CN108446706A - A kind of abrasive grain material automatic identifying method based on color principal Component Extraction - Google Patents
A kind of abrasive grain material automatic identifying method based on color principal Component Extraction Download PDFInfo
- Publication number
- CN108446706A CN108446706A CN201810162476.3A CN201810162476A CN108446706A CN 108446706 A CN108446706 A CN 108446706A CN 201810162476 A CN201810162476 A CN 201810162476A CN 108446706 A CN108446706 A CN 108446706A
- Authority
- CN
- China
- Prior art keywords
- abrasive grain
- color
- image
- value
- identified
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Investigating And Analyzing Materials By Characteristic Methods (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
A kind of abrasive grain material automatic identifying method based on color principal Component Extraction, which is characterized in that step 1:According to the temper color under the color of various materials itself and its temperature at different levels, material standard color sample database is built;Step 2:The abrasive grain area of the Debris Image after the image after RGB image gray processing, S components Debris Image, the three width segmentation corresponding to V component Debris Image is found out using zone marker interconnection algorithm, retain maximum abrasive grain in respective image, three width images of comparison select the wherein maximum Debris Image of abrasive grain area;Step 3:Be partitioned into Debris Image domain color is extracted using K Mean clusters;Step 4:Extracted abrasive grain picture domain color is sought at a distance from standard color sample using the minimum distance classification based on Euclidean distance, realizes the automatic identification of abrasive grain material.Abrasive grain thermal analysis method is combined by the present invention with image processing techniques, based on the minimum distance classification of K Mean cluster and Euclidean distance, realizes the automatic identification of regular abrasive grit material.
Description
Technical field
The invention belongs to the more particularly to a kind of abrasive grain materials based on color principal Component Extraction in Machine Fault Diagnosis field certainly
Dynamic recognition methods.
Background technology
In a variety of causes of mechanical machine part failure, 80% or more mechanical equipment fault and failure are to be by abrasion
Caused by.And abrasive grain carries the important information of equipment attrition situation, usually by the plastic deformation of material, processing hardening, hair
What the complicated process such as thermal oxide and corrosion caused to generate, therefore the abrasive grain in fluid is monitored and analyzed can not only obtain
The degree of wear and tendency information for obtaining equipment can also determine the inducement that the abrasion mechanism of equipment and failure occur, finally
Information can be provided for improve equipment Operation Conditions and design improvement, to improve equipment reliability of operation and safety.Cause
This, Debris Analysis (WDA) technology has become equipment running status monitoring and fault diagnosis most efficient method.
Wherein offline iron spectrum Debris Analysis technology can accurately obtain abrasion machine since its analysis precision is high, accuracy is good
Reason and wear type have become the major technique foundation of mechanical system abrasion analysis to provide profound abrasion reason.
Each friction pair of the generated abrasive grain in equipment during equipment wearing, and different frictions is secondary that it is main
It is again inconsistent to constitute material, therefore generated abrasive grain material is also different in wear process.Based on this by studying abrasive grain
Material the source of abrasive grain can be determined by them, and then judge wear and tear in machines concrete position and abrasion zero
Part.Usual abrasive grain material is generally divided into Ferrious material, non-ferrous metal, metal oxide, lubricant product and pollution by type
Object etc., and most secondary main materials of friction are Ferrious material and non-ferrous metal in mechanical equipment, therefore for the knowledge of material
Ferrious material and its oxide and non-ferrous metal are not studied not mainly.
At present iron spectrogram as abrasive grain Material Identification field it is most widely used be thermal analysis method and scanning electron microscope power spectrum point
Analysis method, wherein thermal analysis method are that the highly effective means of abrasive grain alloying component are identified in Spectral Analysis Technology, small investment,
It is easy to operate, and the specific element composition and constituent content for pursuing abrasive grain fine and smoothly are not needed, only machine need to be understood in advance
The material situation of each parts, it is of interest that its material of the wear particle seen in iron spectral slice matches with which parts.
But the analyst that traditional abrasive grain Material Identification relies primarily on domain expert or has wide experience carries out multi-step temperature to abrasive grain
Heating carries out artificial cognition according to the abrasive grain characterization color under different temperatures to the material of abrasive grain, this requires operator to need to have
There is very high professional knowledge, it is difficult to promote and apply.Since this analysis result is often too dependent on the subjective judgement of analyst
And experience, it is likely to result in prodigious error, and manual analysis efficiency is low, this will lead to human resources to a certain extent
Waste.The Wang Wei China of Shanghai Communications University is special according to the heat transfer of abrasive grain in " abrasive grain notation and analytical technology system research "
Property, the abrasive grain of unlike material is heated to certain temperature, its cooling procedure is observed using infrared temperature-test technology, from particle temperature field
The temperature change of particle surface is observed in the conduction process changed over time to study abrasive grain material.This method is needed by infrared
Thermal imaging system, easily by the ambient temperature of testee, heating temperature selection, object emission rate and particle size etc. because
Element influences, and accuracy rate precision is difficult to ensure, and complicated for operation, and the degree of automation is relatively low.Therefore abrasive grain Material Identification how is improved
Sequencing, automation, to improve the efficiency of abrasive grain Material Identification, meet industrial practical application request then become abrasive grain from
Dynamicization identifies urgent problem to be solved.
