CN111583239A - Honeycomb structure geometric regularity image recognition method and system - Google Patents
Honeycomb structure geometric regularity image recognition method and system Download PDFInfo
- Publication number
- CN111583239A CN111583239A CN202010388353.9A CN202010388353A CN111583239A CN 111583239 A CN111583239 A CN 111583239A CN 202010388353 A CN202010388353 A CN 202010388353A CN 111583239 A CN111583239 A CN 111583239A
- Authority
- CN
- China
- Prior art keywords
- pixel
- point
- image
- vertex
- value
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G03—PHOTOGRAPHY; CINEMATOGRAPHY; ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ELECTROGRAPHY; HOLOGRAPHY
- G03B—APPARATUS OR ARRANGEMENTS FOR TAKING PHOTOGRAPHS OR FOR PROJECTING OR VIEWING THEM; APPARATUS OR ARRANGEMENTS EMPLOYING ANALOGOUS TECHNIQUES USING WAVES OTHER THAN OPTICAL WAVES; ACCESSORIES THEREFOR
- G03B17/00—Details of cameras or camera bodies; Accessories therefor
- G03B17/56—Accessories
- G03B17/566—Accessory clips, holders, shoes to attach accessories to camera
-
- 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
- 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
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02K—DYNAMO-ELECTRIC MACHINES
- H02K7/00—Arrangements for handling mechanical energy structurally associated with dynamo-electric machines, e.g. structural association with mechanical driving motors or auxiliary dynamo-electric machines
- H02K7/06—Means for converting reciprocating motion into rotary motion or vice versa
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30172—Centreline of tubular or elongated structure
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Power Engineering (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for identifying geometric regularity images of a honeycomb structure, wherein the method comprises the steps of obtaining images, processing the images, extracting vertexes, reconstructing cell elements and evaluating quality; setting a step of binaryzation between the step of image processing and the step of vertex extraction, wherein the binaryzation is to set the pixel value of a background in an image to be 0 and set the pixel value of a honeycomb framework in the image to be 1 to form a binaryzation image; the step of "quality evaluation" is based on the cell reconstruction map, and calculates the angular deviation values and the average values thereof, the line deviation values and the average values thereof of all the cells, and then determines whether the cells are qualified. The system comprises a detection table, a digital camera and a computer; the digital camera is electrically connected with the computer; at least one digital camera with resolution not lower than 1080P is provided with telecentric lens and fixed or movable installation mode. The method and the system have the advantages of being scientific, reasonable, simple and easy to implement, high in detection precision, high in working efficiency and the like.
Description
Technical Field
The invention relates to the fields of design, manufacture, application and the like of light structure products of equipment such as transportation, machinery, aerospace, ships and the like, in particular to a method and a system for identifying geometric regularity images of a honeycomb structure.
Background
Lightweight honeycomb products are widely used due to their excellent load-bearing capacity and good energy-absorbing properties, and for example, lightweight honeycomb products are also widely used in high-speed trains. However, the honeycomb product is prone to cell deformation during production, transportation and use, and the deformation has a significant influence on the performance of the honeycomb product. Therefore, the cellular deformation, i.e., the geometric regularity, of the honeycomb product needs to be evaluated so as to judge the quality of the honeycomb product.
Chinese patent application No. 201910503383.7, filed 2019, 9 and 13, discloses a method for identifying honeycomb edges from honeycomb core surface measurement data. The method comprises the following steps: carrying out two-dimensional coordinate transformation on the collected three-dimensional data of the surface of the honeycomb core, and identifying angular points in a two-dimensional plane projection image of the honeycomb core through an angular point detection algorithm, wherein the angular points comprise a human-shaped vertex, a Y-shaped vertex, a pseudo vertex and unidentified missing angular points; based on the proposed corner type judgment algorithm, two end points of the honeycomb edge are sequentially identified, wherein one end point is determined when the adjacent edge is identified, and the other end point is determined through local analysis of the identified vertex; in the realization process of the honeycomb edge, the false vertex can be effectively eliminated, meanwhile, the missing vertex is supplemented, and the stable and high-precision honeycomb edge recognition is realized. After the cellular edges are identified, the geometric regularity of the cellular product can be evaluated, and the quality of the cellular product can be judged according to the geometric regularity. The method has the advantages of high precision and good robustness, but the method needs to scan the surface of the honeycomb core point by point, so that the time is long, and the steps are complicated.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a method and a system for identifying geometric regularity images of a honeycomb structure, which have high working efficiency and high analysis precision.
The purpose of the invention is realized by the following technical scheme:
the method comprises the following steps:
a method for cellular structure geometry regularity image recognition, the method comprising the steps of: acquiring an image, processing the image, extracting a vertex, reconstructing a cell element and evaluating quality; the step of acquiring the image comprises shooting the image and reading the image by a computer; the step of 'vertex extraction' is to find and record the vertex of a cell on the basis of 'image processing'; in the step of reconstructing the cell element, the extracted vertexes are connected according to the mapping relation between the cell element and the vertex to obtain a cell element reconstruction picture;
setting a step of binaryzation between the step of image processing and the step of vertex extraction, wherein the step of binaryzation is to set the pixel value of a background in an image to be 0 and set the pixel value of a honeycomb skeleton in the image to be 1 to form a binaryzation image;
the step of 'quality evaluation' is based on the cell reconstruction graph, calculates the angular deviation values and the average values thereof, the line deviation values and the average values thereof of all the cells, and compares the angular deviation values and the average values with the set tolerance band to judge whether the cell is qualified.
