CN108827197B - Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence - Google Patents

Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence Download PDF

Info

Publication number
CN108827197B
CN108827197B CN201810234750.3A CN201810234750A CN108827197B CN 108827197 B CN108827197 B CN 108827197B CN 201810234750 A CN201810234750 A CN 201810234750A CN 108827197 B CN108827197 B CN 108827197B
Authority
CN
China
Prior art keywords
edge
data
point
psf
image
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.)
Active
Application number
CN201810234750.3A
Other languages
Chinese (zh)
Other versions
CN108827197A (en
Inventor
齐子诚
倪培君
郑颖
郭智敏
唐盛明
李红伟
王晓艳
左欣
乔日东
张维国
付康
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Weapon Science Academy Ningbo Branch
Original Assignee
China Weapon Science Academy Ningbo Branch
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Weapon Science Academy Ningbo Branch filed Critical China Weapon Science Academy Ningbo Branch
Priority to CN201810234750.3A priority Critical patent/CN108827197B/en
Publication of CN108827197A publication Critical patent/CN108827197A/en
Application granted granted Critical
Publication of CN108827197B publication Critical patent/CN108827197B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B15/00—Measuring arrangements characterised by the use of electromagnetic waves or particle radiation, e.g. by the use of microwaves, X-rays, gamma rays or electrons
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10081—Computed x-ray tomography [CT]

Abstract

The invention discloses a linear array industrial CT homogeneous material size measuring method for reducing edge degradation influence, which comprises the steps of placing cylindrical calibration test blocks on homogeneous material objects to be measured together for linear array industrial CT scanning, obtaining an edge degradation curve by utilizing an acquired cylindrical edge CT image, extracting a point diffusion function, establishing edge gray scale degradation models with different slopes by utilizing a single-peak Gaussian response theory, calculating a segmentation gray scale and an edge degradation length component of a current edge by utilizing a nominal cylindrical size, further obtaining a series of segmentation gray scale values and edge degradation lengths with different slopes, establishing an edge degradation evaluation model, preliminarily extracting edge distribution of the test blocks to be measured by utilizing binaryzation, calculating normal gray scale distribution of each point on the edge, and measuring the size of the homogeneous test blocks with unknown size and unknown slope by utilizing the edge degradation evaluation model. The method effectively inhibits the influence of edge degradation caused by different slopes on dimension measurement, and improves the precision of linear array industrial CT dimension measurement.

