CN117893612A - Visual positioning method for medium plate lifting appliance clamping process - Google Patents
Visual positioning method for medium plate lifting appliance clamping process Download PDFInfo
- Publication number
- CN117893612A CN117893612A CN202410295596.6A CN202410295596A CN117893612A CN 117893612 A CN117893612 A CN 117893612A CN 202410295596 A CN202410295596 A CN 202410295596A CN 117893612 A CN117893612 A CN 117893612A
- Authority
- CN
- China
- Prior art keywords
- suspected
- contour
- value
- corner
- medium plate
- 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
- 238000000034 method Methods 0.000 title claims abstract description 73
- 230000000007 visual effect Effects 0.000 title claims abstract description 34
- 238000001514 detection method Methods 0.000 claims abstract description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000003708 edge detection Methods 0.000 abstract description 3
- 238000010801 machine learning Methods 0.000 abstract description 3
- 239000002893 slag Substances 0.000 description 11
- 229910000831 Steel Inorganic materials 0.000 description 10
- 239000010959 steel Substances 0.000 description 10
- 238000005096 rolling process Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000004519 manufacturing process Methods 0.000 description 4
- 238000005098 hot rolling Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000003825 pressing Methods 0.000 description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000005452 bending Methods 0.000 description 2
- 230000006378 damage Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000003723 Smelting Methods 0.000 description 1
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000010791 quenching Methods 0.000 description 1
- 230000000171 quenching effect Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
- G06T7/44—Analysis of texture based on statistical description of texture using image operators, e.g. filters, edge density metrics or local histograms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/761—Proximity, similarity or dissimilarity measures
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image data processing, in particular to a visual positioning method for a medium plate lifting appliance clamping process, which comprises the following steps: acquiring an image of a medium plate; acquiring each contour by using an edge detection algorithm and a corner detection algorithm; distinguishing two sides of the contour according to the normal direction, and constructing two-side long run factor difference coefficients at each pixel point and two-side circumscribed texture contrast ratios of the contour according to texture features at the two sides of the contour at each pixel point; further constructing a texture contrast sequence and local texture similarity of each window; obtaining the contrast fluctuation coefficients of the circumscribed textures at the two sides of each contour; acquiring suspected corner contours, constructing the corner saliency score of each suspected corner contour relative to the rest of the suspected corner contours, and constructing the corner contour saliency of each suspected corner contour based on the corner saliency score; and marking the position of the medium plate by adopting a machine learning algorithm according to the edge contour saliency. The invention can improve the accuracy of positioning the medium plate.
Description
Technical Field
The application relates to the technical field of image data processing, in particular to a visual positioning method for a medium plate lifting appliance clamping process.
Background
The medium plate is a plate widely used in engineering, the thickness of the medium plate is generally between 4.5mm and 25mm and is far smaller than the plane size of the medium plate, and the medium plate is widely applied in the fields of constructional engineering, machine manufacturing, container manufacturing, shipbuilding, bridge construction and the like. In the production process of the medium plate, the medium plate must be suspended by a hanger and accurately fed to a designated position. Because of the relatively large size of the sheet material, large spreaders with hooks are often used for gripping. Although the medium plate has a certain thickness, in the lifting process, the lifting appliance can still cause the problems of hanging injury, hanging bending and the like of the medium plate, thereby influencing the appearance quality of the medium plate. Therefore, in the clamping process, the medium plate needs to be accurately positioned so as to reduce damage caused by lifting to the greatest extent and ensure that the appearance quality of the medium plate meets the requirement.
The medium plates are usually manufactured by a rolling process, in which the billet is passed through a series of rolls, gradually reducing the section and obtaining the desired thickness. Different rolls and rolling parameters may leave streak marks on the surface of the medium plate. The existing method based on visual positioning has the defect that the surface stripes of the medium plate are easily identified as real boundaries by mistake, so that the positioning is inaccurate, and the medium plate is damaged.