In the step of 'vertex extraction', based on a 'binary image', determining wall thickness, assigning pixels, determining cell edge length and annihilating pixels are sequentially executed;
the step of determining the wall thickness is to set a square window with a side length capable of changing from small to large, and after a window with a certain side length traverses the morphological image, if the minimum value of the number of pixels of which the pixel value in the window is 0 is non-zero, define the side length of the square window as the wall thickness L;
the step of pixel assignment is to set an assignment window with a square side length of L, traverse the pixel point with the pixel value of 1 in the morphological image by the assignment window, and then assign the sum of the pixel value of 1 in the window to the pixel point at the center point of the assignment window;
the step of determining the side length of the cell element is to find the pixel point with the largest assignment after completing the pixel assignment, establish a square area by taking the point as the center, calculate the difference absolute value Z of the assignment of each pixel point on four boundaries of the area and the assignment of the center point, traverse the whole morphological image, and record the smallest difference absolute value Z and the coordinate thereof in the round; establishing a new square area by E-E +2 pixels, repeating the process until the minimum difference absolute value Z has an obvious trend of increasing reversely, taking out the pixel point coordinate of the minimum difference absolute value Z and the coordinate of the corresponding area center point at the moment, and solving to obtain the side length A of the honeycomb cell element through the coordinate values of the pixel point coordinate and the corresponding area center point coordinate;
in the step of pixel annihilation, a pixel point with the largest assignment is found, the pixel point is determined as a vertex and recorded, then an annihilation window with a square shape with the side length A of the honeycomb cell as the side length is established with the pixel point as the center, all assignments on the pixel point with the pixel value of 1 in the annihilation window are cleared, on the basis, the pixel point with the largest assignment is found from the rest assignments again, the pixel point with the largest assignment is determined as the vertex and recorded, the operation of annihilation window is repeated until the assignment on the pixel point with the pixel value of 1 is smaller than a given threshold value, and the vertex extraction is finished.
The first step of the step of 'vertex extraction' is to draw lines with the pixel value of 1 into a skeleton diagram by adopting line segments with the line width of 1 pixel on the basis of the form image; traversing pixel points 1 to k on the basis of a skeleton graph with k pixel points, and when a pixel point with a pixel value equal to 1 is encountered, searching for a circle around an eight-neighborhood of the pixel point in a clockwise or anticlockwise direction to obtain the change times of one pixel value; if the pixel value change times is 4, displaying that two straight lines exist in the pixel point, and determining the pixel point as an edge vertex and recording when the two straight lines have a reasonable included angle through coordinate calculation; and if the number of times of the pixel value change is 6, displaying that three straight lines exist in the pixel point, determining the pixel point as a middle vertex and recording.
The first step of the step of 'vertex extraction' is to draw a skeleton drawing by using a line segment with the line width of 1 pixel for a line with the pixel value of 1 on the basis of a 'binary drawing'; secondly, on the basis of the skeleton image, establishing a square window with the size of 5 multiplied by 5 pixels by taking a pixel point with the pixel value of 1 as a central point, if partial area of the window overflows the skeleton image, assigning all the pixel values of the pixel points of the overflow area as 0, then calculating a corner response function value R corresponding to the central point of the window by adopting a Harris algorithm, taking one percent of the maximum value of the corner response function value R as a limit value, setting the corner response function value R smaller than the limit value as zero in all the corner response function values R in the same window, and traversing the whole skeleton image by repeating the operations; and next, establishing a square window with the size of 3x 3 pixels by taking the pixel point with the pixel value equal to 1 and the angular point response function value R larger than zero as a central point, recording the point as a vertex if the R value of the central point of the window is the maximum value in the window, and traversing the whole skeleton map by repeating the operations.
The first step of reconstructing the cell element is that x vertexes are selected as datum points at random at the central position of the binary image, the vertex closest to the datum point is found as an adjacent point at each datum point, the distance between the datum point and the adjacent point is calculated, three records with the shortest distance are reserved, and then the average value of the distances of the 3x records is calculated as the side length A of the cell element; secondly, dividing the obtained vertexes into two categories, namely active connecting points and passive connecting points; namely: dividing the skeleton map into two regions, wherein the region with the width of 1A-2A on the periphery of the skeleton map is an edge region, and the region surrounded by the edge region is a central region; the vertex positioned in the central area is an active connection point, and the vertex positioned in the edge area is a passive connection point; the third step is to connect all active connection points with the three closest vertexes by line segments, namely: each active connection point must only reserve three shortest line segments, and after the connection is completed, a cell reconfiguration image is formed.
The step of reconstructing the cell comprises edge expansion and vertex connection; the step of 'edge expansion' is based on the form image, the width of at least 1 pixel is outwards expanded at the outermost edge of four edges of the form image to form an expansion area, and the pixel values of all pixel points in the expansion area are all set to be 1, so that an expansion image is obtained; in the step of 'vertex connection', an extended image is taken as an object, the extended image is traversed from left to right and from top to bottom, when a pixel point with a pixel value of 0 is encountered, a Moore neighborhood tracking algorithm is adopted to find and record the vertex of the same cell and the connection sequence thereof and record the vertex and the connection sequence under the name of the cell, repeated records are deleted on the principle that at most six vertices of each cell are reserved, vertex connection is carried out according to the reserved records, and a complete cell is drawn; then setting all pixel values of all pixel points in the cell element as 1; on the basis, the pixel point with the next pixel value equal to 0 is searched, the operations are repeated, and the cell reconfiguration is completed while the traversal is completed.
A system suitable for a honeycomb structure geometric regularity image recognition method comprises a detection table, a digital camera and a computer; the digital camera is electrically connected with the computer;
the digital camera is at least one, the resolution ratio is not lower than 1080P, a telecentric lens is configured, and the installation mode is fixed or/and movable.