Description

Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence
Technical Field
The invention relates to the technical field of linear array industrial CT image size measurement, in particular to a method for accurately measuring the size of a homogeneous material, which reduces the degradation influence of a size edge.
Background
With the development of industry, the manufacturing level in China is greatly improved, the aspects of low dimensional accuracy, poor stability and the like still exist in the manufacturing aspect of parts with complex structures, and particularly when the complex structures in the additive manufacturing products are integrally formed, the measurement requirements of internal micro dimensions, curved surface contours and the like are increased day by day, and the measurement technology and level are problems which need to be solved urgently. The method for acquiring the size of the part can be divided into a contact method and a non-contact method, the contact method such as a three-coordinate method has the advantages of high detection precision, high measurement speed and the like, but the method is difficult to measure a tiny and easily-deformed workpiece and an internal structure which cannot be touched by a probe. Non-contact methods such as machine vision measurement have the advantages of non-contact, high efficiency, low cost and the like, but have the problems of low measurement precision, need of presetting a calibration point and the like. The laser measurement method has high precision, but has certain requirements on the surface material of the measured object. The non-contact method can not effectively meet the requirement for measuring the internal dimension, so that the research of the non-contact method capable of accurately measuring the internal dimension has very important significance. The industrial Computed Tomography (CT) technology can nondestructively measure the external and internal structures of an object and the size of a defect, effectively overcomes the defects that the traditional measuring method can only measure the external surface structure of the object and anatomically observe the size of the internal defect, and the like, and is a method commonly used for measuring the internal size of parts at home and abroad at present.
Due to the fact that the shapes of the parts are different, the surfaces of the parts are not perpendicular to a CT scanning plane, the volume effect of a CT image and the like, the edges of the parts in the CT image have wide gray transition areas, and gradual change edges are formed. The measurement of the size of the part based on the industrial CT image method is determined by the selection accuracy of the gray threshold in the image edge segmentation. However, in the actual measurement of the complex-structure parts, the edge gray level degradation is related to the density of the parts, the spatial resolution of the CT device, and the edge slope, and the conventional standard-length pixel calibration measurement method is adopted, so that the dimension measurement close to the standard length is more accurate, and the measurement error is larger when the dimension difference is larger, so that edge degradation compensation is necessary to reduce the dimension measurement error. Based on the current research situation, the invention combines the single-peak Gaussian response theory to establish a method which can reduce the influence of the slope of the edge of the workpiece on the gray scale segmentation of the size measurement, reduce the measurement error and improve the size measurement precision of the industrial CT.
Disclosure of Invention
The invention aims to solve the technical problem of providing a linear array industrial CT homogeneous material size measuring method for reducing edge degradation influence, which can accurately, reliably and nondestructively measure the internal structure size of a complex workpiece when measuring the complex workpiece.
The invention provides a linear array industrial CT homogeneous material size measuring method for reducing edge degradation influence, which solves the technical problems and comprises the following steps:
s1, manufacturing a cylindrical calibration test block with the same material as the measured object by a machining means, wherein the diameter of the calibration test block is more than 100 times larger than the pixel size of the CT image, placing the calibration test block and the measured object together for CT scanning, and obtaining an industrial CT image of the section to be measured;
s2, selecting a CT image area where a calibration test block is located, calculating the centroid of the calibration test block in the image by utilizing threshold segmentation accumulation, selecting a circular area with the radius of 2 times by taking the centroid as the center, extracting line segments with the center as a starting point and the edge of the circular area as an end point in the area, averaging data on all the line segments to obtain an edge ESF (edge diffusion function), deriving to obtain an original PSF (point diffusion function), obtaining an ideal PSF with noise removed by utilizing Gaussian fitting, and recording the diameter length of the calibration test block after measurement.
S3, establishing an edge diffusion gray scale change calculation model without considering noise influence by using a single-peak Gaussian response theory, and calculating the ideal ESF of the calibration test block obtained in the step S1 by using the established edge diffusion gray scale change calculation model. And calculating a vertical edge gray scale division threshold function by combining the gray scale of the calibration test block material, the background mean value and the measuring diameter length obtained in the step S2.
And S4, establishing edge gray segmentation threshold models with different slopes by using the vertical edge gray segmentation threshold function obtained in the step S3.
And S5, preprocessing the CT image of the measured object obtained in the step S1, obtaining a binary image of the measured object by adopting a traditional automatic threshold segmentation method, extracting the edge of the measured object, calculating a normal line of each point of the edge of the measured object, and extracting the gray distribution in the normal line direction. And (4) dividing each normal gray distribution by using the edge gray dividing threshold models with different slopes obtained in the step (S4) to obtain more accurate edge positions.
Preferably, the step S2 specifically includes:
s21, placing the calibration test block and the measured object together for scanning by using an industrial CT system, and obtaining an industrial CT image of the section to be measured, wherein the industrial CT image matrix is omega × omega pixels, the imaging range is sigma × sigma millimeters, the slice thickness of the linear array detector is upsilon millimeters, and the calculated pixel size ps is as follows:
Figure GDA0002482226990000021