Disclosure of Invention
In order to solve the technical problems, the invention provides a visual positioning method for a medium plate lifting appliance clamping process, which aims to solve the existing problems.
The invention discloses a visual positioning method for a medium plate lifting appliance clamping process, which adopts the following technical scheme:
One embodiment of the invention provides a visual positioning method for a medium plate lifting appliance clamping process, which comprises the following steps:
acquiring a medium plate image and acquiring an edge image of the medium plate image;
Detecting edge image corner points by using a corner point detection algorithm, and connecting lines between the corner points to obtain each contour; for each contour, acquiring the normal direction of each pixel point on the contour, distinguishing two sides of the contour according to the normal direction, and constructing two-side long run factor difference coefficients of each pixel point according to texture features of two sides of the contour of each pixel point; constructing contour two-side circumscribed texture contrast ratios at each pixel point according to the two-side long run factor difference coefficients at each pixel point; dividing each contour into windows according to the preset numerical value pixel length, and constructing a texture contrast sequence of each window according to the circumscribed texture contrast of the two sides of the contour at the pixel point in each window; constructing local texture similarity of each window according to the difference of texture comparison sequences of each window and other windows in the outline of each window; constructing two side circumscribed texture contrast fluctuation coefficients of each contour according to the local texture similarity of the windows in each contour; obtaining suspected corner contours according to the contrast fluctuation coefficients of the circumscribed textures at the two sides of each contour; constructing the edge significant fraction of each suspected edge profile relative to the other suspected edge profiles according to the gradient value sequences of each suspected edge profile and the other suspected edge profiles; constructing the edge profile saliency of each suspected edge profile according to the edge saliency fraction of each suspected edge profile relative to the other suspected edge profiles and the external-tangent texture contrast fluctuation coefficients of the two sides of the suspected edge profile;
and marking the position of the medium plate by utilizing a Boosting algorithm according to the edge profile saliency, and realizing visual positioning in the clamping process of the medium plate lifting tool.
Further, the constructing the difference coefficient of the long run factors at two sides of each pixel point according to the texture features at two sides of the outline at each pixel point includes:
For each pixel point, taking the pixel point as a tangent point to respectively make circumscribed circles on two sides of the outline, respectively calculating long run factors of gray scale run matrixes under four preset angles in each circumscribed circle region, respectively calculating average values of all long run factors in each circumscribed circle region, calculating a ratio of the average value of the circumscribed circle on the right side of the outline along the normal direction to the average value of the other circumscribed circle, taking the ratio as an index of an exponential function based on a natural constant, calculating a difference absolute value between a calculation result of the exponential function and 1, and taking the difference absolute value as a difference coefficient of the long run factors on two sides of the pixel point.
Further, the constructing the contour two-side circumscribed texture contrast ratio at each pixel point according to the two-side long run factor difference coefficient at each pixel point includes:
For each pixel point, respectively counting the gray value average value of all the pixel points in each circumscribed circle region of the pixel point, respectively calculating the difference absolute value of the gray value of each pixel point in each circumscribed circle region and the gray value average value of each circumscribed circle, marking the difference absolute value as a first difference absolute value, respectively calculating the ratio of the first difference absolute value of each pixel point in each circumscribed circle region to the total number of the pixel points in each circumscribed circle region, respectively calculating the sum value of the ratio of all the pixel points in each circumscribed circle region, obtaining the difference absolute value of the sum values of the two circumscribed circles, marking the difference absolute value as a second difference absolute value, calculating the product of the difference coefficient of the long run factors at two sides of the pixel point and the second difference absolute value, and taking the product as the contour two-side circumscribed texture contrast at the pixel point.
Further, the constructing a texture contrast sequence of each window according to the contour two-sided circumscribed texture contrast at the pixel point in each window includes:
And for each window, the two sides of the outline at all pixel points in the window are circumscribed with texture contrast to form a texture contrast sequence of the window.