The detection table is a movable working platform and comprises an object placing table, a lifting device and a clamp, and the lifting device is arranged at the bottom of the object placing table; the honeycomb piece to be measured is placed on the object placing table; the lifting device comprises a vertical guide rail, an electric push rod or an electro-hydraulic push rod, and can push the object placing table to move up and down along the vertical guide rail, and adjust the height of the tested honeycomb piece so as to ensure that the upper end surface of the tested honeycomb piece is flush with the upper end surface of the clamp; the control part of the lifting device is electrically connected with the computer;
the clamp consists of four flat plates and a driving device, can be closed to the tested honeycomb piece under the action of the driving device, is locked after being closed to the tested honeycomb piece and is used for positioning and fixing the tested honeycomb piece.
When the digital camera is installed in a movable mode, a walking portal frame, a sliding rail and a moving device are additionally arranged in the system;
the digital camera is arranged on a beam of the walking portal frame and can transversely move along the beam under the driving of the moving device;
the walking portal frame can move longitudinally along the slide rail under the driving of the moving device, and the digital camera and the walking portal frame are controlled by a computer.
Compared with the prior art, the method and the system thereof of the invention are scientific and reasonable. Simple and easy operation, high precision, high working efficiency and the like.
Drawings
FIG. 1 is a block flow diagram of the method of the present invention;
FIG. 2 is a schematic diagram of vertex fetch and cell reconstruction according to an embodiment of the method of the present invention;
FIG. 3 is a schematic diagram of an apparatus configuration according to an embodiment of the system of the present invention;
fig. 4 is a top view of fig. 3.
In the figure: 1-detection table, 2-digital camera, 3-computer, 4-object placing table, 5-lifting device, 6-clamp, 7-walking portal frame, 8-slide rail and 9-moving device.
Detailed Description
The invention is further illustrated by the following figures and examples:
method, with reference to figures 1, 2:
a method for cellular structure geometry regularity image recognition, the method comprising the steps of: acquiring an image, processing the image, extracting a vertex, reconstructing a cell element and evaluating quality; the step of acquiring the image comprises shooting the image and reading the image by a computer; the step of 'vertex extraction' is to find and record the vertex of a cell on the basis of 'image processing'; in the step of reconstructing the cell element, the extracted vertexes are connected according to the mapping relation between the cell element and the vertex to obtain a cell element reconstruction picture;
setting a step of binaryzation between the step of image processing and the step of vertex extraction, wherein the step of binaryzation is to set the pixel value of a background in an image to be 0 and set the pixel value of a honeycomb skeleton in the image to be 1 to form a binaryzation image;
the step of 'quality evaluation' is based on the cell reconstruction graph, calculates the angular deviation values and the average values thereof, the line deviation values and the average values thereof of all the cells, and compares the angular deviation values and the average values with the set tolerance band to judge whether the cell is qualified.
In the step of 'vertex extraction', based on a 'binary image', determining wall thickness, assigning pixels, determining cell edge length and annihilating pixels are sequentially executed;
the step of determining the wall thickness is to set a square window with a side length capable of changing from small to large, and after a window with a certain side length traverses the morphological image, if the minimum value of the number of pixels of which the pixel value in the window is 0 is non-zero, define the side length of the square window as the wall thickness L;
the step of pixel assignment is to set an assignment window with a square side length of L, traverse the pixel point with the pixel value of 1 in the morphological image by the assignment window, and then assign the sum of the pixel value of 1 in the window to the pixel point at the center point of the assignment window;
the step of determining the side length of the cell element is to find the pixel point with the largest assignment after completing the pixel assignment, establish a square area by taking the point as the center, calculate the difference absolute value Z of the assignment of each pixel point on four boundaries of the area and the assignment of the center point, traverse the whole morphological image, and record the smallest difference absolute value Z and the coordinate thereof in the round; establishing a new square area by E-E +2 pixels, repeating the process until the minimum difference absolute value Z has an obvious trend of increasing reversely, taking out the pixel point coordinate of the minimum difference absolute value Z and the coordinate of the corresponding area center point at the moment, and solving to obtain the side length A of the honeycomb cell element through the coordinate values of the pixel point coordinate and the corresponding area center point coordinate;
in the step of pixel annihilation, a pixel point with the largest assignment is found, the pixel point is determined as a vertex and recorded, then an annihilation window with a square shape with the side length A of the honeycomb cell as the side length is established with the pixel point as the center, all assignments on the pixel point with the pixel value of 1 in the annihilation window are cleared, on the basis, the pixel point with the largest assignment is found from the rest assignments again, the pixel point with the largest assignment is determined as the vertex and recorded, the operation of annihilation window is repeated until the assignment on the pixel point with the pixel value of 1 is smaller than a given threshold value, and the vertex extraction is finished.
The first step of the step of 'vertex extraction' is to draw lines with the pixel value of 1 into a skeleton diagram by adopting line segments with the line width of 1 pixel on the basis of the form image; traversing pixel points 1 to k on the basis of a skeleton graph with k pixel points, and when a pixel point with a pixel value equal to 1 is encountered, searching for a circle around an eight-neighborhood of the pixel point in a clockwise or anticlockwise direction to obtain the change times of one pixel value; if the pixel value change times is 4, displaying that two straight lines exist in the pixel point, and determining the pixel point as an edge vertex and recording when the two straight lines have a reasonable included angle through coordinate calculation; and if the number of times of the pixel value change is 6, displaying that three straight lines exist in the pixel point, determining the pixel point as a middle vertex and recording.
