target5
Round fitting is carried out, round fitting result image I is obtained8;
Step 4.1.2) to circle fitting result image I8Carry out binaryzation Boundary filling, i.e. round edge circle and its interior zone structure
At unicom region indicate a candidate target, obtain binaryzation circular colonies target image I9;
Step 4.1.3) obtain binaryzation circular colonies target image I9In each circular colonies target coordinate, and count
Calculate the gray level image I after denoising3The average gray value of circular target at middle respective coordinates, the feature as the circular target
Value;
Step 4.1.4) removal binaryzation circular colonies target image I9In, the average gray value of characteristic value and bacterium colony target
G differs by more than preset parameter a3Circular target, the present embodiment takes a3=10, emulation experiment shows that this parameter to obtain
The first kind candidate target obtained is more accurate, obtains the grid bacterium colony gray level image I after denoising3Middle first kind candidate target label
Image mask1, wherein 1 represents candidate target, 0 indicates background;
Step 4.2) utilizes I3Middle bacterium colony target average gray value g obtains I3In the second class candidate target tag image
Mask2 is as follows:
Step 4.2.1) traverse the grid bacterium colony gray level image I after denoising3, will wherein gray value and average gray value g phases
Difference is less than preset parameter a3Pixel as candidate target pixel, obtain binaryzation bacterium colony target image I10;
Step 4.2.2) to binaryzation bacterium colony target image I10It carries out the filling of binaryzation target and obtains binaryzation bacterium colony target
Image I11;
Step 4.2.3) calculate binaryzation bacterium colony target image I11In each bacterium colony target radius and circularity size, exclude
I11Middle radius differs by more than preset parameter a with mean radius size r4Bacterium colony target, for further exclusive segment
The interference of impurity, calculates the circularity C of bacterium colony target, and excludes circularity C and be less than a5Bacterium colony target, the present embodiment takes a4=5, a5
=0.6, experiment shows that the two parameters can effectively eliminate the candidate bacterium colony target of error detection, obtains the grid bacterium after denoising
Fall gray level image I3In the second class candidate target tag image mask2, wherein 1 represents candidate target, 0 indicates background, another circle
Degree is defined as:
C=4 π S2/L2, the area of S expression bacterium colony targets, the boundary line length of L expression bacterium colony targets, the reflection of circularity size
The degree of closeness of bacterium colony target and circle is indicated closer to 1 closer to circle;
Step 4.3) is by first kind candidate target tag image mask1 and the second class candidate target tag image mask2 phases
Add, obtains the grid bacterium colony gray level image I after denoising3In candidate bacterium colony target marker image mask.
Step 5) obtains the segmentation result of input grid Colony hybridization:
Step 5.1) is by bacterium colony target marker image mask and grid bacterium colony gray level image I1Background value mbUtilize formula:
I12=I3×mask+mb× (1-mask) is calculated, and obtains excluding the bacterium colony target gray image I after grid lines background12;
Step 5.2) obtains the grid bacterium colony gray level image I after denoising3The initial binary image I of middle bacterium colony target13:
To the bacterium colony target gray image I after exclusion grid lines background12Ostu threshold divisions are carried out, after obtaining denoising
Grid bacterium colony gray level image I3The initial binary image I of middle bacterium colony target13;
Step 5.3) is to initial binary image I13In adhesion target be split:
Using the fractional spins based on seed point, to initial binary image I13In adhesion target at
Reason, obtains the grid bacterium colony gray level image I after denoising3The binary image I of middle bacterium colony target14, and by I14As input grid
The segmentation result of Colony hybridization.
Below in conjunction with emulation experiment, the technique effect of the present invention is described further.
1, emulation content and condition:
For verification effectiveness of the invention and correctness, emulation experiment is carried out using two kinds of grid Colony hybridization.
Emulation one, for the experiment being split to the grid Colony hybridization containing a large amount of bacterium colony targets using the present invention;Emulation two, for profit
The experiment that the grid Colony hybridization containing fuzzy edge bacterium colony target is split with the present invention.All emulation experiments exist
It is realized using the libraries VS2015 software+OpenCv under 7 operating systems of Windows.
2, analysis of simulation result:
Shown in result such as Fig. 2 (e) that emulation one generates.