s22, selecting a CT image area G (x, y) where the calibration test block is located, wherein x is an abscissa and y is an ordinate, counting the gray value distribution of the area, dividing the image gray histogram into two groups at a certain threshold by adopting a threshold dividing method, separating the background from the standard test block, setting the interior of the standard test block as 1 and the background as 0. Starting to search from the upper left corner of the image, accumulating all pixel point positions with the gray value of 1 and dividing the pixel point positions by the accumulated number n to obtain the centroid position of the standard test block in the image
Figure GDA0002482226990000031
Selecting a circular area with 2 times of radius by taking the mass center as a centerCalculating the distance from each pixel point to the center of mass of the region, accumulating the gray values of the pixel points with the same distance to obtain an average value, and obtaining one-dimensional edge ESF data lESF. Carrying out forward derivation on ESF, and normalizing to obtain original PSF data lPSF。
S23, applying the original PSF data l obtained in the step S22PSFAnd performing Gaussian fitting to obtain ideal PSF data with noise removed
Figure GDA0002482226990000032
Figure GDA0002482226990000033
Where X [ i ] is the input sequence X, a is the amplitude, u is the center, σ is the standard deviation, and c is the offset.
Calculating ideal Gaussian fit PSF data
Figure GDA0002482226990000034
With the original PSF data lPSFS of (c).
Figure GDA0002482226990000035
Wherein N is PSF data
Figure GDA0002482226990000036
Length of (d) from the original PSF data lPSFThe lengths of the two-way pipe are consistent,
Figure GDA0002482226990000037
is the ith element of an ideal Gaussian fit, liIs the original PSF data lPSFTo find the minimum variance sminCorresponding to
Figure GDA0002482226990000038
The PSF data was fitted to the best gaussian.
Preferably, the step S3 specifically includes:
s31, use ofThe best Gaussian fitting data obtained in the step S23
Figure GDA0002482226990000039
Calculating ideal ESF data of calibration test block by discrete integration method
Figure GDA00024822269900000310
Figure GDA00024822269900000311
Wherein i is 0,1,2
Figure GDA00024822269900000312
The number of samples of (a) to (b),
Figure GDA00024822269900000313
is the first element of the gaussian fit data.
S32, obtaining original PSF data l by utilizing the derivative obtained in S22PSFExtracting lPSFAt the corresponding one-dimensional edge ESF data l, at the start point position a and the end point position bESFThe mean value P of the gray scale of the material of the middle extraction calibration test block and the mean value B of the gray scale of the background are respectively
Figure GDA0002482226990000041
Wherein n is one-dimensional edge ESF data lESFLength. Calculating corrected ideal ESF data
Figure GDA0002482226990000042
Is composed of
Figure GDA0002482226990000043
S33, measuring diameter and length l by using calibration test blockstd(unit: mm), the pixel size ps obtained in the step S21, and the initial threshold value division ratio are calculated
Figure GDA0002482226990000044
Is composed of
Figure GDA0002482226990000045
Preferably, the step S4 specifically includes:
s41, using the optimal Gaussian fitting ideal PSF data obtained in the step S23
Figure GDA0002482226990000046
Method for solving degradation ideal ESF data edge s (t, k) with different slopes and different lengths by using edge gray scale change model without considering noisei) Is composed of
When vertical edges
Figure GDA0002482226990000047
When it is a sloping edge
Figure GDA0002482226990000048
In the formula, hiFor different dimension lengths, kiAre different slopes.
S42, using the edge-degraded ideal ESF data edge S (t, k) obtained in the step S41i) Calculating the degradation region length [ t ] of the degradation ideal ESF data with different slopes and different length edges]Comprises the following steps:
Figure GDA0002482226990000051
s43, utilizing the initial threshold segmentation ratio obtained in the step S33
Figure GDA0002482226990000052
And the edge degradation ideal ESF data S (t, k) with different slopes and lengths obtained in the step S42i) Calculating the gray scale division threshold proportion function P of different slopes and lengthst thd(ki) And the corresponding maximum pixel value reduction scale function Ft(ki)
Figure GDA0002482226990000053
Figure GDA0002482226990000054
Figure GDA0002482226990000055
Preferably, the step S5 specifically includes:
s51, using the CT image of the measured object obtained in the step S1, a binarized image G' (x, y) of the measured object is obtained by an automatic threshold segmentation method, where the internal pixel value of the measured object is 1 and the background pixel value is 0.
S52, the edge of the measurement target is extracted using the binarized image G' (x, y) obtained in step S51. Firstly, finding out a point with the uppermost pixel value of 1 in an image G' (x, y), searching the next adjacent point from the current point to obtain a point on an edge, defining the current initial searching direction as direction 1, and sequentially checking the adjacent points (8 fields) in each direction clockwise from the current direction to see whether a point with the pixel value of 1 exists, if so, the point is the next edge point until the searched next adjacent edge point is the initial starting point, all the edge points are found, and the set of all the found edge points is set as H.
S53, using the normal of each point in the edge point set H obtained in the step S52, the normal takes the edge point as the center length as 2(b-a) in the step S32, extracting the pixel value of the binarized image G' (x, y) in the 3 × 3 area with the edge point as the center, and arranging the positions of the pixel value 1 in the 3 × 3 area in order to form the horizontal coordinate sequence
Figure GDA0002482226990000056
And ordinate series
Figure GDA0002482226990000057
To be provided with
Figure GDA0002482226990000058
And
Figure GDA0002482226990000059
computing a linear fit segment f for the inputi[j]＝aixi[j]+bi，aiCalculating the normal slope of each edge point as
Figure GDA00024822269900000510
Is composed of
Figure GDA0002482226990000061
S54, obtaining the normal slope of each edge point by the step S53
Figure GDA0002482226990000062
And length 2(b-a), extracting the pixel value N on each edge point normal line segment in the CT image G (x, y)i[j]. Calculating the edge degradation length lN[t]And a maximum pixel value reduction ratio F on the normalN[t]
Figure GDA0002482226990000063
Figure GDA0002482226990000064
Obtaining the degradation region length [ t ] of the degradation ideal ESF data with different slopes and length edges by utilizing the step S42]Obtaining the edge degradation length l in the normal directionN[i]Corresponding edge slope kN. According to kNAnd FN[t]Determining the corresponding slope and length edge gray scale division threshold value proportion function P in the step S43t thd。
S55, P obtained by the step S54t thdAnd determining the edge position on the normal line. And accurately positioning all the preprocessed edge points on the whole CT image to obtain a more accurate edge position.