The first step of the step of 'vertex extraction' is to draw a skeleton drawing by using a line segment with the line width of 1 pixel for a line with the pixel value of 1 on the basis of a 'binary drawing'; secondly, on the basis of the skeleton image, establishing a square window with the size of 5 multiplied by 5 pixels by taking a pixel point with the pixel value of 1 as a central point, if partial area of the window overflows the skeleton image, assigning all the pixel values of the pixel points of the overflow area as 0, then calculating a corner response function value R corresponding to the central point of the window by adopting a Harris algorithm, taking one percent of the maximum value of the corner response function value R as a limit value, setting the corner response function value R smaller than the limit value as zero in all the corner response function values R in the same window, and traversing the whole skeleton image by repeating the operations; and next, establishing a square window with the size of 3x 3 pixels by taking the pixel point with the pixel value equal to 1 and the angular point response function value R larger than zero as a central point, recording the point as a vertex if the R value of the central point of the window is the maximum value in the window, and traversing the whole skeleton map by repeating the operations.
The first step of reconstructing the cell element is that x vertexes are selected as datum points at random at the central position of the binary image, the vertex closest to the datum point is found as an adjacent point at each datum point, the distance between the datum point and the adjacent point is calculated, three records with the shortest distance are reserved, and then the average value of the distances of the 3x records is calculated as the side length A of the cell element; secondly, dividing the obtained vertexes into two categories, namely active connecting points and passive connecting points; namely: dividing the skeleton map into two regions, wherein the region with the width of 1A-2A on the periphery of the skeleton map is an edge region, and the region surrounded by the edge region is a central region; the vertex positioned in the central area is an active connection point, and the vertex positioned in the edge area is a passive connection point; the third step is to connect all active connection points with the three closest vertexes by line segments, namely: each active connection point must only reserve three shortest line segments, and after the connection is completed, a cell reconfiguration image is formed.
The step of reconstructing the cell comprises edge expansion and vertex connection; the step of 'edge expansion' is based on the form image, the width of at least 1 pixel is outwards expanded at the outermost edge of four edges of the form image to form an expansion area, and the pixel values of all pixel points in the expansion area are all set to be 1, so that an expansion image is obtained; in the step of 'vertex connection', an extended image is taken as an object, the extended image is traversed from left to right and from top to bottom, when a pixel point with a pixel value of 0 is encountered, a Moore neighborhood tracking algorithm is adopted to find and record the vertex of the same cell and the connection sequence thereof and record the vertex and the connection sequence under the name of the cell, repeated records are deleted on the principle that at most six vertices of each cell are reserved, vertex connection is carried out according to the reserved records, and a complete cell is drawn; then setting all pixel values of all pixel points in the cell element as 1; on the basis, the pixel point with the next pixel value equal to 0 is searched, the operations are repeated, and the cell reconfiguration is completed while the traversal is completed.
System, with reference to figures 3, 4:
a system suitable for a honeycomb structure geometric regularity image recognition method comprises a detection table 1, a digital camera 2 and a computer 3; the digital camera 2 is electrically connected with the computer 3;
the digital camera 2 is at least one, the resolution ratio is not lower than 1080P, a telecentric lens is configured, and the installation mode is fixed or/and movable.
The detection table 1 is a movable working platform and comprises an object placing table 4, a lifting device 5 and a clamp 6, wherein the lifting device 5 is arranged at the bottom of the object placing table (4); the honeycomb piece to be measured is placed on the object placing table 4; the lifting device 5 comprises a vertical guide rail, an electric push rod or an electro-hydraulic push rod, and can push the object placing table 4 to move up and down along the vertical guide rail, and adjust the height of the tested honeycomb piece so as to ensure that the upper end surface of the tested honeycomb piece is flush with the upper end surface of the clamp 6; the control part of the lifting device 5 is electrically connected with the computer 3;
the clamp 6 consists of four flat plates and a driving device, can be closed to the tested honeycomb piece under the action of the driving device, is locked after being closed to the tested honeycomb piece and is used for positioning and fixing the tested honeycomb piece.
When the digital camera 2 is installed in a movable mode, a walking portal frame 7, a sliding rail 8 and a moving device 9 are additionally arranged in the system;
the digital camera 2 is arranged on a beam of the walking portal frame 7 and can transversely move along the beam under the driving of the moving device 9;
the walking portal frame 7 can move longitudinally along the slide rail 8 under the driving of the moving device 9, and the movement of the digital camera 2 and the walking portal frame 7 is controlled by the computer 3.
Method example 1:
a method for cellular structure geometry regularity image recognition, the method comprising the steps of: acquiring an image, processing the image, extracting a vertex, reconstructing a cell element and evaluating quality; the step of acquiring the image comprises shooting the image and reading the image by a computer; the step of 'vertex extraction' is to find and record the vertex of a cell on the basis of 'image processing'; in the step of reconstructing the cell element, the extracted vertexes are connected according to the mapping relation between the cell element and the vertex to obtain a cell element reconstruction picture;
setting a step of binaryzation between the step of image processing and the step of vertex extraction, wherein the step of binaryzation is to set the pixel value of a background in an image to be 0 and set the pixel value of a honeycomb skeleton in the image to be 1 to form a binaryzation image;
the step of 'quality evaluation' is based on the cell reconstruction graph, calculates the angular deviation values and the average values thereof, the line deviation values and the average values thereof of all the cells, and compares the angular deviation values and the average values with the set tolerance band to judge whether the cell is qualified.