Reference Fig. 2,
Fig. 2 (a) is the gray-scale map of the grid Colony hybridization of input;
Fig. 2 (b) be to the pretreated grid bacterium colony gray level image of the grid Colony hybridization of input, can from Fig. 2 (b)
Background heel row is reset to the grid bacterium colony gray level image to input in addition to the non-uniform interference of culture dish boundary gray value,
Simultaneously so that the grey scale pixel value removed outside grid lines and bacterium colony target reaches unanimity;
Fig. 2 (c) is the binaryzation boundary image of bacterium colony target in pretreated grid bacterium colony gray level image, from Fig. 2 (c)
It can be seen that grid lines is excluded substantially;
Fig. 2 (d) is the candidate target that hough-circle transform will be used to obtain, and is shown in grid bacterium colony gray level image after pretreatment
As a result, wherein white connected region indicates candidate bacterium colony target, from Fig. 2 (d) it can be seen that candidate bacterium colony target covers substantially
All bacterium colony targets, but be all not real bacterium colony target;
Fig. 2 (e) is the final segmentation result of the present invention, wherein white connected region indicates bacterium colony target, from Fig. 2 (e)
It can be seen that having eliminated the false target in candidate target, obtained grid Colony hybridization segmentation result is more accurate.
Shown in result such as Fig. 3 (e) that emulation two generates.
Reference Fig. 3,
Fig. 3 (a) is the gray-scale map of the grid Colony hybridization of input;
Fig. 3 (b) be to the pretreated grid bacterium colony gray level image of the grid Colony hybridization of input, can from Fig. 3 (b)
Background heel row is reset to the grid bacterium colony gray level image to input in addition to the non-uniform interference of culture dish boundary gray value,
Simultaneously so that the grey scale pixel value removed outside grid lines and bacterium colony target reaches unanimity;
Fig. 3 (c) is the binaryzation boundary image of bacterium colony target in pretreated grid bacterium colony gray level image, from Fig. 3 (c)
It can be seen that grid lines is excluded substantially, although bacterium colony object edge is more fuzzy, still detected most of
The segment boundary of bacterium colony target;
Fig. 3 (d) is the candidate target that hough-circle transform will be used to obtain, and is shown in pretreated grid bacterium colony gray-scale map
Picture as a result, wherein white connected region indicates candidate bacterium colony target, from Fig. 3 (d) it can be seen that candidate bacterium colony target is contained substantially
All bacterium colony targets have been covered, but have all been not real bacterium colony targets;
Fig. 3 (e) is the final segmentation result of the present invention, wherein white connected region indicates bacterium colony target, from Fig. 3 (e)
It can be seen that having eliminated the false target in candidate target, obtained grid Colony hybridization segmentation result is more accurate.
The above emulation experiment can be seen that either grid Colony hybridization of the segmentation containing a large amount of bacterium colony targets, or divide
The more fuzzy grid Colony hybridization of object boundary is cut, the present invention can obtain accurate segmentation result.
Claims (6)
1. a kind of grid Colony hybridization dividing method based on hough-circle transform, which is characterized in that include the following steps:
(1) grid Colony hybridization is pre-processed:
(1.1) the grid bacterium colony coloured image that the size of input is m × n-pixel is converted into grid bacterium colony gray level image I1,m≥
300 and n >=300;
(1.2) grid bacterium colony gray level image I is recalculated using statistics of histogram method1Background value mb, and utilize calculating
The background value gone out is to grid bacterium colony gray level image I1Background be configured, obtain the grid bacterium colony gray-scale map for resetting background
As I2;
(1.3) to grid bacterium colony gray level image I2Medium filtering is carried out, the grid bacterium colony gray level image I after denoising is obtained3;
(2) the grid bacterium colony gray level image I after denoising is obtained3The binaryzation boundary image I of middle bacterium colony target5:
(2.1) to the grid bacterium colony gray level image I after denoising3Carry out border detection:
Using border detection algorithm to the grid bacterium colony gray level image I after denoising3Border detection is carried out, grid bacterium colony two-value is obtained
Change boundary image I4;
(2.2) removal grid bacterium colony binaryzation boundary image I4Grid Edge boundary line:
Calculate grid bacterium colony binaryzation boundary image I4In each length on boundary and the length and width of minimum enclosed rectangle, and
Exclude I4Middle length is more than preset parameter a1Or the length-width ratio of minimum enclosed rectangle is more than preset parameter a2's
Boundary obtains the grid bacterium colony gray level image I after denoising3The binaryzation boundary image I of middle bacterium colony target5;
(3) hough-circle transform is used to obtain the grid bacterium colony gray level image I after denoising3The mean radius r of middle bacterium colony target and average
Gray value g:
(3.1) to grid bacterium colony gray level image I3The binaryzation boundary image I of middle bacterium colony target5Carry out round fitting:
