CN102494987A - Automatic category rating method for microscopic particles in nodular cast iron - Google Patents

Abstract

The invention discloses an automatic category rating method for microscopic particles in nodular cast iron, which relates to metallographic microscopic image data processing and includes the steps: acquiring images; inputting the estimation value of the category number of the categorical microscopic particles by a client; preprocessing the images; clustering and categorizing the images; and extracting and computing morphological characteristics of the microscopic particles in the images, and distinguishing and rating categories of the microscopic particles in the nodular cast iron according to corresponding national standards. By integrally applying improved K-means clustering algorithm and image partitioning algorithm and by combining metallographic characteristic values, the images are partitioned into the corresponding number of sample images according the categories of the microscopic particles in the images, each partitioned image only contains one category of microscopic particles, and a user can select proper microscopic particles for category rating as needed, so that the automatic category rating method overcomes the shortcoming that processing for metallographic images by means of existing metallographic microscopic analysis technology, particularly processing for the metallographic images containing various textures lacks accuracy and authority of inspection.

Description

The automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron

Technical field

Technical scheme of the present invention relates to the metallurgical microscopic image data processing, specifically the automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron.

Background technology

Metallographic microanalysis is a kind of very important research method in the metal material scientific research, and it can be observed and study in the metal and can't observedly organize details and defective with the macroanalysis method.Because the size of metallographic geometric parameter is very big to effect of material performance, its detection plays a part very important in Analysis of Metallic Materials.

In existing metallographic microanalysis technology; The difference of particle is little in the metallurgical microscopic image, and artificial distinguish is judged, and judged the test stone disunity; The human factor influence is very big, the accuracy of check is controlled with authoritative and article all produced very big negative effect.Often be mingled with a variety of materials in the metallic phase image; As in the spheroidal-graphite cast iron or in certain steel; Often contain multiple materials such as spheroidal graphite, pearlite and ferrite; And metallographic laboratory will go checking measurements with different standards to different materials, and the checking measurements of all putting together will certainly produce disturbing effect like these snotteres, makes to measure the rating result accuracy and reduce.

CN201010199148 discloses a kind of method of coloured image and gray level image being cut apart automatically based on figure cutting.Mainly solving existing figure cutting technique needs the problem of mutual, modeling and manual modification segmentation result in image segmentation, but the processing of the metallic phase image that contains multiple tissue is lacked adaptability and accuracy.

CN201010262281 has disclosed the improvement to the metallographic microstructure image processing method.Processed steps is following: gather metallograph through electron microscope; The image model of regulation metallograph carries out image filtering to the image that transforms to the HIS pattern, and filtered image is carried out the Hotelling conversion; Adopt the artificial neural network clustering algorithm to handle image; Image characteristics extraction, the number of the pixel in all crystal grains element after calculate extracting obtains the area percentage of crystal grain.This method can not accurate recognition with judge the title that cut apart the back tissue, the processing of the metallic phase image that contains multiple tissue is also lacked adaptability and accuracy.

Summary of the invention

Technical matters to be solved by this invention is: the automatic classification ranking method that micro-particle in the spheroidal-graphite cast iron is provided; Employing is with improved K means clustering algorithm and image segmentation algorithm fusion application and combine the means of the eigenwert of metallographic; Be divided into the sample image of respective numbers according to the kind of micro-particle in the image; A spheroidal-graphite cast iron metallographic sample image is come by different micro-separate particles; Only comprise a kind of micro-particle in every image after cutting apart; The user can select the grading of classifying of suitable micro-particle according to the demand of oneself, has overcome the processing of existing metallographic microanalysis technology to metallic phase image, and particularly processing that contain the metallic phase image of multiple tissue lack accuracy and the authoritative shortcoming of checking more.

The present invention solves this technical problem the technical scheme that is adopted: the automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron, and concrete steps are following:

The first step, images acquired

Gather the metallic phase image of spheroidal-graphite cast iron through optical microscope and CCD camera; The resolution of used CCD camera is in 1,300,000～3,000,000 pixels; The communication of employing USB interface; Number of pictures per second is greater than 10 frames during video acquisition, capture effect wherein preferably a frame save as the image of BMP or JPG form, number of pictures per second was 11～31 frames when wherein preferred video was gathered;

Second step, the micro-particle classification number estimated value of client's input category

Consider the quantitative property of the metallographic structure grading of spheroidal-graphite cast iron, the micro-particle clusters number estimated value of confirming client's input category is k=2～5 type;

The 3rd step, the image pre-service

To pass through the image preliminary treatment by the image that the first step is gathered; Its operating process is earlier image to be carried out gray processing to handle; Then the triple channel image transitions is become the single channel image; Select to carry out filtering operation based on the user again and make image sharpening, the filtering method that adopts two kinds of filtering of gaussian filtering and medium filtering to combine here;

The 4th step, the image clustering classification

With improved K means clustering algorithm and image segmentation algorithm fusion application, promptly adopt the dividing method of many threshold values iteration and the K means clustering algorithm is incorporated wherein, to the pretreated image clustering classification of the 3rd step image, concrete steps are:

4.1 for each cluster is confirmed an initial cluster centre, there be k cluster centre in k cluster like this,

4.2 each sample in the sample set is assigned to some in k the cluster according to minimal distance principle,

4.3 the average of using all samples in each cluster is as new cluster centre,

If repeated for 4.2 and 4.3 steps till cluster centre no longer changes 4.4 cluster centre changes,

4.5 the k that obtains an at last cluster centre is exactly a clustering result,

4.6 find out the brightness in the image, through alternative manner its classification be labeled as 1,2,3 ... The K class;

In the 5th step, the morphological feature of the micro-particle in extraction and the computed image is distinguished the kind of micro-particle in the spheroidal-graphite cast iron and is confirmed rank in conjunction with corresponding GB

To extract the morphological feature of micro-particle by the metallic phase image of the good spheroidal-graphite cast iron of the 4th step image clustering classification; The nodularization rate, area occupation ratio, particle size and the pearlite quantity that comprise micro-particle; Through calculating the morphological feature of these micro-particles; Distinguish the kind of micro-particle in the spheroidal-graphite cast iron and confirm rank in conjunction with corresponding national standards, concrete steps are following:

5.1 the differentiation of the micro-particle kind of graphite in the spheroidal-graphite cast iron

Pass through to calculate the nodularization rate and the area occupation ratio of the micro-particle of each graphite earlier,