Invention content
The purpose of the present invention is to provide a kind of abrasive grain material automatic identifying method based on color principal Component Extraction, with solution
The certainly above problem.
To achieve the above object, the present invention uses following technical scheme:
A kind of abrasive grain material automatic identifying method based on color principal Component Extraction, includes the following steps:
Step 1:According to the temper color under the color of various materials itself and its temperature at different levels, material mark is built
Quasi- color card library;
Step 2:RGB image gray processing and S, V component gray processing carried out to Debris Image to be identified, and to gray processing after
Image carry out adaptive threshold fuzziness and Morphological scale-space, find out RGB image gray processing using zone marker interconnection algorithm
The abrasive grain area of Debris Image after rear image, S components Debris Image, the three width segmentation corresponding to V component Debris Image, is protected
Maximum abrasive grain in respective image, three width images of comparison is stayed to select the wherein maximum Debris Image of abrasive grain area;
Step 3:Be partitioned into Debris Image domain color is extracted using K-Mean clusters;
Step 4:Extracted abrasive grain picture domain color and mark are sought using the minimum distance classification based on Euclidean distance
The distance of quasi- color card realizes the automatic identification of abrasive grain material.
Further, material standard color sample database is built in step 1 to include the following steps:
S1:It determines Ferrious material material color card, chooses corresponding under 330 DEG C and 540 DEG C of two temperatures return
Fiery color is as various Ferrious material material standard colors;
S2 determines non-ferrous metal material color card;
S3 determines metal oxide material color card, includes mainly red oxide and black oxide.
3, a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 1,
It is characterized in that, Debris Image is subjected to RGB image gray processing, S components and V component gray processing in step 2, respectively such as formula 1-3
It is shown:
Formula 1:F (x, y)=0.299R (x, y)+0.587G (x, y)+0.114B (x, y)
Formula 2:
Formula 3:
R, G, B of pixel divide at coordinate (x, y) in R (x, y), G (x, y), B (x, y) representative image in formula (1)-(3)
Magnitude;
Further, in step 2 to three width abrasive grain gray level images into row threshold division, be as follows:
1) note T is the segmentation threshold of Debris Image foreground and background, and value range is 0~255, take 0 successively to T~
255 values are calculated;
2) foreground and background is divided the image into according to present threshold value T, it is w to find out foreground points respectively and account for image scaled0,
Average gray is u0;It is w that background points, which account for image scaled,1, average gray u1, the overall average gray scale of image is u;
3) variance of foreground and background image is calculated according to formula 4:
Formula 4:δ2(T)=w0×(u0-u)2+w1×(u1-u)2
4) variance δ corresponding when T takes 0~255 is found out respectively2(T), 256 variance yields are compared, when selecting variance minimum
The threshold value that corresponding T values are divided as Debris Image;
5) gray level image is subjected to binaryzation operation according to formula 5 again, it is black to obtain abrasive grain, and background is white
Mask schemes:
Formula 5:
F (x, y) indicates the gray-scale map of Debris Image in formula 5.
Further, carrying out Morphological scale-space to the Debris Image after segmentation in step 2 is:Reverse process is carried out, fortune is closed
Calculation handles and holes filling operation;Utilizing " eight connected region algorithm " to find out, abrasive grain area is maximum to be retained as research observation
Object, while three width Debris Images finally acquired abrasive grain is compared, it chooses and divides maximum Debris Image as target abrasive grain.
Further, the specific steps of the be partitioned into Debris Image domain color of extraction include in step 3:
S1 defines an iterations upper limit M;
S2 looks for N number of point in figure at random, takes out its rgb value as seed point;
Then S3 finds a rgb value and its most similar seed point to each of figure point, and this point is added to
Rgb value is most close to be obtained in the point group of seed point place;
S4 calculates the average RGB value of seed point group, and using this rgb value as new seed point;Average RGB value is tired
Totalling value/number.
Whether whether equal with old value S5 compares this new value;If equal, seed point convergence is completed, into the 6th
Step;If differed, the 3rd step is continued to execute, until iterations reach M times;
S6, when seed point convergence completion or iterations reach M times, we are a row to the weight of all seed points
Sequence;Weight wherein includes the number of point;
S7 takes out the value of the highest seed point of weight, the dominant hue of abrasive grain picture of this value i.e. needed for us.