Compared with the prior art, the invention has the advantages that: the invention provides a linear array industrial CT homogeneous material size measuring method for reducing edge degradation influence, which calculates edge gray scale degradation caused by factors such as the edge slope of a measured object, the performance of an industrial CT system and the like through a single-peak Gaussian edge degradation theory, actually measures an edge degradation model of the current industrial CT system by using a standard test block, establishes edge segmentation functions with different slopes and lengths, and finally realizes the accurate position of the edge of the measured object in a CT image. The method is used for carrying out experimental verification on 20 reference blocks with different sizes and slopes measured by a three-coordinate method, wherein the relative errors of the sizes measured by the method are less than 2%, and the standard error difference of a series of size measurement results is less than 1.5%. Therefore, by reducing the influence of the slope and the CT performance on the edge degradation, the system error and the random error are reduced, and the precision of the linear array industrial CT system dimension measurement is improved.
Drawings
Fig. 1 is a flowchart of a linear array industrial CT homogeneous material dimension measurement method for reducing edge degradation influence according to the present invention.
FIG. 2 shows the raw PSF data l obtained by normalization in the present inventionPSFAnd best Gaussian fit PSF data
Figure GDA0002482226990000065
FIG. 3 shows a partially different length edge degraded ideal ESF data edge s (t, k) in the present inventioni)。
Fig. 4(a) and 4(b) are graphs of the proportional function of the edge gray scale division threshold value and the corresponding proportional function of the maximum pixel value decrease under different length conditions with a certain slope in the present invention.
FIG. 5 is a theoretical edge map of different slopes in accordance with the present invention.
FIG. 6 is a graph of different slope theoretical degradation edges in accordance with the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
In the specific embodiment, a cylindrical test block made of 304 stainless steel is selected, and in order to establish an edge degradation gray scale change calculation model, the test block with a known diameter and length is used as a reference. The industrial CT system adopting the linear array detector scans the circular test block and the measured object together to obtain an industrial CT image of the section to be measured, and finally, further analysis and modeling are carried out on a computer.
Fig. 1 is a flowchart of a linear array industrial CT homogeneous material dimension measurement method for reducing edge degradation influence according to the present invention, and the modeling and measurement steps are as follows.
S1, machining by adopting a slow cutting process to manufacture a cylindrical test block, wherein the diameter of the calibration test block is more than 100 times larger than the pixel size of the CT image, and the test block is not more than half of the cross section area of the object to be measured. The roundness, the size precision, the cylinder perpendicularity and the surface finish degree of the cylindrical test block are ensured by a production process (such as heat treatment) under the same conditions as an object to be measured, and the diameter of the cylindrical test block is measured by a three-coordinate measuring machine, wherein the measurement precision of the three-coordinate measuring machine is higher than the minimum pixel size of an industrial CT system.
S2, selecting a CT image area where a calibration test block is located, calculating the centroid of the calibration test block in the image by utilizing threshold segmentation accumulation, selecting a circular area with the radius of 2 times by taking the centroid as the center, extracting line segments with the center as a starting point and the edge of the circular area as an end point in the area, averaging data on all the line segments to obtain an edge ESF (edge diffusion function), deriving to obtain an original PSF (point diffusion function), obtaining an ideal PSF with noise removed by utilizing Gaussian fitting, and recording the diameter length of the calibration test block after measurement. Specifically;
s21, placing the calibration test block and the measured object together for scanning by using an industrial CT system, and obtaining an industrial CT image of the section to be measured, wherein the industrial CT image matrix is omega × omega pixels, the imaging range is sigma × sigma millimeters, the slice thickness of the linear array detector is upsilon millimeters, and the calculated pixel size ps is as follows:
Figure GDA0002482226990000071
s22, selecting a CT image area G (x, y) where the calibration test block is located, wherein x is an abscissa and y is an ordinate, counting the gray value distribution of the area, dividing the image gray histogram into two groups at a certain threshold by adopting a threshold dividing method, separating the background from the standard test block, setting the interior of the standard test block as 1 and the background as 0. Starting to search from the upper left corner of the image, accumulating all pixel point positions with the gray value of 1 and dividing the pixel point positions by the accumulated number n to obtain the centroid position of the standard test block in the image
Figure GDA0002482226990000081
Selecting a circular area with the radius of 2 times by taking the center of mass as a center, calculating the distance from each pixel point in the area to the center of mass, accumulating the gray values of the pixel points with the same distance, and averaging to obtain one-dimensional edge ESF data lESF. Carrying out forward derivation on ESF, and normalizing to obtain original PSF data lPSF。
S23, applying the original PSF data l obtained in the step S22PSFAnd performing Gaussian fitting to obtain ideal PSF data with noise removed
Figure GDA0002482226990000082
Figure GDA0002482226990000083
Where X [ i ] is the input sequence X, a is the amplitude, u is the center, σ is the standard deviation, and c is the offset.
Calculating ideal Gaussian fit PSF data
Figure GDA0002482226990000084
With the original PSF data lPSFS of (c).
Figure GDA0002482226990000085
Wherein N is PSF data
Figure GDA0002482226990000086
Length of (d) from the original PSF data lPSFThe lengths of the two-way pipe are consistent,
Figure GDA0002482226990000087
is the ith element of an ideal Gaussian fit, liIs the original PSF data lPSFTo find the minimum variance sminCorresponding to
Figure GDA0002482226990000088
The PSF data was fitted to the best gaussian.
FIG. 2 shows the raw PSF data l obtained by normalization in the present inventionPSFAnd best Gaussian fit PSF data
Figure GDA0002482226990000089
S3, establishing an edge diffusion gray scale change calculation model without considering noise influence by using a single-peak Gaussian response theory, and calculating the ideal ESF of the calibration test block obtained in the step S1 by using the established edge diffusion gray scale change calculation model. And calculating a vertical edge gray scale division threshold function by combining the gray scale of the calibration test block material, the background mean value and the measuring diameter length obtained in the step S2. Specifically;