Setting circle fitting judges parameter b1, and using hough-circle transform to the binaryzation boundary image I of bacterium colony target5It is quasi- to carry out circle
It closes, obtains the binaryzation boundary image I containing part bacterium colony target6;
(3.2) the binaryzation boundary image I containing part bacterium colony target is utilized6, calculate the grid bacterium colony gray level image I after denoising3
The mean radius r and average gray value g of middle bacterium colony target;
(4) hough-circle transform is used to obtain the grid bacterium colony gray level image I after denoising3In candidate bacterium colony target marker image
mask:
(4.1) to the binaryzation boundary image I of bacterium colony target5Hough-circle transform is carried out, I is obtained3In first kind candidate target mark
Remember image mask1;
(4.2) I is utilized3Middle bacterium colony target average gray value g obtains I3In the second class candidate target tag image mask2;
(4.3) first kind candidate target tag image mask1 and the second class candidate target tag image mask2 are added, are obtained
Grid bacterium colony gray level image I after denoising3In candidate bacterium colony target marker image mask;
(5) segmentation result of input grid Colony hybridization is obtained:
(5.1) by bacterium colony target marker image mask and grid bacterium colony gray level image I1Background value mbCarry out certain mathematics fortune
It calculates, obtains excluding the bacterium colony target gray image I after grid lines background12;
(5.2) the grid bacterium colony gray level image I after denoising is obtained3The initial binary image I of middle bacterium colony target13:
To the bacterium colony target gray image I after exclusion grid lines background12Thresholding processing is carried out, the grid bacterium colony after denoising is obtained
Gray level image I3The initial binary image I of middle bacterium colony target13;
(5.3) to initial binary image I13In adhesion target be split:
Using adhesion partitioning algorithm, to initial binary image I13In adhesion target handled, obtain the grid after denoising
Bacterium colony gray level image I3The binary image I of middle bacterium colony target14, and by I14Segmentation result as input grid Colony hybridization.
2. the grid Colony hybridization dividing method according to claim 1 based on hough-circle transform, which is characterized in that step
(1.2) the use statistics of histogram method described in recalculates grid bacterium colony gray level image I1Background value mb, and utilize
Calculated background value is to grid bacterium colony gray level image I1Background be configured, obtain resetting the grid bacterium colony ash of background
Spend image I2, realize that steps are as follows:
In grid bacterium colony gray level image I1In include culture dish edge any position delimit arbitrary shape effective operating space
Domain, and the grey level histogram of effective operating area is counted, then using the corresponding gray value of peak value in grey level histogram as grid bacterium
Fall gray level image I1Background value mb, then reset grid bacterium colony gray level image I1All pixels outside middle operating area
Gray value be background value mb, obtain the grid bacterium colony gray level image I for resetting background2。
3. the grid Colony hybridization dividing method according to claim 1 based on hough-circle transform, which is characterized in that step
(3.2) binaryzation boundary image I of the utilization containing part bacterium colony target described in6, calculate the grid bacterium colony gray scale after denoising
Image I3The mean radius r and average gray value g of middle bacterium colony target, are as follows:
(3.2.1) is to the binaryzation boundary image I containing part bacterium colony target6Binaryzation Boundary filling is carried out, is obtained containing part
The binary image I of bacterium colony target7;
(3.2.2) extracts I7In each non-zero connected region coordinate, and count I3In respective coordinates at bacterium colony target radius
And gray value, the radius of bacterium colony target and the average value of gray value is calculated, obtains the grid bacterium colony gray level image after denoising
I3The mean radius r and average gray value g of middle bacterium colony target.
4. the grid Colony hybridization dividing method according to claim 1 based on hough-circle transform, which is characterized in that step
(4.1) the binaryzation boundary image I to bacterium colony target described in5Hough-circle transform is carried out, I is obtained3In first kind candidate's mesh
Tag image mask1 is marked, is as follows:
(4.1.1) sets circle fitting and judges parameter as b2, using hough-circle transform to the binaryzation boundary image I of bacterium colony target5Into
Row circle fitting obtains round fitting result image I8;
(4.1.2) is to circle fitting result image I8Binaryzation Boundary filling is carried out, binaryzation circular colonies target image I is obtained9;
(4.1.3) obtains binaryzation circular colonies target image I9In each circular colonies target coordinate, and after calculating denoising
Gray level image I3The average gray value of circular target at middle respective coordinates, the characteristic value as the circular target;
(4.1.4) removes binaryzation circular colonies target image I9In, the average gray value g of characteristic value and bacterium colony target differs greatly
In preset parameter a3Circular target, obtain the grid bacterium colony gray level image I after denoising3Middle first kind candidate target mark
Remember image mask1.