Further, it identifies and is as follows in step 4:
S1 seeks extracted abrasive grain picture domain color and the Ferrious material material in material color value according to formula 6
Black oxidation in red oxide material color or metal oxide in matter color, non-ferrous metal material color, metal oxide
The Euclidean distance of object material color;
Formula 6:
Formula midpoint (R0, G0, B0) it is the abrasive grain picture domain color pixel value extracted, point (R1, G1, B1) it is abrasive grain material mark
Quasi- color value;
S2, if point (R1, G1, B1) represent Ferrious material material color, point (R2, G2, B2) non-ferrous metal material color is represented,
Point (R3, G3, B3) represent red oxide material color in metal oxide, point (R4, G4, B4) represent black in metal oxide
Oxide material color, then according to obtained four distance value (D01、D02、D03、D04) analyzed and determined:
If min { D01, D02, D03, D04}=D01, then abrasive grain material to be identified is the material corresponding to Ferrious material material color
Matter;
If min { D01, D02, D03, D04}=D02, then abrasive grain material to be identified is the material corresponding to non-ferrous metal material color
Matter;
If min { D01, D02, D03, D04}=D03, then abrasive grain material to be identified is red oxide material in metal oxide
Material corresponding to color;
If min { D01, D02, D03, D04}=D04, then abrasive grain material to be identified is black oxide material in metal oxide
Material corresponding to color;
Further, since Ferrious material medium carbon steel, low-alloy steel, cast iron, Langaloy or high-alloy steel do not heat shape
State is brilliant white, if identified abrasive grain material is the white metals such as Ferrious material or aluminium, by abrasive grain by heat to 330 DEG C of temperature
Afterwards, then abrasive grain domain color (R ' to be identified is extracted0, G '0, B '0), carbon steel or low alloy steel is blue at 330 DEG C, and cast iron is straw colour
For color to bronze colour, Langaloy and the high alloy steel capital are white;It is sought away from olive drab(O.D) (R5, G5, B5), brilliant white (R1, G1, B1)
And blue (R6, G6, B6) distance value (D '05、D′01、D′06) analyzed and determined:
If min { D '05, D '01, D '06}=D '05, then abrasive grain material to be identified is cast iron;
If min { D '05, D '01, D '06}=D '01, then abrasive grain material to be identified is high-alloy steel or Langaloy or aluminium;
If min { D '05, D '01, D '06}=D '06, then abrasive grain material to be identified is carbon steel or low alloy steel;
Further, if identified abrasive grain material is high-alloy steel or Langaloy or aluminium, by abrasive grain by heat to 540 DEG C
After temperature, then extract abrasive grain domain color (R " to be identified0, G "0, B "0), high-alloy steel is olive drab(O.D) to bronze colour at 540 DEG C, high
Nickel alloy is blue;And it is sought away from olive drab(O.D) (R5, G5, B5), brilliant white (R1, G1, B1) and blue (R6, G6, B6) distance
It is worth (D "05、D″01、D″06) analyzed and determined:
If min { D "05, D "01, D "06}=D "05, then abrasive grain material to be identified is high-alloy steel;
If min { D "05, D "01, D "06}=D "01, then abrasive grain material to be identified is aluminium;
If min { D "05, D "01, D "06}=D "06, then abrasive grain material to be identified is Langaloy;
If identified abrasive grain material is ferroso-ferric oxide or tin/metal in S2, repeatedly step 3, seeks weight second
The value of big seed point is tin/metal if the value is blue or is orange, is otherwise ferroso-ferric oxide.
Compared with prior art, the present invention has following technique effect:
Abrasive grain thermal analysis method is combined by the present invention with image processing techniques, based on K-Mean clusters and Euclidean distance
Minimum distance classification realizes the automatic identification of regular abrasive grit material;
The present invention realizes point of susceptible condition abrasive grain using the abrasive grain half-tone information image under two kinds of color space coordinates
It cuts, all particle partitions Debris Analysis is composed suitable for offline iron;
The present invention accurately identifies the abrasive grain material of offline iron spectrogram picture, what raising Debris Analysis and equipment fault positioned
Efficiency promotes the popularization and application of Debris Analysis automation.
Description of the drawings
Fig. 1 is the abrasive grain material automatic identification overview flow chart based on color principal Component Extraction;
Fig. 2 (a)-(c) is respectively copper Debris Image, copper abrasive grain domain color schematic diagram after original copper Debris Image, segmentation;
Fig. 2 (d)-(f) is respectively aluminium Debris Image, aluminium abrasive grain domain color schematic diagram after original aluminum Debris Image, segmentation;
Fig. 3 is K-Mean cluster extraction Debris Image domain color flow charts;
Fig. 4 is the abrasive grain Material Identification flow chart of the minimum distance classification based on abrasive grain domain color and Euclidean distance;
Specific implementation mode
This method is illustrated below in conjunction with the accompanying drawings.