The nodularization rate is calculated: $v=\frac{1\×n_{1.0}+0.8\×n_{0.8}+0.6\×n_{0.6}+0.3\×n_{0.3}+0\×n_0}{n_{1.0}+n_{0.8}+n_{0.6}+n_{0.3}+n_0}---\left(I\right)$

In the formula, n _1.0, n _0.8, n _0.6, n _0.3, n ₀The micro-particle of a graphite number of representing five kinds of spherical correction factors respectively,

Area occupation ratio calculates: $u=\frac{S_i}{{\mathrm{\π\; r}}^2}---\left(\mathrm{II}\right)$

In the following formula, S _iBe the real area of the micro-particle of each graphite, unit is mm ², r is the circumradius of the micro-particle of each graphite, unit is mm,

(I) the micro-particle of a graphite number of five kinds of spherical correction factors in the formula and (II) the formula corresponding relation that calculates the area occupation ratio magnitude range of gained is respectively: n _1.0Corresponding u>=0.81, n _0.8Corresponding u=0.80～0.61, n _0.6Corresponding u=0.60～0.41, n _0.3Corresponding u=0.40～0.21 and n ₀Corresponding u≤0.20,

Distinguish the kind of the micro-particle of graphite in the spheroidal-graphite cast iron nodularization again with area occupation ratio, contrast with GB9441-88 by the area occupation ratio numerical value that calculates, the micro-particle area occupation ratio of graphite less than 0.2 be flake graphite; Greater than 0.2 to less than 0.4 be quasiflake graphite; Greater than 0.4 to less than 0.6 be temper carbon; Greater than 0.6 to less than 0.8 be nodular graphite; Greater than 0.8 to less than 1.0 be globular graphite;

5.2 the size of the micro-particle of graphite grading in the spheroidal-graphite cast iron

The computing formula of calculating the micro-average particle diameter of each graphite

does

$\stackrel{\‾}{D}=2\×\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{S}_i}{\mathrm{\π}(n_0+n_1+n_2+n_3+n_4+n_5)}}---\left(\mathrm{III}\right)$

In the following formula, S _iBe the real area of the micro-particle of each graphite, unit is mm ², n ₀, n ₁, n ₂, n ₃, n ₄And n ₅The scope of the micro-particle diameter D of corresponding respectively graphite is: greater than 25 to 50, greater than 12 to 25, greater than 6 to 12, greater than 4 to 6, greater than 1.5 to 4 and≤1.5 the micro-particle number of graphite,

Mean diameter

numerical value and GB GB 9441-88 contrast by the micro-particle size of graphite in the spheroidal-graphite cast iron that calculates; Judge the affiliated rank of the micro-particle size of different graphite; Here stipulate according to GB GB 9441-88; The scope of the micro-average particle diameter of graphite

is: the rank greater than 25 to 50 is 3 grades; Rank greater than 12 to 25 is 4 grades; Rank greater than 6 to 12 is 5 grades; Rank greater than 3 to 6 is 6 grades, and the rank greater than 1.5 to 3 is 7 grades, and≤1.5 rank is 8 grades;

5.3 spheroidal-graphite cast iron medium pearlite quantity grading

Calculate pearlitic quantity number percent M, computing formula is:

$M=\frac{\underset{i}{\mathrm{\Σ}}{S}_i}{S_{\mathrm{all}}}\×100\%---\left(\mathrm{IV}\right)$

S in the following formula _iBe the real area of the micro-particle of each pearlite, S _AllBe the size of entire image, unit is mm ²,

With pearlite quantity number percent that is calculated and GB GB 9441-88 contrast, judge pearlite quantity rank, GB GB 9441-88 regulation; Pearlite quantity percentage range more than or equal to 90 be pearl 95, more than or equal to 80 to less than 90 be pearl 85, more than or equal to 70 to less than 80 be pearl 75; More than or equal to 60 to less than 70 be pearl 65, more than or equal to 50 to less than 60 be pearl 55, more than or equal to 40 to less than 50 be pearl 45; More than or equal to 30 to less than 40 be pearl 35, ≈'s 25 is pearl 25, ≈'s 15 is pearl 15; ≈'s 10 is pearl 10, and ≈'s 5 is pearl 5.

The automatic classification ranking method of micro-particle in the above-mentioned spheroidal-graphite cast iron; The flow process of said image segmentation algorithm is: begin → confirm cluster centre → distribution sample; Is sample average that new cluster centre → center changes? → be, return the distribution sample, sample average is new cluster centre; Not, find out the brightness of image characteristic, iteration is sorted out.

The automatic classification ranking method of micro-particle in the above-mentioned spheroidal-graphite cast iron; Wherein the enforcement of five steps is to be controlled according to following techniqueflow by computing machine: beginning → images acquired → image pre-service → confirm cluster centre → distribution sample; Is sample average that new cluster centre → center changes? → be; Return the distribution sample, sample average is new cluster centre; Not, find out the brightness of image characteristic, iteration is sorted out.

Method of the present invention also be applicable in other alloys in the automatic classification grading of micro-particle, for example the automatic classification of micro-particle is graded in gray iron and the aluminium alloy.Can be applicable to that material analysis and other fields relate to less than 6 kinds of classification of substances analyses.

The invention has the beneficial effects as follows: in existing metallographic microanalysis technology, the identification of metallography microscope particle also still is a technological gap and difficult point, and the inventive method proposes to show that for the classification of sample this is the initiative in present technique field.

Compared with prior art, the substantive distinguishing features that the automatic classification ranking method of micro-particle is had in the spheroidal-graphite cast iron of the present invention is:

(1) purpose of " second step, the micro-particle classification number estimated value of client's input category " of the inventive method is for improving the degree of accuracy of classification.Because the client imports the micro-particle classification number of image to be classified, just can the contained micro-particle of metallic phase image be classified automatically, and extract the measurement grading to different micro-particles respectively.Here be the supervision behavior, need the micro-particle classification number of client's input category.After classifying automatically, need to measure which kind of micro-particle by the people for going judgement again.

(2) in " the 3rd step, the image pre-service " of the inventive method, gaussian filtering comes down to a kind of wave filter of signal, and its purposes is the smoothing processing of signal, is used for the noise problem of processing digital images later stage application, obtains the better image edge; Medium filtering is based on the theoretical a kind of nonlinear signal processing technology that can effectively suppress noise of sequencing statistical; The ultimate principle of medium filtering is to replace the value of any in digital picture or the Serial No. with the Mesophyticum of each point value in the neighborhood of this point; The approaching actual value of pixel value around letting, thus isolated noise spot eliminated.The filtering method that adopts gaussian filtering and two kinds of filtering of medium filtering to combine makes image pattern clear, is convenient to classification processing.If uneven illumination is even, image has the situation of shade obvious difference, causes subsequent operation to produce deviation easily, can carry out the equalization operation here, makes the gray balance of image.