Method example 2:
essentially the same as in "method example 1" except that: in the step of 'vertex extraction', based on a 'binary image', determining wall thickness, assigning pixels, determining cell edge length and annihilating pixels are sequentially executed;
the step of determining the wall thickness is to set a square window with a side length capable of changing from small to large, and after a window with a certain side length traverses the morphological image, if the minimum value of the number of pixels of which the pixel value in the window is 0 is non-zero, define the side length of the square window as the wall thickness L;
the step of pixel assignment is to set an assignment window with a square side length of L, traverse the pixel point with the pixel value of 1 in the morphological image by the assignment window, and then assign the sum of the pixel value of 1 in the window to the pixel point at the center point of the assignment window;
the step of determining the side length of the cell element is to find the pixel point with the largest assignment after completing the pixel assignment, establish a square area by taking the point as the center, calculate the difference absolute value Z of the assignment of each pixel point on four boundaries of the area and the assignment of the center point, traverse the whole morphological image, and record the smallest difference absolute value Z and the coordinate thereof in the round; establishing a new square area by E-E +2 pixels, repeating the process until the minimum difference absolute value Z has an obvious trend of increasing reversely, taking out the pixel point coordinate of the minimum difference absolute value Z and the coordinate of the corresponding area center point at the moment, and solving to obtain the side length A of the honeycomb cell element through the coordinate values of the pixel point coordinate and the corresponding area center point coordinate;
in the step of pixel annihilation, a pixel point with the largest assignment is found, the pixel point is determined as a vertex and recorded, then an annihilation window with a square shape with the side length A of the honeycomb cell as the side length is established with the pixel point as the center, all assignments on the pixel point with the pixel value of 1 in the annihilation window are cleared, on the basis, the pixel point with the largest assignment is found from the rest assignments again, the pixel point with the largest assignment is determined as the vertex and recorded, the operation of annihilation window is repeated until the assignment on the pixel point with the pixel value of 1 is smaller than a given threshold value, and the vertex extraction is finished.
Method example 3:
essentially the same as in "method example 1" except that: the first step of the step of 'vertex extraction' is to draw lines with the pixel value of 1 into a skeleton diagram by adopting line segments with the line width of 1 pixel on the basis of the form image; traversing pixel points 1 to k on the basis of a skeleton graph with k pixel points, and when a pixel point with a pixel value equal to 1 is encountered, searching for a circle around an eight-neighborhood of the pixel point in a clockwise or anticlockwise direction to obtain the change times of one pixel value; if the pixel value change times is 4, displaying that two straight lines exist in the pixel point, and determining the pixel point as an edge vertex and recording when the two straight lines have a reasonable included angle through coordinate calculation; and if the number of times of the pixel value change is 6, displaying that three straight lines exist in the pixel point, determining the pixel point as a middle vertex and recording.
Method example 4:
essentially the same as in "method example 1" except that: the first step of the step of 'vertex extraction' is to draw a skeleton drawing by using a line segment with the line width of 1 pixel for a line with the pixel value of 1 on the basis of a 'binary drawing'; secondly, on the basis of the skeleton image, establishing a square window with the size of 5 multiplied by 5 pixels by taking a pixel point with the pixel value of 1 as a central point, if partial area of the window overflows the skeleton image, assigning all the pixel values of the pixel points of the overflow area as 0, then calculating a corner response function value R corresponding to the central point of the window by adopting a Harris algorithm, taking one percent of the maximum value of the corner response function value R as a limit value, setting the corner response function value R smaller than the limit value as zero in all the corner response function values R in the same window, and traversing the whole skeleton image by repeating the operations; and next, establishing a square window with the size of 3x 3 pixels by taking the pixel point with the pixel value equal to 1 and the angular point response function value R larger than zero as a central point, recording the point as a vertex if the R value of the central point of the window is the maximum value in the window, and traversing the whole skeleton map by repeating the operations.
Method examples 5 to 8:
essentially the same as in "method examples 1 to 4", respectively, except that: the first step of reconstructing the cell element is that x vertexes are selected as datum points at random at the central position of the binary image, the vertex closest to the datum point is found as an adjacent point at each datum point, the distance between the datum point and the adjacent point is calculated, three records with the shortest distance are reserved, and then the average value of the distances of the 3x records is calculated as the side length A of the cell element; secondly, dividing the obtained vertexes into two categories, namely active connecting points and passive connecting points; namely: dividing the skeleton map into two regions, wherein the region with the width of 1A-2A on the periphery of the skeleton map is an edge region, and the region surrounded by the edge region is a central region; the vertex positioned in the central area is an active connection point, and the vertex positioned in the edge area is a passive connection point; the third step is to connect all active connection points with the three closest vertexes by line segments, namely: each active connection point must only reserve three shortest line segments, and after the connection is completed, a cell reconfiguration image is formed.
Method examples 9 to 12:
essentially the same as in "method examples 1 to 4", respectively, except that: the step of reconstructing the cell comprises edge expansion and vertex connection; the step of 'edge expansion' is based on the form image, the width of at least 1 pixel is outwards expanded at the outermost edge of four edges of the form image to form an expansion area, and the pixel values of all pixel points in the expansion area are all set to be 1, so that an expansion image is obtained; in the step of 'vertex connection', an extended image is taken as an object, the extended image is traversed from left to right and from top to bottom, when a pixel point with a pixel value of 0 is encountered, a Moore neighborhood tracking algorithm is adopted to find and record the vertex of the same cell and the connection sequence thereof and record the vertex and the connection sequence under the name of the cell, repeated records are deleted on the principle that at most six vertices of each cell are reserved, vertex connection is carried out according to the reserved records, and a complete cell is drawn; then setting all pixel values of all pixel points in the cell element as 1; on the basis, the pixel point with the next pixel value equal to 0 is searched, the operations are repeated, and the cell reconfiguration is completed while the traversal is completed.
System example 1:
a system suitable for a honeycomb structure geometric regularity image recognition method comprises a detection table 1, a digital camera 2 and a computer 3; the digital camera 2 is electrically connected with the computer 3;
the digital camera 2 is at least one, the resolution ratio is not lower than 1080P, a telecentric lens is configured, and the installation mode is fixed or/and movable.