Referring to Fig.1, a kind of abrasive grain material automatic identifying method based on color principal Component Extraction, includes the following steps:
Step 1:Most secondary main materials of friction are Ferrious material and non-ferrous metal in mechanical equipment, therefore are directed to material
Ferrious material and its oxide and non-ferrous metal are mainly studied in the identification of matter.By studying the color of various materials itself and its each
Temper color under grade temperature, builds material standard color sample database.It is as follows:
S1 determines Ferrious material material color card, includes mainly carbon steel and low alloy steel, cast iron, Langaloy, height
Steel alloy, it is brilliant white not heat all usually, therefore chooses tempering face corresponding under 330 DEG C and 540 DEG C of two temperatures
Color is blue as carbon steel and low alloy steel at wherein 330 DEG C of various Ferrious material material standard colors, cast iron be olive drab(O.D) extremely
Bronze colour, Langaloy and the high alloy steel capital are white;At 540 DEG C high-alloy steel be olive drab(O.D) to bronze colour, Langaloy is
Blue.
S2 determines non-ferrous metal material color card, and including mainly the whites such as copper alloy, tin/metal, aluminium alloy has
Non-ferrous metal.Wherein copper is orange colour;It is in black, surface visible blue or orange colored spots that tin/metal, which is under white light,;Aluminium closes
The white non-ferrous metal such as gold is white.
S3 determines metal oxide material color card, includes mainly red oxide (di-iron trioxide) and black
Oxide (ferroso-ferric oxide).Wherein red oxide is salmon pink;Black oxide is brownish black.
Step 2:RGB image gray processing and S, V component gray processing carried out to Debris Image to be identified, and to gray processing after
Image carry out adaptive threshold fuzziness and Morphological scale-space, find out RGB image gray processing using zone marker interconnection algorithm
The abrasive grain area of Debris Image after rear image, S components Debris Image, the three width segmentation corresponding to V component Debris Image, is protected
Maximum abrasive grain in respective image, three width images of comparison is stayed to select the wherein maximum Debris Image of abrasive grain area, segmentation effect
As shown in Fig. 2 (b) and (e), it is as follows:
Debris Image is carried out RGB image gray processing, S components and V component gray processing, respectively such as formula (1)-(3) institute by S1
Show:
Formula (1):F (x, y)=0.299R (x, y)+0.587G (x, y)+0.114B (x, y)
Formula (2):
Formula (3):
R, G, B of pixel divide at coordinate (x, y) in R (x, y), G (x, y), B (x, y) representative image in formula (1)-(3)
Magnitude;
S2 is as follows to three width abrasive grain gray level images into row threshold division:
1) note T is the segmentation threshold of Debris Image foreground and background, and value range is 0~255, take 0 successively to T~
255 values are calculated;
2) foreground and background is divided the image into according to present threshold value T, it is w to find out foreground points respectively and account for image scaled0,
Average gray is u0;It is w that background points, which account for image scaled,1, average gray u1, the overall average gray scale of image is u;
3) variance of foreground and background image is calculated according to formula (4):
Formula (4):δ2(T)=w0×(u0-u)2+w1×(u1-u)2
4) variance δ corresponding when T takes 0~255 is found out respectively2(T), 256 variance yields are compared, when selecting variance minimum
The threshold value that corresponding T values are divided as Debris Image;
5) gray level image is subjected to binaryzation operation according to formula (5) again, it is black to obtain abrasive grain, and background is white
Mask schemes:
Formula (5):
F (x, y) indicates the gray-scale map of Debris Image in formula (5);
S3, Morphological scale-space is carried out to the Debris Image after segmentation, that is, carries out " reversed " processing, and " closed operation " is handled, with
And carry out " holes filling " using " unrestrained water is filled " algorithm and operate, it is ensured that the integrality of divided Debris Image.
S4, utilizing " eight connected region algorithm " to find out, abrasive grain area is maximum to be retained as research observation object, compares simultaneously
Finally acquired abrasive grain, selection divide maximum Debris Image as target abrasive grain to three width Debris Images.
Step 3:The distribution of color of abrasive grain and uneven in usual Debris Image, therefore extracted using K-Mean clusters
The Debris Image domain color being partitioned into is to characterize the color of abrasive grain material, the domain color extracted such as Fig. 2 (c) and (f)
Shown, flow chart is as shown in figure 3, be as follows:
S1 defines an iterations upper limit M.
S2 looks for N number of point in figure at random, takes out its rgb value as seed point.
Then S3 finds a rgb value and its most similar seed point to each of figure point, and this point is added to
Rgb value is most close to be obtained in the point group of seed point place.
S4 calculates the average RGB value (cumulative total value/number) of seed point group, and using this rgb value as new kind
Sub- point.
Whether whether equal with old value S5 compares this new value.
If 1) equal, seed point convergence is completed, into the 6th step.
If 2) etc., continue to execute the 3rd step, until iterations reach M times.
S6, when seed point convergence completion or iterations reach M times, we are a row to the weight of all seed points
Sequence (weight wherein includes the number of point).
S7 takes out the value of the highest seed point of weight, the dominant hue of abrasive grain picture of this value i.e. needed for us.