Compared with prior art, the marked improvement of the automatic classification ranking method of micro-particle is in the spheroidal-graphite cast iron of the present invention:

(1) the image intelligent sorting technique is applied in the metallographic examination software, shows the different sample image in classification back, reduced client's artificial judgement, alleviated the workload of metallographic examination greatly to the client.

(2) the inventive method has improved the accuracy of measuring greatly, has eliminated the interference of artificial subjectivity and extraneous objective factor, and follow-up measurement grading is all laid a good foundation.In supervision accurately and under the qualified situation of picture, the accuracy rate of classification＞95%.

(3) increase work efficiency, reduced the time of metal lographic examination greatly.

Description of drawings

Below in conjunction with accompanying drawing and embodiment the present invention is further specified.

Fig. 1 is the implementation step route map of the automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron of the present invention.

The Computer Control Technology process flow diagram that Fig. 2 implements for the automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron of the present invention.

Fig. 3 is the process flow diagram of the image segmentation algorithm of the automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron of the present invention.

Fig. 4 is the recognition effect figure of the spheroidal-graphite cast iron sample 1 of the inventive method embodiment 1.

Fig. 5 is the classifying quality figure of the spheroidal-graphite cast iron sample 2 of the inventive method embodiment 2.

Fig. 6 is the classifying quality figure of the spheroidal-graphite cast iron sample 3 of the inventive method embodiment 3.

Among the figure, 1. exhibit red is a flake graphite, shows that 2. yellow is a quasiflake graphite, shows that 3. blueness is a temper carbon, shows that 4. green is a nodular graphite, shows that 5. purple is a globular graphite.

Embodiment

Embodiment illustrated in fig. 1 showing, the implementation step of the automatic classification ranking method of micro-particle is in the spheroidal-graphite cast iron of the present invention:

The first step; Import individual sample image; Individual sample image of this input is a metallic phase image of gathering spheroidal-graphite cast iron through optical microscope and CCD camera, and the resolution of used CCD camera adopts USB interface communication in 1,300,000～3,000,000 pixels; Number of pictures per second is greater than 10 frames during video acquisition, capture effect wherein preferably a frame save as the image of BMP or JPG form;

In second step, the micro-particle classification number estimated value of client's input category, this estimated value are k=2～5 type;

The 3rd step, image pre-service, the filtering method that adopts gaussian filtering and two kinds of filtering of medium filtering to combine;

In the 4th step, the image clustering classification with improved K means clustering algorithm and image segmentation algorithm fusion application, is promptly adopted the dividing method of many threshold values iteration, and the K means clustering algorithm is incorporated wherein flat;

In the 5th step, the morphological feature of the micro-particle in extraction and the computed image is distinguished the kind of micro-particle in the spheroidal-graphite cast iron and is confirmed rank in conjunction with corresponding GB.

Embodiment illustrated in fig. 2 showing; The Computer Control Technology flow process that five steps are implemented in the automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron of the present invention is: beginning → images acquired → image pre-service → confirm cluster centre → distribution sample; Is sample average that new cluster centre → center changes? → be; Return the distribution sample, sample average is new cluster centre; Not, find out the brightness of image characteristic, iteration is sorted out.

Embodiment illustrated in fig. 3 showing; The flow process of the image segmentation algorithm of the automatic classification ranking method of micro-particle is in the spheroidal-graphite cast iron of the present invention: begin → confirm cluster centre → distribution sample; Is sample average that new cluster centre → center changes? → be; Return the distribution sample, sample average is new cluster centre; Not, find out the brightness of image characteristic, iteration is sorted out.

Embodiment 1

The automatic classification grading of micro-particle in the spheroidal-graphite cast iron sample 1

The first step, images acquired

Gather the metallic phase image of tested spheroidal-graphite cast iron through optical microscope and CCD camera; The resolution of used CCD camera is in 2,000,000 pixels; Employing USB interface communication, number of pictures per second is 21 frames during video acquisition, capture effect wherein preferably a frame save as the image of BMP form.

Second step, the micro-particle classification number estimated value of client's input category

Consider the quantitative property of the metallographic structure grading of spheroidal-graphite cast iron, confirm that the micro-particle clusters number estimated value of client's input category is the K=2 class.

The 3rd step, the image pre-service

To pass through the image pre-service by the image that the first step is gathered; Its operating process is earlier image to be carried out gray processing to handle; Then the triple channel image transitions is become the single channel image; Select to carry out filtering operation according to the user again and make image sharpening, the filtering method that adopts two kinds of filtering of gaussian filtering and medium filtering to combine here.

The 4th step, the image clustering classification

With improved K means clustering algorithm and image segmentation algorithm fusion application, promptly adopt the dividing method of many threshold values iteration and the K means clustering algorithm is incorporated wherein, the pretreated image of the 3rd step image is carried out Cluster Classification, concrete steps are:

4.1 for each cluster is confirmed an initial cluster centre, there are 2 cluster centres in such 2 clusters,

4.2 each sample in the sample set is assigned to some in 2 clusters according to minimal distance principle,

4.3 the average of using all samples in each cluster is as new cluster centre,

If repeated for 4.2 and 4.3 steps till cluster centre no longer changes 4.4 cluster centre changes,

4.5 2 cluster centres that obtain at last are exactly clustering result,

4.6 find out the brightness in the image, its classification be labeled as 2 types through alternative manner.

In the 5th step, the morphological feature of the micro-particle in extraction and the computed image is distinguished the kind of micro-particle in the spheroidal-graphite cast iron and is confirmed rank in conjunction with corresponding GB

To extract the morphological feature of micro-particle by the metallic phase image of the good spheroidal-graphite cast iron of the 4th step image clustering classification; The nodularization rate, area occupation ratio, particle size and the pearlite quantity that comprise micro-particle; Through calculating the morphological feature of these micro-particles; Distinguish the kind of micro-particle and confirm rank in conjunction with corresponding national standards, concrete steps are following:

5.1 the differentiation of the micro-particle kind of graphite in the spheroidal-graphite cast iron

Pass through to calculate the nodularization rate and the area occupation ratio of the micro-particle of each graphite earlier,

The nodularization rate is calculated: $v=\frac{1\×48+0.8\×27+0.6\×30+0.3\×25+0\×13}{48+27+30+25+13}=0.67---\left(I\right)$

In the formula, the micro-particle of a graphite number of five kinds of spherical correction factors of expression is respectively n _1.0=48, n _0.8=27, n _0.6=30, n _0.3=25, n ₀=13.