5. the grid Colony hybridization dividing method according to claim 1 based on hough-circle transform, which is characterized in that step
(4.2) I is utilized described in3Middle bacterium colony target average gray value g obtains I3In the second class candidate target tag image mask2,
It is as follows:
(4.2.1) traverses the grid bacterium colony gray level image I after denoising3, wherein gray value and average gray value g differences is less than in advance
The parameter a first set4Pixel as candidate target pixel, obtain binaryzation bacterium colony target image I10;
(4.2.2) is to binaryzation bacterium colony target image I10It carries out the filling of binaryzation target and obtains binaryzation bacterium colony target image I11;
(4.2.3) calculates binaryzation bacterium colony target image I11In each bacterium colony target radius and circularity size, exclude I11In half
Diameter differs by more than preset parameter a with mean radius size r5Bacterium colony target, and exclude binaryzation bacterium colony target image
I11Middle circularity C is less than preset parameter a6Bacterium colony target, obtain the grid bacterium colony gray level image I after denoising3In
Two class candidate target tag image mask2, wherein circularity are defined as:
C=4 π S2/L2, the area of S expression bacterium colony targets, the boundary line length of L expression bacterium colony targets.
6. the grid Colony hybridization dividing method according to claim 1 based on hough-circle transform, which is characterized in that step
(5.1) described in by bacterium colony target marker image mask and grid bacterium colony gray level image I1Background value mbCarry out certain number
Student movement is calculated, and obtains excluding the bacterium colony target gray image I after grid lines background12, concrete operations are as follows:
Utilize formula:I12=I3×mask+mb× (1-mask) obtains excluding the bacterium colony target gray image after grid lines background
I12。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810086989.0A CN108376403B (en) | 2018-01-30 | 2018-01-30 | Grid colony image segmentation method based on Hough circle transformation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810086989.0A CN108376403B (en) | 2018-01-30 | 2018-01-30 | Grid colony image segmentation method based on Hough circle transformation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108376403A true CN108376403A (en) | 2018-08-07 |
CN108376403B CN108376403B (en) | 2020-02-07 |
Family
ID=63016979
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810086989.0A Expired - Fee Related CN108376403B (en) | 2018-01-30 | 2018-01-30 | Grid colony image segmentation method based on Hough circle transformation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108376403B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110334601A (en) * | 2019-06-04 | 2019-10-15 | 武汉极目智能技术有限公司 | A kind of speed(-)limit sign board recognition methods of combination machine learning and computer vision |
CN111368643A (en) * | 2020-02-12 | 2020-07-03 | 杭州电子科技大学 | Escherichia coli dynamic growth monitoring method |
CN111429361A (en) * | 2020-02-17 | 2020-07-17 | 马健锦 | Bacterial colony counting method based on machine vision |
CN113110265A (en) * | 2021-05-26 | 2021-07-13 | 焦作大学 | Fermentation environment adjusting method for biochemical industry by adopting Internet of things |
CN113817585A (en) * | 2021-09-10 | 2021-12-21 | 武汉理工大学 | Bacterial colony selection comprehensive workstation and selection method |
CN114723804A (en) * | 2022-04-13 | 2022-07-08 | 盐城工学院 | Colony counting method based on image classification |
CN116309193A (en) * | 2023-05-23 | 2023-06-23 | 武汉大学人民医院(湖北省人民医院) | Image processing method and device for immune sample |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2546803A1 (en) * | 2008-09-18 | 2013-01-16 | Molecular Devices (New Milton) Ltd | Colony detection |
CN103164858A (en) * | 2013-03-20 | 2013-06-19 | 浙江大学 | Adhered crowd segmenting and tracking methods based on superpixel and graph model |
CN105574528A (en) * | 2015-12-15 | 2016-05-11 | 安徽工业大学 | Synechia cell image segmenting method based on polyphase mutual exclusion level set |
CN106447020A (en) * | 2016-09-13 | 2017-02-22 | 上海海事大学 | Intelligent bacterium colony counting method |
-
2018
- 2018-01-30 CN CN201810086989.0A patent/CN108376403B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2546803A1 (en) * | 2008-09-18 | 2013-01-16 | Molecular Devices (New Milton) Ltd | Colony detection |
CN103164858A (en) * | 2013-03-20 | 2013-06-19 | 浙江大学 | Adhered crowd segmenting and tracking methods based on superpixel and graph model |