Step 4:Extracted abrasive grain picture domain color and mark are sought using the minimum distance classification based on Euclidean distance
The distance of quasi- color card realizes the automatic identification of abrasive grain material, and identification process is as shown in figure 4, be as follows:
S1 seeks extracted abrasive grain picture domain color and the brilliant white in material color value, tangerine according to formula (6)
The Euclidean distance of this four classes color of yellow, salmon pink and black.
Formula (6):
Formula midpoint (R0, G0, B0) it is the abrasive grain picture domain color pixel value extracted, point (R1, G1, B1) it is abrasive grain material mark
Quasi- color value.
S2, if point (R1, G1, B1) represent brilliant white, point (R2, G2, B2) represent orange colour, point (R3, G3, B3) represent tangerine
Red, point (R4, G4, B4) represent black, then according to obtained four distance value (D01、D02、D03、D04) analyzed and determined,
It is as follows:
If 1) min { D01, D02, D03, D04}=D01, then abrasive grain material to be identified is the white metals such as Ferrious material or aluminium.
If 2) min { D01, D02, D03, D04}=D02, then abrasive grain material to be identified is copper.
If 3) min { D01, D02, D03, D04}=D03, then abrasive grain material to be identified is di-iron trioxide.
If 4) min { D01, D02, D03, D04}=D04, then abrasive grain material to be identified is ferroso-ferric oxide or tin/metal.
S3, if identified abrasive grain material is the white metals such as Ferrious material or aluminium in S2, by abrasive grain by heat to 330 DEG C of temperature
After degree, then extract abrasive grain domain color (R ' to be identified0, G '0, B '0), and it is sought away from olive drab(O.D) (R5, G5, B5), brilliant white (R1,
G1, B1) and blue (R6, G6, B6) distance value (D '05、D′01、D′06) analyzed and determined, it is as follows:
If 1) min { D '05, D '01, D '06}=D '05, then abrasive grain material to be identified is cast iron.
If 2) min { D '05, D '01, D '06}=D '01, then abrasive grain material to be identified is high-alloy steel
Or the white metals such as Langaloy or aluminium.
If 3) min { D '05, D '01, D '06}=D '06, then abrasive grain material to be identified is carbon steel and low alloy steel.
S4 incites somebody to action abrasive grain if identified abrasive grain material is high-alloy steel or the white metals such as Langaloy or aluminium in S3
After heat to 540 DEG C of temperature, then extract abrasive grain domain color (R " to be identified0, G "0, B "0), and it is sought away from olive drab(O.D) (R5, G5, B5)、
Brilliant white (R1, G1, B1) and blue (R6, G6, B6) distance value (D "05、D″01、D″06) analyzed and determined, specific steps are such as
Under:
If 1) min { D "05, D "01, D "06}=D "05, then abrasive grain material to be identified is high-alloy steel.
If 2) min { D "05, D "01, D "06}=D "01, then abrasive grain material to be identified is the white metals such as aluminium.
If 3) min { D "05, D "01, D "06}=D "06, then abrasive grain material to be identified is Langaloy.
S5, if identified abrasive grain material is ferroso-ferric oxide or tin/metal in S2, repeatedly step 3, seeks weight
The value of second largest seed point is tin/metal if the value is blue or is orange, is otherwise ferroso-ferric oxide.
Claims (9)
1. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction, which is characterized in that include the following steps:
Step 1:According to the temper color under the color of various materials itself and its temperature at different levels, material standard face is built
Colo(u)r atlas library;
Step 2:RGB image gray processing and S, V component gray processing are carried out to Debris Image to be identified, and to the figure after gray processing
As carrying out adaptive threshold fuzziness and Morphological scale-space, after finding out RGB image gray processing using zone marker interconnection algorithm
The abrasive grain area of Debris Image after image, S components Debris Image, the three width segmentation corresponding to V component Debris Image, retains each
The maximum abrasive grain from image, three width images of comparison select the wherein maximum Debris Image of abrasive grain area;
Step 3:Be partitioned into Debris Image domain color is extracted using K-Mean clusters;
Step 4:Extracted abrasive grain picture domain color and standard face are sought using the minimum distance classification based on Euclidean distance
The distance of colo(u)r atlas realizes the automatic identification of abrasive grain material.
2. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 1, feature
It is, material standard color sample database is built in step 1 and is included the following steps:
S1:It determines Ferrious material material color card, chooses tempering face corresponding under 330 DEG C and 540 DEG C of two temperatures
Color is as various Ferrious material material standard colors;
S2 determines non-ferrous metal material color card;
S3 determines metal oxide material color card, includes mainly red oxide and black oxide.
3. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 1, feature
It is, Debris Image is subjected to RGB image gray processing, S components and V component gray processing in step 2, respectively such as formula 1-3 institutes
Show:
Formula 1:F (x, y)=0.299R (x, y)+0.587G (x, y)+0.114B (x, y)
Formula 2:
Formula 3:
In formula (1)-(3) in R (x, y), G (x, y), B (x, y) representative image at coordinate (x, y) pixel R, G, B component
Value.
4. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 1, feature
It is, three width abrasive grain gray level images is as follows into row threshold division in step 2:
1) note T is the segmentation threshold of Debris Image foreground and background, and value range is 0~255, takes 0~255 value successively to T
It is calculated;
2) foreground and background is divided the image into according to present threshold value T, it is w to find out foreground points respectively and account for image scaled0, average
Gray scale is u0;It is w that background points, which account for image scaled,1, average gray u1, the overall average gray scale of image is u;
3) variance of foreground and background image is calculated according to formula 4:
Formula 4:δ2(T)=w0×(u0-u)2+w1×(u1-u)2
4) variance δ corresponding when T takes 0~255 is found out respectively2(T), 256 variance yields are compared, select variance minimum when institute right
The threshold value that the T values answered are divided as Debris Image;
5) gray level image is subjected to binaryzation operation according to formula 5 again, it is black to obtain abrasive grain, and background is the mask of white
Figure:
Formula 5:
F (x, y) indicates the gray-scale map of Debris Image in formula 5.
5. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 1, feature
It is, carrying out Morphological scale-space to the Debris Image after segmentation in step 2 is:Reverse process is carried out, closed operation is handled, and
Holes filling operation is carried out using " unrestrained water is filled " algorithm;Utilizing " eight connected region algorithm " to find out, abrasive grain area is maximum to be reserved for
Object is observed for research, while comparing three width Debris Images finally acquired abrasive grain, chooses and divides maximum Debris Image conduct
Target abrasive grain.
6. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 1, feature
It is, the specific steps that be partitioned into Debris Image domain color is extracted in step 3 include:
S1 defines an iterations upper limit M;
S2 looks for N number of point in figure at random, takes out its rgb value as seed point;
Then S3 finds a rgb value and its most similar seed point to each of figure point, and this point is added to rgb value
It is most close to obtain in the point group of seed point place;
S4 calculates the average RGB value of seed point group, and using this rgb value as new seed point;Average RGB value is cumulative total
Value/number;
Whether whether equal with old value S5 compares this new value;If equal, seed point convergence is completed, into the 6th step;
If differed, the 3rd step is continued to execute, until iterations reach M times;
S6, when seed point convergence completion or iterations reach M times, we do a sequence to the weight of all seed points;
Weight wherein includes the number of point;
S7 takes out the value of the highest seed point of weight, the dominant hue of abrasive grain picture of this value i.e. needed for us.
7. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 1, feature
It is, identifies and be as follows in step 4:
S1 seeks extracted abrasive grain picture domain color and the Ferrious material material face in material color value according to formula 6
Black oxide material in red oxide material color or metal oxide in color, non-ferrous metal material color, metal oxide
The Euclidean distance of matter color;
Formula 6:
Formula midpoint (R0,G0,B0) it is the abrasive grain picture domain color pixel value extracted, point (R1,G1,B1) it is abrasive grain material standard face
Color value;
S2, if point (R1,G1,B1) represent Ferrious material material color, point (R2,G2,B2) represent non-ferrous metal material color, point
(R3,G3,B3) represent red oxide material color in metal oxide, point (R4,G4,B4) represent black oxygen in metal oxide
Compound material color, then according to obtained four distance value (D01、D02、D03、D04) analyzed and determined:
If min { D01,D02,D03,D04}=D01, then abrasive grain material to be identified is the material corresponding to Ferrious material material color;
If min { D01,D02,D03,D04}=D02, then abrasive grain material to be identified is the material corresponding to non-ferrous metal material color;
If min { D01,D02,D03,D04}=D03, then abrasive grain material to be identified is red oxide material color in metal oxide
Corresponding material;
If min { D01,D02,D03,D04}=D04, then abrasive grain material to be identified is black oxide material color in metal oxide
Corresponding material.
8. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 7, feature
It is, if identified abrasive grain material is the white metals such as Ferrious material or aluminium, after abrasive grain is heated to 330 DEG C of temperature, then carries
Take abrasive grain domain color (R' to be identified0,G'0,B'0), carbon steel or low alloy steel is blue at 330 DEG C, and cast iron is olive drab(O.D) to bronze
Color, Langaloy and the high alloy steel capital are white;Carbon steel, low-alloy steel, cast iron, Langaloy or the non-heated condition of high-alloy steel
It is brilliant white;It is sought away from olive drab(O.D) (R5,G5,B5), brilliant white (R1,G1,B1) and blue (R6,G6,B6) distance value
(D'05、D'01、D'06) analyzed and determined:
If min { D'05,D'01,D'06}=D'05, then abrasive grain material to be identified is cast iron;
If min { D'05,D'01,D'06}=D'01, then abrasive grain material to be identified is high-alloy steel or Langaloy or aluminium;
If min { D'05,D'01,D'06}=D'06, then abrasive grain material to be identified is carbon steel or low alloy steel.