s31, utilizing the best Gaussian fitting data obtained in the step S23
Figure GDA00024822269900000810
Calculating ideal ESF data of calibration test block by discrete integration method
Figure GDA00024822269900000811
Figure GDA00024822269900000812
Wherein i is 0,1,2
Figure GDA00024822269900000813
The number of samples of (a) to (b),
Figure GDA00024822269900000814
is the first element of the gaussian fit data.
S32, obtaining original PSF data l by utilizing the derivative obtained in S22PSFExtracting lPSFAt the corresponding one-dimensional edge ESF data l, at the start point position a and the end point position bESFThe mean value P of the gray scale of the material of the middle extraction calibration test block and the mean value B of the gray scale of the background are respectively
Figure GDA0002482226990000091
Wherein n is one-dimensional edge ESF data lESFLength. Calculating corrected ideal ESF data
Figure GDA0002482226990000092
Is composed of
Figure GDA0002482226990000093
S33, measuring diameter and length l by using calibration test blockstd(unit: mm), the pixel size ps obtained in the step S21, and the initial threshold value division ratio are calculated
Figure GDA0002482226990000094
Is composed of
Figure GDA0002482226990000095
S4, establishing edge gray segmentation threshold models with different slopes by using the vertical edge gray segmentation threshold function obtained in the step S3, and referring to FIG. 5, the invention is a theoretical edge map with different slopes. Specifically;
s41, as shown in FIG. 6, is a graph of different slope theoretical degradation edges in the present invention. Using the best Gauss obtained in the step S23Fitting ideal PSF data
Figure GDA0002482226990000096
Method for solving degradation ideal ESF data edge s (t, k) with different slopes and different lengths by using edge gray scale change model without considering noisei) Is composed of
When vertical edges
Figure GDA0002482226990000097
When it is a sloping edge
Figure GDA0002482226990000101
In the formula, hiFor different dimension lengths, kiAre different slopes.
FIG. 3 shows an edge s (t, k) of the ESF data with partially degraded different length edges according to the present inventioni)。
S42, using the edge-degraded ideal ESF data edge S (t, k) obtained in the step S41i) Calculating the degradation region length [ t ] of the degradation ideal ESF data with different slopes and different length edges]Comprises the following steps:
Figure GDA0002482226990000102
s43, utilizing the initial threshold segmentation ratio obtained in the step S33
Figure GDA0002482226990000103
And the edge degradation ideal ESF data S (t, k) with different slopes and lengths obtained in the step S42i) Calculating the gray scale division threshold proportion function P of different slopes and lengthst thd(ki) And the corresponding maximum pixel value reduction scale function Ft(ki)
Figure GDA0002482226990000104
Figure GDA0002482226990000105
Figure GDA0002482226990000106
Fig. 4(a) and 4(b) are graphs of the proportional function of the edge gray scale division threshold value and the corresponding proportional function of the maximum pixel value decrease under different length conditions with a certain slope in the present invention.
And S5, preprocessing the CT image of the measured object obtained in the step S1, obtaining a binary image of the measured object by adopting a traditional automatic threshold segmentation method, extracting the edge of the measured object, calculating a normal line of each point of the edge of the measured object, and extracting the gray distribution in the normal line direction. And (4) dividing each normal gray distribution by using the edge gray dividing threshold models with different slopes obtained in the step (S4) to obtain more accurate edge positions. Specifically;
s51, using the CT image of the measured object obtained in the step S1, a binarized image G' (x, y) of the measured object is obtained by an automatic threshold segmentation method, where the internal pixel value of the measured object is 1 and the background pixel value is 0.
S52, the edge of the measurement target is extracted using the binarized image G' (x, y) obtained in step S51. Firstly, finding out a point with the uppermost pixel value of 1 in an image G' (x, y), searching the next adjacent point from the current point to obtain a point on an edge, defining the current initial searching direction as direction 1, and sequentially checking the adjacent points (8 fields) in each direction clockwise from the current direction to see whether a point with the pixel value of 1 exists, if so, the point is the next edge point until the searched next adjacent edge point is the initial starting point, all the edge points are found, and the set of all the found edge points is set as H.
S53, using the normal of each point in the edge point set H obtained in the step S52, the length of the normal taking the edge point as the center is 2(b-a) in the step S32The pixel values of the binarized image G' (x, y) in the 3 × 3 region centered on the edge point are extracted, and the positions of 1 point in the pixel value in the 3 × 3 region are arranged in order to form an abscissa series
Figure GDA0002482226990000111
And ordinate series
Figure GDA0002482226990000112
To be provided with
Figure GDA0002482226990000113
And
Figure GDA0002482226990000114
computing a linear fit segment f for the inputi[j]＝aixi[j]+bi，aiCalculating the normal slope of each edge point as
Figure GDA0002482226990000115
Is composed of
Figure GDA0002482226990000116
S54, obtaining the normal slope of each edge point by the step S53
Figure GDA0002482226990000117
And length 2(b-a), extracting the pixel value N on each edge point normal line segment in the CT image G (x, y)i[j]. Calculating the edge degradation length lN[t]And a maximum pixel value reduction ratio F on the normalN[t]
Figure GDA0002482226990000118
Figure GDA0002482226990000119
Obtaining the different inclinations using the step S42Rate and Length edge degradation the degradation region length t of ideal ESF data]Obtaining the edge degradation length l in the normal directionN[i]Corresponding edge slope kN. According to kNAnd FN[t]Determining the corresponding slope and length edge gray scale division threshold value proportion function P in the step S43t thd。
S55, P obtained by the step S54t thdAnd determining the edge position on the normal line. And accurately positioning all the preprocessed edge points on the whole CT image to obtain a more accurate edge position.
A series of length blocks were measured and the results are shown in table 1.
Figure GDA0002482226990000121
Through comparison, the measurement method of the invention has small measurement error, which shows that the measurement method of the invention can reduce system error and random error, effectively inhibit the influence of edge degradation caused by different slopes on the dimension measurement, and improve the precision of the linear array industrial CT dimension measurement.