CN105574528A (en) * | 2015-12-15 | 2016-05-11 | 安徽工业大学 | Synechia cell image segmenting method based on polyphase mutual exclusion level set |
CN106447020A (en) * | 2016-09-13 | 2017-02-22 | 上海海事大学 | Intelligent bacterium colony counting method |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110334601A (en) * | 2019-06-04 | 2019-10-15 | 武汉极目智能技术有限公司 | A kind of speed(-)limit sign board recognition methods of combination machine learning and computer vision |
CN111368643A (en) * | 2020-02-12 | 2020-07-03 | 杭州电子科技大学 | Escherichia coli dynamic growth monitoring method |
CN111368643B (en) * | 2020-02-12 | 2023-04-14 | 杭州电子科技大学 | Escherichia coli dynamic growth monitoring method |
CN111429361A (en) * | 2020-02-17 | 2020-07-17 | 马健锦 | Bacterial colony counting method based on machine vision |
CN113110265A (en) * | 2021-05-26 | 2021-07-13 | 焦作大学 | Fermentation environment adjusting method for biochemical industry by adopting Internet of things |
CN113817585A (en) * | 2021-09-10 | 2021-12-21 | 武汉理工大学 | Bacterial colony selection comprehensive workstation and selection method |
CN114723804A (en) * | 2022-04-13 | 2022-07-08 | 盐城工学院 | Colony counting method based on image classification |
CN116309193A (en) * | 2023-05-23 | 2023-06-23 | 武汉大学人民医院(湖北省人民医院) | Image processing method and device for immune sample |
CN116309193B (en) * | 2023-05-23 | 2023-08-25 | 武汉大学人民医院(湖北省人民医院) | Image processing method and device for immune sample |
Also Published As
Publication number | Publication date |
---|---|
CN108376403B (en) | 2020-02-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108376403A (en) | Grid Colony hybridization dividing method based on hough-circle transform | |
Malambo et al. | Automated detection and measurement of individual sorghum panicles using density-based clustering of terrestrial lidar data | |
CN107292310B (en) | Visual positioning and automatic reading method for circular pointer type dial plate | |
CN109325935B (en) | Power transmission line detection method based on unmanned aerial vehicle image | |
CN104282026B (en) | Distributing homogeneity appraisal procedure based on watershed algorithm and minimum spanning tree | |
US9330435B2 (en) | Bare earth finding and feature extraction for 3D point clouds | |
CN104182985B (en) | Remote sensing image change detection method | |
CN105913081B (en) | SAR image classification method based on improved PCAnet | |
CN104990925A (en) | Defect detecting method based on gradient multiple threshold value optimization | |
CN106951905A (en) | Apple identification and localization method on a kind of tree based on TOF camera | |
CN107507170A (en) | A kind of airfield runway crack detection method based on multi-scale image information fusion | |
CN109523565A (en) | A kind of diffraction light-free Moire fringe center positioning method and system | |
CN108052886A (en) | A kind of puccinia striiformis uredospore programming count method of counting | |
CN103745478A (en) | Machine vision determination method for wheat germination rate | |
CN107220647B (en) | Crop center point positioning method and system under blade crossing condition | |
CN115115627B (en) | Soil saline-alkali soil monitoring method based on data processing | |
CN112085675B (en) | Depth image denoising method, foreground segmentation method and human motion monitoring method | |
CN106778749A (en) | Based on the touring operating area boundary extraction method that concentration class and Delaunay triangles are reconstructed | |
CN105512622B (en) | A kind of visible remote sensing image sea land dividing method based on figure segmentation and supervised learning | |
Robb et al. | A semi-automated method for mapping glacial geomorphology tested at Breiðamerkurjökull, Iceland | |
CN112666527B (en) | Weather radar filtering method fusing live precipitation data | |
CN104123569B (en) | Video person number information statistics method based on supervised learning | |
Zeng et al. | Detecting and measuring fine roots in minirhizotron images using matched filtering and local entropy thresholding | |
CN106991441A (en) | Merge the plant specimen sorting technique and system of multiple dimensioned direction textural characteristics | |
CN108447045B (en) | SAR remote sensing image water area detection method based on SAT integral graph |
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 | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20200901 Address after: 400030, stone road 70, Shapingba District, Chongqing, attached to 4-15-3 Patentee after: CHONGQING JIANGXUE TECHNOLOGY Co.,Ltd. Address before: 710071, No. 2 Taibai South Road, Yanta District, Shaanxi, Xi'an Patentee before: XIDIAN University |
|
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200207 Termination date: 20210130 |