9. a kind of abrasive grain material automatic identifying method based on color principal Component Extraction according to claim 8, feature
It is, if identified abrasive grain material is high-alloy steel or Langaloy or aluminium, by abrasive grain by after heat to 540 DEG C of temperature, then carries
Take abrasive grain domain color (R " to be identified0,G”0,B”0), high-alloy steel is olive drab(O.D) to bronze colour at 540 DEG C, and Langaloy is indigo plant
Color;And it is sought away from olive drab(O.D) (R5,G5,B5), brilliant white (R1,G1,B1) and blue (R6,G6,B6) distance value (D "05、
D”01、D”06) analyzed and determined:
If min { D "05,D”01,D”06}=D "05, then abrasive grain material to be identified is high-alloy steel;
If min { D "05,D”01,D”06}=D "01, then abrasive grain material to be identified is aluminium;
If min { D "05,D”01,D”06}=D "06, then abrasive grain material to be identified is Langaloy;
If identified abrasive grain material is ferroso-ferric oxide or tin/metal in S2, repeatedly step 3, it is second largest to seek weight
The value of seed point is tin/metal if the value is blue or is orange, is otherwise ferroso-ferric oxide.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810162476.3A CN108446706B (en) | 2018-02-27 | 2018-02-27 | Automatic abrasive grain material identification method based on principal component extraction of colors |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810162476.3A CN108446706B (en) | 2018-02-27 | 2018-02-27 | Automatic abrasive grain material identification method based on principal component extraction of colors |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108446706A true CN108446706A (en) | 2018-08-24 |
CN108446706B CN108446706B (en) | 2021-01-19 |
Family
ID=63192686
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810162476.3A Active CN108446706B (en) | 2018-02-27 | 2018-02-27 | Automatic abrasive grain material identification method based on principal component extraction of colors |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108446706B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110717899A (en) * | 2019-09-25 | 2020-01-21 | 武汉易视维科技有限公司 | Online visual detection method for bead explosion defects of transparent filter rod |
WO2020135056A1 (en) * | 2018-12-29 | 2020-07-02 | 北京金山安全软件有限公司 | Image processing method and device |
CN111428544A (en) * | 2019-01-10 | 2020-07-17 | 百度在线网络技术(北京)有限公司 | Scene recognition method and device, electronic equipment and storage medium |
CN115018845A (en) * | 2022-08-09 | 2022-09-06 | 聊城市泓润能源科技有限公司 | Method for detecting quality of lubricating oil abrasive particles |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6516100B1 (en) * | 1998-10-29 | 2003-02-04 | Sharp Laboratories Of America, Inc. | Method for image characterization using color and texture statistics with embedded spatial information |
CN101908142A (en) * | 2010-08-04 | 2010-12-08 | 丁天 | Feature analysis-based video flame detecting method |
CN101964293B (en) * | 2010-08-23 | 2012-01-18 | 西安航空动力股份有限公司 | Metallographical microstructural image processing method |
CN102528661A (en) * | 2011-11-10 | 2012-07-04 | 浙江工业大学 | Observation method for fluid precision finishing of surfaces of micro-structures of molds |
CN102830121A (en) * | 2012-08-17 | 2012-12-19 | 浙江工业大学 | Real-time detection method of soft abrasive flow abrasive group |
CN103278431A (en) * | 2013-04-22 | 2013-09-04 | 西安交通大学 | On-line monitoring and analyzing method of lubrication oil wear particles based on video acquisition |
CN103886579A (en) * | 2013-12-11 | 2014-06-25 | 西安交通大学 | Abrasive particle chain self-adaptive segmentation method orienting online ferrographic image automatic identification |
CN104484675A (en) * | 2014-12-15 | 2015-04-01 | 上海海事大学 | Method for extraction of texture features and pattern recognition of ferrographic wear particles |
CN105758862A (en) * | 2016-01-26 | 2016-07-13 | 西安交通大学 | Online oxidative wear state monitoring method based on color extraction of wear particle image |
CN107677216A (en) * | 2017-09-06 | 2018-02-09 | 西安交通大学 | A kind of multiple abrasive particle three-dimensional appearance synchronous obtaining methods based on photometric stereo vision |
-
2018
- 2018-02-27 CN CN201810162476.3A patent/CN108446706B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6516100B1 (en) * | 1998-10-29 | 2003-02-04 | Sharp Laboratories Of America, Inc. | Method for image characterization using color and texture statistics with embedded spatial information |
CN101908142A (en) * | 2010-08-04 | 2010-12-08 | 丁天 | Feature analysis-based video flame detecting method |
CN101964293B (en) * | 2010-08-23 | 2012-01-18 | 西安航空动力股份有限公司 | Metallographical microstructural image processing method |
CN102528661A (en) * | 2011-11-10 | 2012-07-04 | 浙江工业大学 | Observation method for fluid precision finishing of surfaces of micro-structures of molds |