Area occupation ratio calculates: $u=\frac{S_i}{{\mathrm{\π\; r}}^2}---\left(\mathrm{II}\right)$

In the formula, S _iBe the real area of the micro-particle of each graphite, unit is mm ², r is the circumradius of the micro-particle of each graphite, unit is mm.

From the micro-particle of the graphite of spheroidal-graphite cast iron sample 1, choose 10 micro-particles of graphite according to above-mentioned area occupation ratio computing formula (II) and carry out area occupation ratio calculating, the result is following:

S ₁=0.6731, r ₁=0.8044 o'clock, u ₁=0.3311;

S ₂=2.4768, r ₂=1.8640 o'clock, u ₂=0.2269;

S ₃=2.7857, r ₃=2.0049 o'clock, u ₃=0.2206;

S ₄=1.1189, r ₄=0.8016 o'clock, u ₄=0.5543;

S ₅=3.2694, r ₅=2.3391 o'clock, u ₅=0.1902;

S ₆=2.6196, r ₆=1.0688 o'clock, u ₆=0.7300;

S ₇=2.5642, r ₇=0.9542 o'clock, u ₇=0.8964;

S ₈=1.0170, r ₈=0.9605 o'clock, u ₈=0.3509;

S ₉=0.9587, r ₉=0.7017 o'clock, u ₉=0.6197;

S ₁₀=1.8474, r ₁₀=0.8157 o'clock, u ₁₀=0.8837.

(I) the micro-particle of a graphite number of five kinds of spherical correction factors in the formula and (II) the formula corresponding relation that calculates the area occupation ratio magnitude range of gained is respectively: n _1.0Corresponding u>=0.81, n _0.8Corresponding u=0.80～0.61, n _0.6Corresponding u=0.60～0.41, n _0.3Corresponding u=0.40～0.21 and n ₀Corresponding u≤0.20.

Distinguish the kind of the micro-particle of graphite in the spheroidal-graphite cast iron nodularization again with area occupation ratio, by area occupation ratio numerical value that calculates and GB 9441-88 contrast, in the spheroidal-graphite cast iron sample 1, n ₀The corresponding u of the micro-particle of=13 coccoliths China ink ₅=0.1902 is flake graphite, n _0.3The corresponding u of the micro-particle of=25 coccoliths China ink ₁=0.3311, u ₂=0.2269, u ₃=0.2206 and u ₈=0.3509 is quasiflake graphite, n _0.6The corresponding u of the micro-particle of=30 coccoliths China ink ₄=0.5543 is temper carbon, n _0.8The corresponding u of the micro-particle of=27 coccoliths China ink ₉=0.6197 and u ₆=0.7300 is nodular graphite, n _1.0The corresponding u of the micro-particle of=48 coccoliths China ink ₇=0.8964 and u ₁₀=0.8837 is globular graphite.

5.2 the size of the micro-particle of graphite grading in the spheroidal-graphite cast iron

The computing formula of calculating the micro-average particle diameter of each graphite does

$\stackrel{\‾}{D}=2\×\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{S}_i}{\mathrm{\π}(n_0+n_1+n_2+n_3+n_4+n_5)}}---\left(\mathrm{III}\right)$

S in the formula _iBe the area of the micro-particle of each graphite, unit is mm ², n ₀, n ₁, n ₂, n ₃, n ₄And n ₅The scope of the micro-particle diameter D of corresponding respectively graphite is: greater than 25 to 50, greater than 12 to 25, greater than 6 to 12, greater than 4 to 6, greater than 1.5 to 4 and≤1.5 the micro-particle number of graphite, n here ₀=0, n ₁=0, n ₂=0, n ₃=0, n ₄=12, n ₅=131, n=0+0+0+0+12+131=143 calculates

Mean diameter

numerical value and GB GB 9441-88 contrast by the micro-particle size of graphite in the spheroidal-graphite cast iron that calculates; Judge the affiliated rank of the micro-particle size of different graphite; Here stipulate according to GB GB 9441-88; The scope of the micro-average particle diameter of graphite

is: the rank greater than 25 to 50 is 3 grades; Rank greater than 12 to 25 is 4 grades; Rank greater than 6 to 12 is 5 grades; Rank greater than 3 to 6 is 6 grades, and the rank greater than 1.5 to 3 is 7 grades, and≤1.5 rank is 8 grades.The micro-particle size rank of judging thus in the spheroidal-graphite cast iron sample 1 of graphite is 8 grades.

5.3 spheroidal-graphite cast iron medium pearlite quantity grading

Calculate pearlitic quantity number percent m, computing formula is:

$m=\frac{\underset{i}{\mathrm{\Σ}}{S}_i}{S_{\mathrm{all}}}\×100\%=98.55\%,---\left(\mathrm{IV}\right)$

S wherein _iBe the area of the micro-particle of each pearlite, S _AllBe the size of image, unit is mm ²

With pearlite quantity number percent that is calculated and GB GB 9441-88 contrast, judge pearlite quantity rank, GB GB 9441-88 regulation; Pearlite quantity percentage range more than or equal to 90 be pearl 95, more than or equal to 80 to less than 90 be pearl 85, more than or equal to 70 to less than 80 be pearl 75; More than or equal to 60 to less than 70 be pearl 65, more than or equal to 50 to less than 60 be pearl 55, more than or equal to 40 to less than 50 be pearl 45; More than or equal to 30 to less than 40 be pearl 35, ≈'s 25 is pearl 25, ≈'s 15 is pearl 15; ≈'s 10 is pearl 10, and ≈'s 5 is pearl 5, judges that thus the pearlite name in the spheroidal-graphite cast iron sample 1 is called pearl 95.

Fig. 4 shows the recognition effect of present embodiment to the micro-particle of graphite in the spheroidal-graphite cast iron sample 1; Wherein exhibit red is a flake graphite 1, shows that yellow is a quasiflake graphite 2, shows that blueness is a temper carbon 3; Show that green is a nodular graphite 4, show that purple is a globular graphite 5.