System example 2:
essentially the same as in system example 1, except that: the detection table 1 is a movable working platform and comprises an object placing table 4, a lifting device 5 and a clamp 6, wherein the lifting device 5 is arranged at the bottom of the object placing table (4); the honeycomb piece to be measured is placed on the object placing table 4; the lifting device 5 comprises a vertical guide rail, an electric push rod or an electro-hydraulic push rod, and can push the object placing table 4 to move up and down along the vertical guide rail, and adjust the height of the tested honeycomb piece so as to ensure that the upper end surface of the tested honeycomb piece is flush with the upper end surface of the clamp 6; the control part of the lifting device 5 is electrically connected with the computer 3;
the clamp 6 consists of four flat plates and a driving device, can be closed to the tested honeycomb piece under the action of the driving device, is locked after being closed to the tested honeycomb piece and is used for positioning and fixing the tested honeycomb piece.
System examples 3, 4:
essentially the same as in system examples 1, 2, respectively, except that: when the digital camera 2 is installed in a movable mode, a walking portal frame 7, a sliding rail 8 and a moving device 9 are additionally arranged in the system;
the digital camera 2 is arranged on a beam of the walking portal frame 7 and can transversely move along the beam under the driving of the moving device 9;
the walking portal frame 7 can move longitudinally along the slide rail 8 under the driving of the moving device 9, and the movement of the digital camera 2 and the walking portal frame 7 is controlled by the computer 3.
The peak extraction and cell reconstruction results of one embodiment of the method of the present invention are shown in fig. 2, and the calculation results are: the maximum value of the interior angle deviation is 14.42, the average value of the interior angle deviation is 2.62, the standard deviation of the interior angle deviation is 2.42, and the product quality is judged to be qualified when the interior angle deviation is within the range of the set value.
Claims (9)
1. A method for cellular structure geometry regularity image recognition, the method comprising the steps of: acquiring an image, processing the image, extracting a vertex, reconstructing a cell element and evaluating quality; the step of acquiring the image comprises shooting the image and reading the image by a computer; the step of 'vertex extraction' is to find and record the vertex of a cell on the basis of 'image processing'; in the step of reconstructing the cell element, the extracted vertexes are connected according to the mapping relation between the cell element and the vertex to obtain a cell element reconstruction picture; the method is characterized in that:
setting a step of binaryzation between the step of image processing and the step of vertex extraction, wherein the step of binaryzation is to set the pixel value of a background in an image to be 0 and set the pixel value of a honeycomb skeleton in the image to be 1 to form a binaryzation image;
the step of 'quality evaluation' is based on the cell reconstruction graph, calculates the angular deviation values and the average values thereof, the line deviation values and the average values thereof of all the cells, and compares the angular deviation values and the average values with the set tolerance band to judge whether the cell is qualified.
2. The method of claim 1, wherein: in the step of 'vertex extraction', based on a 'binary image', determining wall thickness, assigning pixels, determining cell edge length and annihilating pixels are sequentially executed;
the step of determining the wall thickness is to set a square window with a side length capable of changing from small to large, and after a window with a certain side length traverses the morphological image, if the minimum value of the number of pixels of which the pixel value in the window is 0 is non-zero, define the side length of the square window as the wall thickness L;
the step of pixel assignment is to set an assignment window with a square side length of L, traverse the pixel point with the pixel value of 1 in the morphological image by the assignment window, and then assign the sum of the pixel value of 1 in the window to the pixel point at the center point of the assignment window;
the step of determining the side length of the cell element is to find the pixel point with the largest assignment after completing the pixel assignment, establish a square area by taking the point as the center, calculate the difference absolute value Z of the assignment of each pixel point on four boundaries of the area and the assignment of the center point, traverse the whole morphological image, and record the smallest difference absolute value Z and the coordinate thereof in the round; establishing a new square area by E-E +2 pixels, repeating the process until the minimum difference absolute value Z has an obvious trend of increasing reversely, taking out the pixel point coordinate of the minimum difference absolute value Z and the coordinate of the corresponding area center point at the moment, and solving to obtain the side length A of the honeycomb cell element through the coordinate values of the pixel point coordinate and the corresponding area center point coordinate;
in the step of pixel annihilation, a pixel point with the largest assignment is found, the pixel point is determined as a vertex and recorded, then an annihilation window with a square shape with the side length A of the honeycomb cell as the side length is established with the pixel point as the center, all assignments on the pixel point with the pixel value of 1 in the annihilation window are cleared, on the basis, the pixel point with the largest assignment is found from the rest assignments again, the pixel point with the largest assignment is determined as the vertex and recorded, the operation of annihilation window is repeated until the assignment on the pixel point with the pixel value of 1 is smaller than a given threshold value, and the vertex extraction is finished.
3. The method of claim 1, wherein: the first step of the step of 'vertex extraction' is to draw lines with the pixel value of 1 into a skeleton diagram by adopting line segments with the line width of 1 pixel on the basis of the form image; traversing pixel points 1 to k on the basis of a skeleton graph with k pixel points, and when a pixel point with a pixel value equal to 1 is encountered, searching for a circle around an eight-neighborhood of the pixel point in a clockwise or anticlockwise direction to obtain the change times of one pixel value; if the pixel value change times is 4, displaying that two straight lines exist in the pixel point, and determining the pixel point as an edge vertex and recording when the two straight lines have a reasonable included angle through coordinate calculation; and if the number of times of the pixel value change is 6, displaying that three straight lines exist in the pixel point, determining the pixel point as a middle vertex and recording.