CN102830121A (en) * | 2012-08-17 | 2012-12-19 | 浙江工业大学 | Real-time detection method of soft abrasive flow abrasive group |
CN103278431A (en) * | 2013-04-22 | 2013-09-04 | 西安交通大学 | On-line monitoring and analyzing method of lubrication oil wear particles based on video acquisition |
CN103886579A (en) * | 2013-12-11 | 2014-06-25 | 西安交通大学 | Abrasive particle chain self-adaptive segmentation method orienting online ferrographic image automatic identification |
CN104484675A (en) * | 2014-12-15 | 2015-04-01 | 上海海事大学 | Method for extraction of texture features and pattern recognition of ferrographic wear particles |
CN105758862A (en) * | 2016-01-26 | 2016-07-13 | 西安交通大学 | Online oxidative wear state monitoring method based on color extraction of wear particle image |
CN107677216A (en) * | 2017-09-06 | 2018-02-09 | 西安交通大学 | A kind of multiple abrasive particle three-dimensional appearance synchronous obtaining methods based on photometric stereo vision |
Non-Patent Citations (1)
Title |
---|
王静秋: "铁谱图像分割及磨粒识别关键技术研究", 《中国优秀博士学位论文全文数据库》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020135056A1 (en) * | 2018-12-29 | 2020-07-02 | 北京金山安全软件有限公司 | Image processing method and device |
CN111428544A (en) * | 2019-01-10 | 2020-07-17 | 百度在线网络技术(北京)有限公司 | Scene recognition method and device, electronic equipment and storage medium |
CN111428544B (en) * | 2019-01-10 | 2023-05-26 | 百度在线网络技术(北京)有限公司 | Scene recognition method and device, electronic equipment and storage medium |
CN110717899A (en) * | 2019-09-25 | 2020-01-21 | 武汉易视维科技有限公司 | Online visual detection method for bead explosion defects of transparent filter rod |
CN115018845A (en) * | 2022-08-09 | 2022-09-06 | 聊城市泓润能源科技有限公司 | Method for detecting quality of lubricating oil abrasive particles |
CN115018845B (en) * | 2022-08-09 | 2022-10-25 | 聊城市泓润能源科技有限公司 | Method for detecting quality of lubricating oil abrasive particles |
Also Published As
Publication number | Publication date |
---|---|
CN108446706B (en) | 2021-01-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108446706A (en) | A kind of abrasive grain material automatic identifying method based on color principal Component Extraction | |
Stan et al. | Optimizing convolutional neural networks to perform semantic segmentation on large materials imaging datasets: X-ray tomography and serial sectioning | |
Singh et al. | Application of image processing and radial basis neural network techniques for ore sorting and ore classification | |
CN104680519B (en) | Seven-piece puzzle recognition methods based on profile and color | |
CN108305259B (en) | Multi-texture feature fusion type automatic abrasive particle type identification method | |
Peng et al. | Oxidation wear monitoring based on the color extraction of on-line wear debris | |
US8023737B2 (en) | Method and device for the segmentation of regions and related computer program product | |
CN105389581B (en) | A kind of rice germ plumule integrity degree intelligent identifying system and its recognition methods | |
CN106203496B (en) | Hydrographic curve extracting method based on machine learning | |
Danti | Segmentation and classification of raw arecanuts based on three sigma control limits | |
Duarte et al. | Segmenting high-quality digital images of stomata using the wavelet spot detection and the watershed transform | |
CN110728286B (en) | Abrasive belt grinding material removal rate identification method based on spark image | |
Changhui et al. | Overlapped fruit recognition for citrus harvesting robot in natural scenes | |
Arunachalam et al. | Identification of defects in fruits using digital image processing | |
Komenda | Automatic recognition of complex microstructures using the Image Classifier | |
CN111402321A (en) | Method for target identification and volume estimation of solid particles in high-temperature molten pool | |
JP2005532557A (en) | Method for on-line measurement of the melt phase | |
Ganesan et al. | Automatic detection of optic disc and blood vessel in retinal images using morphological operations and Ipachi model | |
Mishra et al. | Computer Vision Algorithm for the detection of fracture cracks in Oil Hardening Non-Shrinking (OHNS) die steel after machining process | |
Indra | Skin detection using color distance measurement and thresholding | |
Ganesan et al. | Detection and segmentation of retinal blood vessel in digital RGB and CIELUV color space fundus images | |
Rismayana et al. | Facial Skin Color Segmentation Using Otsu Thresholding Algorithm | |
Abrol et al. | Automated segmentation of leukocytes using marker-based watershed algorithm from blood smear images | |
CN112287831A (en) | Following robot multi-target identification system and method based on coded thermal infrared mark | |
Fu | Research on recognition method of ferrographic image wear particle based on image segmentation |
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 |