Embodiment 2

The automatic classification grading of micro-particle in the spheroidal-graphite cast iron sample 2

The first step, images acquired

Gather the metallic phase image of tested spheroidal-graphite cast iron sample 2 through optical microscope and CCD camera; The resolution of used CCD camera is in 3,000,000 pixels; Employing USB interface communication, number of pictures per second is 31 frames during video acquisition, capture effect wherein preferably a frame save as the image of JPG form.

Second step, the micro-particle classification number estimated value of client's input category

Consider the quantitative property of the metallographic structure grading of spheroidal-graphite cast iron, confirm that the micro-particle clusters number estimated value of client's input category is the K=3 class.

The 3rd step, the image pre-service

To pass through the image preliminary treatment by the image that the first step is gathered; Its operating process is earlier image to be carried out gray processing to handle; Then the triple channel image transitions is become the single channel image; Select to carry out filtering operation based on the user again and make image sharpening, the filtering method that adopts two kinds of filtering of gaussian filtering and medium filtering to combine here;

The 4th step, the image clustering classification

With improved K means clustering algorithm and image segmentation algorithm fusion application, promptly adopt the dividing method of many threshold values iteration and the K means clustering algorithm is incorporated wherein, to the pretreated image clustering classification of the 3rd step image, concrete steps are:

4.1 for each cluster is confirmed an initial cluster centre, there are 3 cluster centres in such 3 clusters,

4.2 each sample in the sample set is assigned to some in 3 clusters according to minimal distance principle,

4.3 the average of using all samples in each cluster is as new cluster centre,

If repeated for 4.2 and 4.3 steps till cluster centre no longer changes 4.4 cluster centre changes,

4.5 3 cluster centres that obtain at last are exactly clustering result,

4.6 find out the brightness in the image, through alternative manner its classification is labeled as 3 types,

In the 5th step, the morphological feature of the micro-particle in extraction and the computed image is distinguished the kind of micro-particle in the spheroidal-graphite cast iron and is confirmed rank in conjunction with corresponding GB

To extract the morphological feature of micro-particle by the metallic phase image of the good spheroidal-graphite cast iron of the 4th step image clustering classification; The nodularization rate, area occupation ratio, particle size and the pearlite quantity that comprise micro-particle; Through calculating the morphological feature of these micro-particles; Distinguish the kind of micro-particle and confirm rank in conjunction with corresponding national standards, concrete steps are following:

5.1 the differentiation of the micro-particle kind of graphite in the spheroidal-graphite cast iron

Pass through to calculate the nodularization rate and the area occupation ratio of the micro-particle of each graphite earlier,

The nodularization rate is calculated: $v=\frac{1\×3+0.8\×3+0.6\×1+0.3\×5+0\×1}{3+3+1+5+1}=0.58---\left(I\right)$

In the formula, the micro-particle of a graphite number of five kinds of spherical correction factors of expression is respectively n _1.0=3, n _0.8=3, n _0.6=1, n _0.3=5, n ₀=1.

Area occupation ratio calculates: $u=\frac{S_i}{{\mathrm{\π\; r}}^2}---\left(\mathrm{II}\right)$

In the formula, S _iBe the real area of the micro-particle of each graphite, unit is mm ², r is the circumradius of the micro-particle of each graphite, unit is mm.

From the micro-particle of the graphite of spheroidal-graphite cast iron sample 2, choose 10 micro-particles of graphite according to above-mentioned area occupation ratio computing formula (II) and carry out area occupation ratio calculating, the result is following:

S ₁=0.6032, r ₁=1.1069 o'clock, u ₁=0.1567;

S ₂=4.7730, r ₂=1.7364 o'clock, u ₂=0.5039;

S ₃=0.0612, r ₃=0.1620 o'clock, u ₃=0.7427;

S ₄=8.7709, r ₄=3.5116 o'clock, u ₄=0.2264;

S ₅=1.9756, r ₅=1.4886 o'clock, u ₅=0.2838;

S ₆=17.3494, r ₆=2.5956 o'clock, u ₆=0.8197;

S ₇=19.6368, r ₇=2.9391 o'clock, u ₇=0.7236;

S ₈=27.7054, r ₈=2.9773 o'clock, u ₈=0.9949;

S ₉=2.5555, r ₉=1.6794 o'clock, u ₉=0.2884;

S ₁₀=155.3228, r ₁₀=13.7633 o'clock, u ₁₀=0.2610;

(I) the micro-particle of a graphite number of five kinds of spherical correction factors in the formula and (II) the formula corresponding relation that calculates the area occupation ratio scope of gained is respectively: n _1.0Corresponding u>=0.81, n _0.8Corresponding u=0.80～0.61, n _0.6Corresponding u=0.60～0.41, n _0.3Corresponding u=0.40～0.21, n ₀Corresponding u≤0.20.

Distinguish the kind of the micro-particle of graphite in the spheroidal-graphite cast iron nodularization again with area occupation ratio, by area occupation ratio numerical value that calculates and GB 9441-88 contrast, in the spheroidal-graphite cast iron sample 1, n ₀The corresponding u of the micro-particle of=1 coccolith China ink ₁=0.1567 is flake graphite, n _0.3The corresponding u of the micro-particle of=4 coccoliths China ink ₄=0.2264, u ₅=0.2838, u ₉=0.2884, u ₁₀=0.2610 is quasiflake graphite, n _0.6The corresponding u of the micro-particle of=1 coccolith China ink ₂=0.5039 is temper carbon, n _0.8The corresponding u of the micro-particle of=2 coccoliths China ink ₃=0.7427, u ₇=0.7236 is nodular graphite, n _1.0The corresponding u of the micro-particle of=2 coccoliths China ink ₆=0.8197, u ₈=0.9949 is globular graphite.

5.2 to the micro-particle size grading of spheroidal-graphite cast iron graphite

The computing formula of calculating the micro-average particle diameter of each graphite

does

$\stackrel{\‾}{D}=2\×\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{S}_i}{\mathrm{\π}(n_0+n_1+n_2+n_3+n_4+n_5)}}---\left(\mathrm{III}\right)$

S in the formula _iBe the area of the micro-particle of each graphite, unit is mm ², n ₀, n ₁, n ₂, n ₃, n ₄And n ₅The scope of the micro-particle diameter D of corresponding respectively graphite is: greater than 25 to 50, greater than 12 to 25, greater than 6 to 12, greater than 4 to 6, greater than 1.5 to 4 and≤1.5 the micro-particle number of graphite, n here ₀=0, n ₁=0, n ₂=7, n ₃=1, n ₄=8, n ₅=6, n=0+0+7+1+8+6=22 calculates

Mean diameter

numerical value and GB GB 9441-88 contrast by the micro-particle size of graphite in the spheroidal-graphite cast iron that calculates; Judge the affiliated rank of the micro-particle size of different graphite; Here stipulate according to GB GB 9441-88; The scope of the micro-average particle diameter of graphite is: the rank greater than 25 to 50 is 3 grades; Rank greater than 12 to 25 is 4 grades; Rank greater than 6 to 12 is 5 grades; Rank greater than 3 to 6 is 6 grades, and the rank greater than 1.5 to 3 is 7 grades, and≤1.5 rank is 8 grades.The micro-particle size rank of judging thus in the spheroidal-graphite cast iron sample 2 of graphite is 6 grades.