Claims (5)

1. A linear array industrial CT homogeneous material size measuring method for reducing edge degradation influence is characterized in that: comprises the following steps:
s1, manufacturing a cylindrical calibration test block with the same material as the measured object by a machining means, wherein the diameter of the calibration test block is more than 100 times larger than the pixel size of the CT image, placing the calibration test block and the measured object together for CT scanning to obtain an industrial CT image of the section to be measured, and recording the diameter length measured by the calibration test block;
s2, selecting a CT image area where a calibration test block is located, calculating the centroid of the calibration test block in the image by utilizing threshold segmentation accumulation, selecting a circular area with the radius of 2 times by taking the centroid as the center, extracting line segments with the center as a starting point and the edge of the circular area as an end point in the area, averaging the data on all the line segments to obtain edge ESF (edge diffusion function) data, and deriving and normalizing the edge ESF data to obtain original PSF (point diffusion function) data; obtaining ideal PSF data with noise removed by using Gaussian fitting, and finally calculating by using the ideal PSF data and the original PSF data to obtain optimal ideal PSF data;
s3, calculating the ideal ESF data of the calibration test block by using the optimal ideal PSF data obtained in the step S2; extracting a calibration test block material gray level mean value P and a background gray level mean value B by using the edge ESF data and the original PSF data obtained in the step S2, and calculating an initial threshold segmentation ratio by combining the calibration test block material gray level mean value P, the background gray level mean value B and the diameter length measured in the step S1;
s4, establishing an edge diffusion gray scale change calculation model without considering noise influence by using a single-peak Gaussian response theory, and calculating the ideal ESF data of the calibration test block obtained in the step S2 by using the established edge diffusion gray scale change calculation model; establishing edge gray scale segmentation threshold models with different slopes and lengths by using the initial threshold segmentation proportion obtained in the step S3;
s5, preprocessing the CT image of the measured object obtained in the step S1, obtaining a binary image of the measured object by adopting a traditional automatic threshold segmentation method, extracting the edge of the measured object, calculating a normal line of each point of the edge of the measured object, and extracting the gray distribution in the normal line direction; and (4) dividing each normal gray distribution by using the edge gray dividing threshold models with different slopes and lengths obtained in the step (S4) to obtain more accurate edge positions.
2. The measurement method according to claim 1, characterized in that: the step S2 specifically includes:
s21, placing the calibration test block and the measured object together for scanning by using an industrial CT system, and obtaining an industrial CT image of the section to be measured, wherein the industrial CT image matrix is omega × omega pixels, the imaging range is sigma × sigma millimeters, the slice thickness of the linear array detector is upsilon millimeters, and the calculated pixel size ps is as follows:
Figure RE-FDA0002497527090000011
s22, selecting a CT image area G (x, y) where a calibration test block is located, wherein x is a horizontal coordinate, y is a vertical coordinate, counting the gray value distribution of the area, dividing an image gray histogram into two groups at a certain threshold by adopting a threshold dividing method, separating a background from a standard test block, setting the interior of the standard test block as 1 and the background as 0; starting to search from the upper left corner of the image, accumulating all pixel point positions with the gray value of 1 and dividing the pixel point positions by the accumulated number n to obtain the centroid position of the standard test block in the image
Figure RE-FDA0002497527090000021
Selecting a circular area with the radius of 2 times by taking the center of mass as a center, calculating the distance from each pixel point in the area to the center of mass, accumulating the gray values of the pixel points with the same distance, and averaging to obtain one-dimensional edge ESF data lESFForward derivation is carried out on ESF, and original PSF data l is obtained by normalizationPSF；
S23, applying the original PSF data l obtained in the step S22PSFAnd performing Gaussian fitting to obtain ideal PSF data with noise removed
Figure RE-FDA0002497527090000022
Figure RE-FDA0002497527090000023
Where X [ i ] is the input sequence X, a is the amplitude, u is the center, σ is the standard deviation, and c is the offset;
calculating ideal PSF data
Figure RE-FDA0002497527090000024
With the original PSF data lPSFS;
Figure RE-FDA0002497527090000025
where N is ideal PSF data
Figure RE-FDA0002497527090000026
Length of (d) from the original PSF data lPSFThe lengths of the two-way pipe are consistent,
Figure RE-FDA0002497527090000027
is ideal PSF data
Figure RE-FDA0002497527090000028
The ith element of (1)iIs the original PSF data lPSFTo find the minimum variance sminCorresponding to
Figure RE-FDA0002497527090000029
For the best ideal PSF data, note as
Figure RE-FDA00024975270900000210
3. The measurement method according to claim 2, characterized in that: the step S3 specifically includes:
s31, using the optimal ideal PSF data obtained in the step S23
Figure RE-FDA00024975270900000211