4. The method of claim 1, wherein: the first step of the step of 'vertex extraction' is to draw a skeleton drawing by using a line segment with the line width of 1 pixel for a line with the pixel value of 1 on the basis of a 'binary drawing'; secondly, on the basis of the skeleton image, establishing a square window with the size of 5 multiplied by 5 pixels by taking a pixel point with the pixel value of 1 as a central point, if partial area of the window overflows the skeleton image, assigning all the pixel values of the pixel points of the overflow area as 0, then calculating a corner response function value R corresponding to the central point of the window by adopting a Harris algorithm, taking one percent of the maximum value of the corner response function value R as a limit value, setting the corner response function value R smaller than the limit value as zero in all the corner response function values R in the same window, and traversing the whole skeleton image by repeating the operations; and next, establishing a square window with the size of 3x 3 pixels by taking the pixel point with the pixel value equal to 1 and the angular point response function value R larger than zero as a central point, recording the point as a vertex if the R value of the central point of the window is the maximum value in the window, and traversing the whole skeleton map by repeating the operations.
5. The method according to any one of claims 1 to 4, characterized in that: the first step of reconstructing the cell element is that x vertexes are selected as datum points at random at the central position of the binary image, the vertex closest to the datum point is found as an adjacent point at each datum point, the distance between the datum point and the adjacent point is calculated, three records with the shortest distance are reserved, and then the average value of the distances of the 3x records is calculated as the side length A of the cell element; secondly, dividing the obtained vertexes into two categories, namely active connecting points and passive connecting points; namely: dividing the skeleton map into two regions, wherein the region with the width of 1A-2A on the periphery of the skeleton map is an edge region, and the region surrounded by the edge region is a central region; the vertex positioned in the central area is an active connection point, and the vertex positioned in the edge area is a passive connection point; the third step is to connect all active connection points with the three closest vertexes by line segments, namely: each active connection point must only reserve three shortest line segments, and after the connection is completed, a cell reconfiguration image is formed.
6. The method of any one of claims 1-4, wherein: the step of reconstructing the cell comprises edge expansion and vertex connection; the step of 'edge expansion' is based on the form image, the width of at least 1 pixel is outwards expanded at the outermost edge of four edges of the form image to form an expansion area, and the pixel values of all pixel points in the expansion area are all set to be 1, so that an expansion image is obtained; in the step of 'vertex connection', an extended image is taken as an object, the extended image is traversed from left to right and from top to bottom, when a pixel point with a pixel value of 0 is encountered, a Moore neighborhood tracking algorithm is adopted to find and record the vertex of the same cell and the connection sequence thereof and record the vertex and the connection sequence under the name of the cell, repeated records are deleted on the principle that at most six vertices of each cell are reserved, vertex connection is carried out according to the reserved records, and a complete cell is drawn; then setting all pixel values of all pixel points in the cell element as 1; on the basis, the pixel point with the next pixel value equal to 0 is searched, the operations are repeated, and the cell reconfiguration is completed while the traversal is completed.
7. A system suitable for use in the method of any one of claims 1 to 6, the system comprising an inspection station (1), a digital camera (2) and a computer (3); the digital camera (2) is electrically connected with the computer (3); the method is characterized in that:
the digital camera (2) is at least one, the resolution ratio is not lower than 1080P, a telecentric lens is configured, and the installation mode is fixed or/and movable.
8. The system of claim 7, wherein: the detection table (1) is a movable working platform and comprises an object placing table (4), a lifting device (5) and a clamp (6), wherein the lifting device (5) is arranged at the bottom of the object placing table (4); the honeycomb piece to be measured is placed on the object placing table (4); the lifting device (5) comprises a vertical guide rail, an electric push rod or an electro-hydraulic push rod, and can push the object placing table (4) to move up and down along the vertical guide rail, and the height of the tested honeycomb piece is adjusted to ensure that the upper end surface of the tested honeycomb piece is flush with the upper end surface of the clamp (6); the control part of the lifting device (5) is electrically connected with the computer (3);
the clamp (6) consists of four flat plates and a driving device, can be closed to the tested honeycomb piece under the action of the driving device and is locked after being closed to the tested honeycomb piece, and is used for positioning and fixing the tested honeycomb piece.
9. The system according to claim 7 or 8, wherein: when the digital camera (2) is installed in a movable mode, a walking portal frame (7), a sliding rail (8) and a moving device (9) are additionally arranged in the system;
the digital camera (2) is arranged on a beam of the walking portal frame (7) and can transversely move along the beam under the driving of the moving device (9);
the walking type portal frame (7) can move longitudinally along the sliding rail (8) under the driving of the moving device (9), and the movement of the digital camera (2) and the walking type portal frame (7) is controlled by the computer (3).