5.3 spheroidal-graphite cast iron medium pearlite quantity grading

Calculate pearlitic quantity number percent M, computing formula is:

$M=\frac{\underset{i}{\mathrm{\Σ}}{S}_i}{S_{\mathrm{all}}}\×100\%=35.42\%,---\left(\mathrm{IV}\right)$

S wherein _iBe the area of the micro-particle of each pearlite, S _AllBe the size of image, unit is mm ²

With pearlite quantity number percent that is calculated and GB GB 9441-88 contrast, judge pearlite quantity rank, GB GB 9441-88 regulation; Pearlite quantity percentage range more than or equal to 90 be pearl 95, more than or equal to 80 to less than 90 be pearl 85, more than or equal to 70 to less than 80 be pearl 75; More than or equal to 60 to less than 70 be pearl 65, more than or equal to 50 to less than 60 be pearl 55, more than or equal to 40 to less than 50 be pearl 45; More than or equal to 30 to less than 40 be pearl 35, ≈'s 25 is pearl 25, ≈'s 15 is pearl 15; ≈'s 10 is pearl 10, and ≈'s 5 is pearl 5, judges that thus the pearlite name in the spheroidal-graphite cast iron sample 2 is called pearl 35.

Fig. 5 is the classifying quality figure of spheroidal-graphite cast iron sample 2 in the present embodiment.Wherein the picture left above is the preceding initial picture of identification, and top right plot is the micro-particle figure of graphite that is partitioned into, and left figure below is the micro-particle figure of pearlite that is partitioned into, and bottom-right graph is the micro-particle figure of ferrite that is partitioned into.This proof the inventive method can be effectively carried out discriminator to the micro-particle of the different shape characteristic that contains in the spheroidal-graphite cast iron.

Embodiment 3

The automatic classification grading of micro-particle in the spheroidal-graphite cast iron sample 3

The first step, images acquired

Gather the metallic phase image of tested spheroidal-graphite cast iron sample 3 through optical microscope and CCD camera; The resolution of used CCD camera is in 1,300,000 pixels; Employing USB interface communication, number of pictures per second is 11 frames during video acquisition, capture effect wherein preferably a frame save as the image of BMP form.

Second step, the micro-particle classification number estimated value of client's input category

Consider the quantitative property of the metallographic structure grading of spheroidal-graphite cast iron, confirm that the micro-particle clusters number estimated value of client's input category is the K=3 class.

The 3rd step, the image pre-service

To pass through the image pre-service by the image that the first step is gathered; Its operating process is earlier image to be carried out gray processing to handle; Then the triple channel image transitions is become the single channel image; Select to carry out filtering operation according to the user again and make image sharpening, the filtering method that adopts two kinds of filtering of gaussian filtering and medium filtering to combine here.

The 4th step, the image clustering classification

With improved K means clustering algorithm and image segmentation algorithm fusion application, promptly adopt the dividing method of many threshold values iteration and the K means clustering algorithm is incorporated wherein, the pretreated image of the 3rd step image is carried out Cluster Classification, concrete steps are:

4.1 for each cluster is confirmed an initial cluster centre, there are 3 cluster centres in such 3 clusters,

4.2 each sample in the sample set is assigned to some in 3 clusters according to minimal distance principle,

4.3 the average of using all samples in each cluster is as new cluster centre,

If repeated for 4.2 and 4.3 steps till cluster centre no longer changes 4.4 cluster centre changes,

4.5 3 cluster centres that obtain at last are exactly clustering result,

4.6 find out the brightness in the image, through alternative manner its classification is labeled as 3 types,

In the 5th step, the morphological feature of the micro-particle in extraction and the computed image is distinguished the kind of micro-particle in the spheroidal-graphite cast iron and is confirmed rank in conjunction with corresponding GB

To extract the morphological feature of micro-particle by the metallic phase image of the good spheroidal-graphite cast iron of the 4th step image clustering classification; The nodularization rate, area occupation ratio, particle size and the pearlite quantity that comprise micro-particle; Through calculating the morphological feature of these micro-particles; Distinguish the kind of micro-particle and confirm rank in conjunction with corresponding national standards, concrete steps are following:

5.1 the differentiation of the micro-particle kind of graphite in the spheroidal-graphite cast iron

Pass through to calculate the nodularization rate and the area occupation ratio of the micro-particle of each graphite earlier,

The nodularization rate is calculated: $v=\frac{1\×11+0.8\×6+0.6\×1+0.3\×1+0\×0}{11+6+1+1+0}=0.88---\left(I\right)$

In the formula, the micro-particle of a graphite number of five kinds of spherical correction factors of expression is respectively n _1.0=11, n _0.8=6, n _0.6=1, n _0.3=1, n ₀=0.

Area occupation ratio calculates: $u=\frac{S_i}{{\mathrm{\π\; r}}^2}---\left(\mathrm{II}\right)$

In the formula, S _iBe the real area of the micro-particle of each graphite, unit is mm ², r is the circumradius of the micro-particle of each graphite, unit is mm,

From the micro-particle of the graphite of spheroidal-graphite cast iron sample 3, choose 10 micro-particles of graphite according to above-mentioned area occupation ratio computing formula (II) and carry out area occupation ratio calculating, the result is following:

S ₁=0.1544, r ₁=0.4580 o'clock, u ₁=0.2343;

S ₂=0.5886, r ₂=0.4580 o'clock, u ₂=0.8930;

S ₃=0.2739, r ₃=0.3621 o'clock, u ₃=0.6649;

S ₄=0.7518, r ₄=0.4962 o'clock, u ₄=0.9719;

S ₅=1.4803, r ₅=0.6960 o'clock, u ₅=0.9728;

S ₆=0.7809, r ₆=0.5938 o'clock, u ₆=0.7050;

S ₇=1.8445, r ₇=0.7764 o'clock, u ₇=0.9739;

S ₈=0.7955, r ₈=0.5914 o'clock, u ₈=0.7240;

S ₉=1.4132, r ₉=0.7287 o'clock, u ₉=0.8471;

S ₁₀=1.2909, r ₁₀=0.7253 o'clock, u ₁₀=0.7812;

(I) the micro-particle of a graphite number of five kinds of spherical correction factors in the formula and (II) the formula corresponding relation that calculates the area occupation ratio scope of gained is respectively: n _1.0Corresponding u>=0.81, n _0.8Corresponding u=0.80～0.61, n _0.6Corresponding u=0.60～0.41, n _0.3Corresponding u=0.40～0.21, n ₀Corresponding u≤0.20.