Calculating ideal ESF data of calibration test block by discrete integral
Figure RE-FDA00024975270900000212
Figure RE-FDA00024975270900000213
Wherein i is 0,1,2
Figure RE-FDA00024975270900000214
The number of samples of (a) to (b),
Figure RE-FDA00024975270900000215
is the optimal ideal PSF data
Figure RE-FDA00024975270900000216
The first element of (a);
s32, original PSF data l obtained by using the S22PSFExtracting lPSFAt the corresponding one-dimensional edge ESF data l, at the start point position a and the end point position bESFThe mean value P of the gray scale of the material of the middle extraction calibration test block and the mean value B of the gray scale of the background are respectively
Figure RE-FDA0002497527090000031
Wherein n is one-dimensional edge ESF data lESFA length; calculating corrected ideal ESF data
Figure RE-FDA0002497527090000032
Is composed of
Figure RE-FDA0002497527090000033
S33 diameter length l measured by using calibration test blockstd(unit: mm), the pixel size ps obtained in the step S21, and the initial threshold value division ratio are calculated
Figure RE-FDA0002497527090000034
Is composed of
Figure RE-FDA0002497527090000035
4. A measuring method according to claim 3, characterized in that: the step S4 specifically includes:
s41, using the optimal ideal PSF data obtained in the step S23
Figure RE-FDA0002497527090000036
Method for solving degradation ideal ESF data edge s (t, k) with different slopes and different lengths by using edge gray scale change model without considering noisei) Is composed of
When vertical edges
Figure RE-FDA0002497527090000037
When it is a sloping edge
Figure RE-FDA0002497527090000038
In the formula, hiFor different dimension lengths, kiThe slope is different, and upsilon is the slice thickness of the linear array detector;
s42, using the edge-degraded ideal ESF data edge S (t, k) obtained in the step S41i) Calculating the degradation region length [ t ] of the degradation ideal ESF data with different slopes and different length edges]Comprises the following steps:
Figure RE-FDA0002497527090000041
s43, utilizing the initial threshold segmentation ratio obtained in the step S33
Figure RE-FDA0002497527090000042
And the edge degradation ideal ESF data S (t, k) with different slopes and lengths obtained in the step S42i) Calculating the gray scale division threshold proportion function P of different slopes and lengthst thd(ki) And the corresponding maximum pixel value reduction scale function Ft(ki)
Figure RE-FDA0002497527090000043
5. The measurement method according to claim 4, characterized in that: the step S5 specifically includes:
s51, obtaining a measured object binary image G' (x, y) by using the measured object CT image obtained in the step S1 and adopting an automatic threshold segmentation method, wherein the internal pixel value of the measured object is 1 and the background pixel value is 0;
s52, extracting the edge of the object to be measured using the binarized image G' (x, y) obtained in step S51; firstly, finding out a point with the uppermost pixel value of 1 in an image G' (x, y), searching the next adjacent point from the current point to obtain a point on an edge, defining the current initial searching direction as direction 1, and clockwise sequentially checking the adjacent points in each direction from the current direction to see whether a point with the pixel value of 1 exists, if so, determining the point as the next edge point until the next adjacent edge point is the initial starting point, all the edge points are found, and setting the set of all the found edge points as H;
s53, using the normal of each point in the edge point set H obtained in the step S52, the normal takes the edge point as the center and the length is 2(b-a) in the step S32, extracting the pixel value of the binarized image G' (x, y) in the 3 × 3 area with the edge point as the center, and arranging the positions of the pixel value 1 in the 3 × 3 area in order to form the horizontal coordinate sequence
Figure RE-FDA0002497527090000044
And ordinate series
Figure RE-FDA0002497527090000051
To be provided with
Figure RE-FDA0002497527090000052
And
Figure RE-FDA0002497527090000053
computing a linear fit segment f for the inputi[j]＝aixi[j]+bi，aiCalculating the normal slope of each edge point for the fitting slope of the tangent line corresponding to each edge pointIs composed of
Figure RE-FDA0002497527090000054
Is composed of
Figure RE-FDA0002497527090000055
S54, obtaining the normal slope of each edge point by the step S53
Figure RE-FDA0002497527090000056
And length 2(b-a), extracting the pixel value N on each edge point normal line segment in the CT image G (x, y)i[j]Calculating the edge degradation length lN[t]And a maximum pixel value reduction ratio F on the normalN[t]
Figure RE-FDA0002497527090000057
Figure RE-FDA0002497527090000058
Obtaining the degradation region length [ t ] of the degradation ideal ESF data with different slopes and length edges by utilizing the step S42]Obtaining the edge degradation length l in the normal directionN[i]Corresponding edge slope kN(ii) a According to kNAnd FN[t]Determining the corresponding slope and length edge gray scale division threshold value proportion function P in the step S43t thd；
S55, P obtained by the step S54t thdDetermining the edge position on the normal; and accurately positioning all the preprocessed edge points on the whole CT image to obtain a more accurate edge position.
CN201810234750.3A 2018-03-21 2018-03-21 Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence Active CN108827197B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810234750.3A CN108827197B (en) 2018-03-21 2018-03-21 Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810234750.3A CN108827197B (en) 2018-03-21 2018-03-21 Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence

Publications (2)

Publication Number Publication Date
CN108827197A CN108827197A (en) 2018-11-16
CN108827197B true CN108827197B (en) 2020-07-17

Family

ID=64154188

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810234750.3A Active CN108827197B (en) 2018-03-21 2018-03-21 Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence

Country Status (1)

Country Link
CN (1) CN108827197B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110017797B (en) * 2019-04-24 2020-08-21 中国兵器科学研究院宁波分院 Dimension measurement result uncertainty evaluation method based on image isosurface segmentation method
CN113547208A (en) * 2021-07-27 2021-10-26 安徽速达数控设备有限责任公司 Gas supply system adjusting method of laser cutting machine
CN117058220B (en) * 2023-10-11 2024-01-26 天津施格自动化科技有限公司 Fitting algorithm for measuring diameter and central point of metal rod by using distance sensor

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4915166B2 (en) * 2006-08-04 2012-04-11 日本電気株式会社 Blur filter design method
JP5416377B2 (en) * 2008-08-28 2014-02-12 アンリツ産機システム株式会社 Image processing apparatus, X-ray foreign object detection apparatus including the same, and image processing method
CN106447684A (en) * 2016-08-10 2017-02-22 中国兵器科学研究院宁波分院 Industrial CT image weak edge dimension measuring method
CN106841115A (en) * 2017-02-07 2017-06-13 南京理工大学 Offshore spilled oil detection method based on spatial frequency features

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4915166B2 (en) * 2006-08-04 2012-04-11 日本電気株式会社 Blur filter design method
JP5416377B2 (en) * 2008-08-28 2014-02-12 アンリツ産機システム株式会社 Image processing apparatus, X-ray foreign object detection apparatus including the same, and image processing method
CN106447684A (en) * 2016-08-10 2017-02-22 中国兵器科学研究院宁波分院 Industrial CT image weak edge dimension measuring method
CN106841115A (en) * 2017-02-07 2017-06-13 南京理工大学 Offshore spilled oil detection method based on spatial frequency features

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
涡轮叶片CT图像边缘提取的最优算子研究;孙晶晶等;《航空动力学报》;20100131;第25卷(第1期);第175-179页 *
金属材料内部缺陷精确工业CT测量方法;齐子诚等;《强激光与粒子束》;20180228;第30卷(第2期);第025102-1-025102-7页 *

Also Published As

Publication number Publication date
CN108827197A (en) 2018-11-16

Similar Documents

Publication Publication Date Title
CN110992346B (en) Fatigue crack length online detection method based on DIP and DICM
CN108827197B (en) Linear array industrial CT homogeneous material size measurement method capable of reducing edge degradation influence
CN108613630B (en) Two-wire tube level bubble offset measurement method based on image processing technology
CN115100203A (en) Steel bar polishing and rust removing quality detection method
CN103292701A (en) Machine-vision-based online dimensional measurement method of precise instrument
CN109115800B (en) Method for rapidly detecting burrs of product and accurately measuring length
CN110906875B (en) Visual processing method for aperture measurement
CN107796826B (en) Micro duplicate gear broken tooth defect detection method based on tooth center distance curve analysis
CN110060293B (en) Defect detection performance limit evaluation method of CT detection system
CN113689415A (en) Steel pipe wall thickness online detection method based on machine vision
CN103759672A (en) Vision measurement method for ice cream stick plane contour dimensions
CN111047588A (en) Imaging measurement method for size of shaft type small part
Duan et al. High precision edge detection algorithm for mechanical parts
CN108876860A (en) A kind of image calibration method for pipe bubble offset measurement
CN114577131A (en) 3D structured light camera-based vehicle body clearance detection method and system
CN110017797B (en) Dimension measurement result uncertainty evaluation method based on image isosurface segmentation method
CN102506753B (en) Fourteen-point spherical wavelet transformation-based shape difference detection method for irregular parts
CN110220967B (en) Automatic identification method for cluster crack surface profile by ACFM technology
CN110544276A (en) Least square method ellipse fitting piston skirt maximum point size measurement method
CN112669460B (en) Workpiece defect detection method, system and computer readable storage medium
CN108844469B (en) Method and system for testing workpiece step height based on laser
CN110084786B (en) Automatic defect identification method for digital X-ray image with gradually-changed background
CN111815575A (en) Bearing steel ball part detection method based on machine vision
CN109238198B (en) Method for obtaining maximum processing error of curved surface parts processed in batch
CN113607058B (en) Straight blade size detection method and system based on machine vision

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