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010388353.9A CN111583239B (en) | 2020-05-09 | 2020-05-09 | Honeycomb structure geometric regularity image recognition method and system |
| PCT/CN2020/109726 WO2021227285A1 (en) | 2020-05-09 | 2020-08-18 | Image recognition method for geometrical regularity of cellular structure, and system |
| US17/422,720 US11893768B2 (en) | 2020-05-09 | 2020-08-18 | Method and system for recognizing geometric regularity image of honeycomb structure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010388353.9A CN111583239B (en) | 2020-05-09 | 2020-05-09 | Honeycomb structure geometric regularity image recognition method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111583239A true CN111583239A (en) | 2020-08-25 |
| CN111583239B CN111583239B (en) | 2021-03-30 |
Family
ID=72112205
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010388353.9A Active CN111583239B (en) | 2020-05-09 | 2020-05-09 | Honeycomb structure geometric regularity image recognition method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111583239B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1975759A (en) * | 2006-12-15 | 2007-06-06 | 中山大学 | Human face identifying method based on structural principal element analysis |
| KR100743936B1 (en) * | 2006-08-11 | 2007-07-30 | 박홍욱 | Reinforced materials with honeycomb type for earthquake-resistance, repair, rehabilitation of concrete structures |
| CN103793712A (en) * | 2014-02-19 | 2014-05-14 | 华中科技大学 | Image recognition method and system based on edge geometric features |
| CN104369188A (en) * | 2014-11-20 | 2015-02-25 | 中国计量学院 | Workpiece grabbing device and method based on machine vision and ultrasonic transducer |
| CN105091744A (en) * | 2015-05-07 | 2015-11-25 | 中国科学院自动化研究所 | Pose detection apparatus and method based on visual sensor and laser range finder |
| CN106030614A (en) * | 2014-04-22 | 2016-10-12 | 史內普艾德有限公司 | System and method for controlling a camera based on processing an image captured by other camera |
| CN110658196A (en) * | 2018-06-29 | 2020-01-07 | 上海微电子装备（集团）股份有限公司 | Defect detection device and defect detection method |
| CN210253172U (en) * | 2019-07-19 | 2020-04-07 | 东莞恩浩电子有限公司 | Automatic system for visual inspection |
-
2020
- 2020-05-09 CN CN202010388353.9A patent/CN111583239B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100743936B1 (en) * | 2006-08-11 | 2007-07-30 | 박홍욱 | Reinforced materials with honeycomb type for earthquake-resistance, repair, rehabilitation of concrete structures |
| CN1975759A (en) * | 2006-12-15 | 2007-06-06 | 中山大学 | Human face identifying method based on structural principal element analysis |
| CN103793712A (en) * | 2014-02-19 | 2014-05-14 | 华中科技大学 | Image recognition method and system based on edge geometric features |
| CN106030614A (en) * | 2014-04-22 | 2016-10-12 | 史內普艾德有限公司 | System and method for controlling a camera based on processing an image captured by other camera |
| CN104369188A (en) * | 2014-11-20 | 2015-02-25 | 中国计量学院 | Workpiece grabbing device and method based on machine vision and ultrasonic transducer |
| CN105091744A (en) * | 2015-05-07 | 2015-11-25 | 中国科学院自动化研究所 | Pose detection apparatus and method based on visual sensor and laser range finder |
| CN110658196A (en) * | 2018-06-29 | 2020-01-07 | 上海微电子装备（集团）股份有限公司 | Defect detection device and defect detection method |
| CN210253172U (en) * | 2019-07-19 | 2020-04-07 | 东莞恩浩电子有限公司 | Automatic system for visual inspection |
Non-Patent Citations (1)
| Title |
|---|
| CUI CAN ET AL.: "Branch point algorithm for structural irregularity determination of honeycomb", 《COMPOSITES PART B： ENGINEERING》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111583239B (en) | 2021-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109886939B (en) | Bridge crack detection method based on tensor voting | |
| CN106441087B (en) | A kind of more size Multi-parameter Measurement Methods of workpiece based on image procossing | |
| CN107392849B (en) | Target identification and positioning method based on image subdivision | |
| CN107358628B (en) | Linear array image processing method based on target | |
| CN110986877B (en) | Railway engineering clearance detection method based on high-precision vehicle-mounted laser mobile measurement system | |
| CN111524134B (en) | Method and device for detecting regularity of honeycomb products on production line | |
| CN108592816A (en) | A kind of three-dimensional measuring apparatus and method for large scale surface | |
| CN111583239B (en) | Honeycomb structure geometric regularity image recognition method and system | |
| CN111583243B (en) | Method for reconstructing adjacent point cell for detecting cellular regularity | |
| CN111583235B (en) | Branch point identification vertex extraction method and system for detecting cellular regularity | |
| CN111583237B (en) | Neighborhood window recursion cell reconstruction method for detecting cellular regularity | |
| WO2021227285A1 (en) | Image recognition method for geometrical regularity of cellular structure, and system | |
| WO2021227288A1 (en) | Method for evaluating geometrical morphology of honeycomb product, and detection and evaluation system | |
| CN111583241A (en) | Mobile detection method and device for regularity of honeycomb product with super-large area | |
| CN111524133B (en) | Window statistical vertex extraction method and system for honeycomb regularity detection | |
| CN115790440A (en) | Profile tolerance measuring method and measuring system based on 3D scanning | |
| CN111583234B (en) | Geometric form evaluation method, detection and evaluation system for honeycomb products | |
| CN115222893A (en) | Three-dimensional reconstruction splicing method for large-size components based on structured light measurement | |
| CN110411384B (en) | Method for detecting surface roughness of internal rod piece of three-dimensional printing lattice structure | |
| CN112258585B (en) | Calibration field design and image processing method for image distortion partition solution | |
| CN111536890A (en) | Harris vertex extraction method and system for detecting cellular regularity | |
| CN106959101A (en) | A kind of indoor orientation method based on optical flow method | |
| WO2021227286A1 (en) | Mobile testing method and apparatus for regularity of honeycomb product with ultra-large area | |
| CN111583242B (en) | Morphological vertex extraction method and detection device for honeycomb regularity detection | |
| CN117848238B (en) | Rail transit station actual limit measurement method based on laser point cloud |
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 | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20200825 Assignee: HUNAN MINGXIANG TECHNOLOGY DEVELOPMENT Co.,Ltd. Assignor: CENTRAL SOUTH University Contract record no.: X2023980038143 Denomination of invention: Image recognition method and system for geometric regularity of honeycomb structures Granted publication date: 20210330 License type: Common License Record date: 20230720 |