Distinguish the kind of the micro-particle of graphite in the spheroidal-graphite cast iron nodularization again with area occupation ratio, by area occupation ratio numerical value that calculates and GB 9441-88 contrast, in the spheroidal-graphite cast iron sample 1, n ₀The micro-particle of=0 coccolith China ink is a flake graphite, n _0.3The corresponding u of the micro-particle of=1 coccolith China ink ₁=0.2343 is quasiflake graphite, n _0.6The micro-particle of=0 coccolith China ink is a temper carbon, n _0.8The corresponding u of the micro-particle of=4 coccoliths China ink ₃=0.6649, u ₆=0.7050, u ₈=0.7240, u ₁₀=0.7812 is nodular graphite, n _1.0The corresponding u of the micro-particle of=5 coccoliths China ink ₂=0.8930, u ₄=0.9719, u ₅=0.9728, u ₇=0.9739, u ₁₀=0.7812 is globular graphite.

5.2 the size of the micro-particle of graphite grading in the spheroidal-graphite cast iron

The computing formula of calculating the micro-average particle diameter of each graphite does

$\stackrel{\‾}{D}=2\×\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{S}_i}{\mathrm{\π}(n_0+n_1+n_2+n_3+n_4+n_5)}}---\left(\mathrm{III}\right)$

S in the formula _iBe the area of the micro-particle of each graphite, unit is mm ², n ₀, n ₁, n ₂, n ₃, n ₄And n ₅The scope of the micro-particle diameter D of corresponding respectively graphite is: greater than 25 to 50, greater than 12 to 25, greater than 6 to 12, greater than 4 to 6, greater than 1.5 to 4 and≤1.5 the micro-particle number of graphite, n here ₀=0, n ₁=1, n ₂=0, n ₃=0, n ₄=1, n ₅=17, n=0+1+0+0+1+17=19 calculates

Mean diameter

numerical value and GB GB 9441-88 contrast by the micro-particle size of graphite in the spheroidal-graphite cast iron that calculates; Judge the affiliated rank of the micro-particle size of different graphite; Here stipulate according to GB GB 9441-88; The scope of the micro-average particle diameter of graphite

is: the rank greater than 25 to 50 is 3 grades; Rank greater than 12 to 25 is 4 grades; Rank greater than 6 to 12 is 5 grades; Rank greater than 3 to 6 is 6 grades, and the rank greater than 1.5 to 3 is 7 grades, and≤1.5 rank is 8 grades.The micro-particle size rank of judging thus in the spheroidal-graphite cast iron sample 3 of graphite is 6 grades.

5.3 spheroidal-graphite cast iron medium pearlite quantity grading

Calculate pearlitic quantity number percent M, computing formula is:

$M=\frac{\underset{i}{\mathrm{\Σ}}{S}_i}{S_{\mathrm{all}}}\×100\%=25.38\%,---\left(\mathrm{IV}\right)$

S wherein _iBe the area of the micro-particle of each pearlite, S _AllBe the size of image, unit is mm ²

With pearlite quantity number percent that is calculated and GB GB 9441-88 contrast, judge pearlite quantity rank, GB GB 9441-88 regulation; Pearlite quantity percentage range more than or equal to 90 be pearl 95, more than or equal to 80 to less than 90 be pearl 85, more than or equal to 70 to less than 80 be pearl 75; More than or equal to 60 to less than 70 be pearl 65, more than or equal to 50 to less than 60 be pearl 55, more than or equal to 40 to less than 50 be pearl 45; More than or equal to 30 to less than 40 be pearl 35, ≈'s 25 is pearl 25, ≈'s 15 is pearl 15; ≈'s 10 is pearl 10, and ≈'s 5 is pearl 5, judges that thus the pearlite name in the spheroidal-graphite cast iron sample 3 is called pearl 25.

Fig. 6 is the classifying quality figure of spheroidal-graphite cast iron sample 3 in the present embodiment.The picture left above is the preceding initial graph of identification, and top right plot is the micro-particle figure of cutting apart of ferrite, and left figure below is the micro-particle figure of cutting apart of graphite, and bottom-right graph is the micro-particle figure of cutting apart of pearlite.This proof the inventive method can be effectively carried out discriminator to the micro-particle of the different shape characteristic that contains in the spheroidal-graphite cast iron.

The foregoing description 1, embodiment 2 and embodiment 3 are example with spheroidal-graphite cast iron; Graphite, ring-type ferrite and matrix pearlite during sorting technique shown in it can be formed it accurately are separately; After the classification, the client can rule of thumb judge the sort of component materials that needs in the Measurement and analysis spheroidal-graphite cast iron.

Claims (3)

1. the automatic classification ranking method of micro-particle in the spheroidal-graphite cast iron is characterized in that concrete steps are following:

The first step, images acquired

Gather the metallic phase image of spheroidal-graphite cast iron through optical microscope and CCD camera; The resolution of used CCD camera is in 1,300,000～3,000,000 pixels; Employing USB interface communication, number of pictures per second is 11～31 frames during video acquisition, capture effect wherein preferably a frame save as the image of BMP or JPG form;

Second step, the micro-particle classification number estimated value of client's input category

Consider the quantitative property of the metallographic structure grading of spheroidal-graphite cast iron, the micro-particle clusters number estimated value of confirming client's input category is k=2～5 type;

The 3rd step, the image pre-service

To pass through the image preliminary treatment by the image that the first step is gathered; Its operating process is earlier image to be carried out gray processing to handle; Then the triple channel image transitions is become the single channel image; Select to carry out filtering operation based on the user again and make image sharpening, the filtering method that adopts two kinds of filtering of gaussian filtering and medium filtering to combine here;

The 4th step, the image clustering classification

With improved K means clustering algorithm and image segmentation algorithm fusion application, promptly adopt the dividing method of many threshold values iteration and the K means clustering algorithm is incorporated wherein, to the pretreated image clustering classification of the 3rd step image, concrete steps are:

4.1 for each cluster is confirmed an initial cluster centre, there be k cluster centre in k cluster like this,

4.2 each sample in the sample set is assigned to some in k the cluster according to minimal distance principle,

4.3 the average of using all samples in each cluster is as new cluster centre,

If repeated for 4.2 and 4.3 steps till cluster centre no longer changes 4.4 cluster centre changes,

4.5 the k that obtains an at last cluster centre is exactly a clustering result,

4.6 find out the brightness in the image, through alternative manner its classification be labeled as 1,2,3 ... The K class;

In the 5th step, the morphological feature of the micro-particle in extraction and the computed image is distinguished the kind of micro-particle in the spheroidal-graphite cast iron and is confirmed rank in conjunction with corresponding GB

To extract the morphological feature of micro-particle by the metallic phase image of the good spheroidal-graphite cast iron of the 4th step image clustering classification; The nodularization rate, area occupation ratio, particle size and the pearlite quantity that comprise micro-particle; Through calculating the morphological feature of these micro-particles; Distinguish the kind of micro-particle in the spheroidal-graphite cast iron and confirm rank in conjunction with corresponding national standards, concrete steps are following:

5.1 the differentiation of the micro-particle kind of graphite in the spheroidal-graphite cast iron

Pass through to calculate the nodularization rate and the area occupation ratio of the micro-particle of each graphite earlier,

The nodularization rate is calculated: $v=\frac{1\×n_{1.0}+0.8\×n_{0.8}+0.6\×n_{0.6}+0.3\×n_{0.3}+0\×n_0}{n_{1.0}+n_{0.8}+n_{0.6}+n_{0.3}+n_0}---\left(I\right)$

In the formula, n _1.0, n _0.8, n _0.6, n _0.3, n ₀The micro-particle of a graphite number of representing five kinds of spherical correction factors respectively,

Area occupation ratio calculates: $u=\frac{S_i}{{\mathrm{\π\; r}}^2}---\left(\mathrm{II}\right)$

In the following formula, S _iBe the real area of the micro-particle of each graphite, unit is mm ², r is the circumradius of the micro-particle of each graphite, unit is mm,

(I) the micro-particle of a graphite number of five kinds of spherical correction factors in the formula and (II) the formula corresponding relation that calculates the area occupation ratio magnitude range of gained is respectively: n _1.0Corresponding u>=0.81, n _0.8Corresponding u=0.80～0.61, n _0.6Corresponding u=0.60～0.41, n _0.3Corresponding u=0.40～0.21 and n ₀Corresponding u≤0.20,

Distinguish the kind of the micro-particle of graphite in the spheroidal-graphite cast iron nodularization again with area occupation ratio, contrast with GB9441-88 by the area occupation ratio numerical value that calculates, the micro-particle area occupation ratio of graphite less than 0.2 be flake graphite; Greater than 0.2 to less than 0.4 be quasiflake graphite; Greater than 0.4 to less than 0.6 be temper carbon; Greater than 0.6 to less than 0.8 be nodular graphite; Greater than 0.8 to less than 1.0 be globular graphite;

5.2 the size of the micro-particle of graphite grading in the spheroidal-graphite cast iron

The computing formula of calculating the micro-average particle diameter of each graphite

does

$\stackrel{\‾}{D}=2\×\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{S}_i}{\mathrm{\π}(n_0+n_1+n_2+n_3+n_4+n_5)}}---\left(\mathrm{III}\right)$

In the following formula, S _iBe the real area of the micro-particle of each graphite, unit is mm ², n ₀, n ₁, n ₂, n ₃, n ₄And n ₅The scope of the micro-particle diameter D of corresponding respectively graphite is: greater than 25 to 50, greater than 12 to 25, greater than 6 to 12, greater than 4 to 6, greater than 1.5 to 4 and≤1.5 the micro-particle number of graphite,

Mean diameter

numerical value and GB GB 9441-88 contrast by the micro-particle size of graphite in the spheroidal-graphite cast iron that calculates; Judge the affiliated rank of the micro-particle size of different graphite; Here stipulate according to GB GB 9441-88; The scope of the micro-average particle diameter of graphite

is: the rank greater than 25 to 50 is 3 grades; Rank greater than 12 to 25 is 4 grades; Rank greater than 6 to 12 is 5 grades; Rank greater than 3 to 6 is 6 grades; Rank greater than 1.5 to 3 is 7 grades, and≤1.5 rank is 8 grades;

5.3 spheroidal-graphite cast iron medium pearlite quantity grading

Calculate pearlitic quantity number percent M, computing formula is:

$M=\frac{\underset{i}{\mathrm{\Σ}}{S}_i}{S_{\mathrm{all}}}\×100\%---\left(\mathrm{IV}\right)$

S in the following formula _iBe the real area of the micro-particle of each pearlite, S _AllBe the size of entire image, unit is mm ²,

With pearlite quantity number percent that is calculated and GB GB 9441-88 contrast, judge pearlite quantity rank, GB GB 9441-88 regulation; Pearlite quantity percentage range more than or equal to 90 be pearl 95, more than or equal to 80 to less than 90 be pearl 85, more than or equal to 70 to less than 80 be pearl 75; More than or equal to 60 to less than 70 be pearl 65, more than or equal to 50 to less than 60 be pearl 55, more than or equal to 40 to less than 50 be pearl 45; More than or equal to 30 to less than 40 be pearl 35, ≈'s 25 is pearl 25, ≈'s 15 is pearl 15; ≈'s 10 is pearl 10, and ≈'s 5 is pearl 5.

2. the automatic classification ranking method of micro-particle in the said spheroidal-graphite cast iron of claim 1; It is characterized in that: the flow process of said image segmentation algorithm is: begin → confirm cluster centre → distribution sample; Is sample average that new cluster centre → center changes? → be; Return the distribution sample, sample average is new cluster centre; Not, find out the brightness of image characteristic, iteration is sorted out.

3. the automatic classification ranking method of micro-particle in the said spheroidal-graphite cast iron of claim 1; It is characterized in that: wherein the enforcement of five steps is to be controlled according to following techniqueflow by computing machine: beginning → images acquired → image pre-service → confirm cluster centre → distribution sample; Is sample average that new cluster centre → center changes? → be; Return the distribution sample, sample average is new cluster centre; Not, find out the brightness of image characteristic, iteration is sorted out.

Images