Further, the constructing the local texture similarity of each window according to the difference of the texture contrast sequences of each window and the rest windows in the outline where each window is located includes:
and for each window, calculating the DTW distance between the window and the texture contrast sequences of the other windows in the outline of the window, calculating the sum value of all the DTW distances of the window, and taking the sum value as the local texture similarity of the window.
Further, the constructing the two-side circumscribed texture contrast fluctuation coefficients of each contour according to the local texture similarity of the windows in each contour includes:
For each contour, forming a local texture similarity sequence of all windows in the contour according to a window sequence, calculating the sum value of all elements in the local texture similarity sequence, marking the sum value as a first sum value, calculating the ratio of the first sum value to the number of windows in the contour, marking the ratio as a first ratio, calculating the mean value of all elements in the local texture similarity sequence, calculating the difference value of each element in the local texture similarity sequence and the mean value, calculating the variance of all elements in the local texture similarity sequence, calculating the ratio of the difference value of each element in the local texture similarity sequence to the variance, marking the ratio as a second ratio, marking the sum value of the second ratio of all elements in the local texture similarity sequence as a second sum value, and taking the product of the first ratio and the second sum value as a contour two-side circumscribed texture contrast fluctuation coefficient.
Further, the obtaining the suspected corner profile according to the two-side circumscribed texture contrast fluctuation coefficients of each profile includes:
And arranging the contours from small to large according to the fluctuation coefficients of the external-cut texture contrast ratios at the two sides, and taking the contours corresponding to the fluctuation coefficients of the external-cut texture contrast ratios at the two sides of the preset proportion as suspected corner contours.
Further, the constructing the significant score of each suspected corner contour relative to the other suspected corner contours according to the gradient value sequence of each suspected corner contour and the other suspected corner contours includes:
For each suspected corner contour, acquiring gradient values of each pixel point of the suspected corner contour, and arranging the gradient values according to the sequence of the pixel points to form a gradient value sequence of the suspected corner contour to acquire the average value of all elements in the gradient value sequence of the suspected corner contour; calculating the absolute value of the difference value of the average value of the suspected corner outline and the rest suspected corner outlines, and when the absolute value of the difference value of the suspected corner outlines is smaller than or equal to 0.1, the significant fraction of the suspected corner outlines relative to the corners of the suspected corner outlines is 1;
Calculating the product of the average value of the suspected corner profile and the rest of the suspected corner profiles, calculating the difference value between-1 and a preset value larger than zero, and calculating the sum value of-1 and the preset value larger than zero, wherein when the product of the suspected corner profiles is larger than the difference value and smaller than the sum value, the significant fraction of the suspected corner profile relative to the corners of the suspected corner profiles is 1;
Otherwise, the significant fraction of the suspected corner profile relative to the corners of the remaining suspected corner profiles is 0.1.
Further, the constructing the edge profile saliency of each suspected edge profile according to the edge saliency score of each suspected edge profile relative to other suspected edge profiles and the external-tangent texture contrast fluctuation coefficient of two sides of the suspected edge profile includes:
And calculating the sum value of the edge significance scores of the suspected edge contours relative to the other suspected edge contours, marking the sum value as a first sum value, calculating the absolute value of the difference value of the external-tangent texture contrast fluctuation coefficients of the two sides of the suspected edge contours and the other suspected edge contours, calculating the sum value of all the absolute values of the difference values of the suspected edge contours, marking the sum value as a second sum value, calculating the sum value of the second sum value and a preset value larger than zero, marking the sum value as a third sum value, and taking the ratio of the first sum value to the third sum value as the edge contour significance of the suspected edge contours.
Further, the marking of the position of the medium plate by using Boosting algorithm according to the edge profile saliency, to realize the visual positioning of the medium plate in the lifting tool clamping process, includes:
The input of the Boosting algorithm is the 4 outlines with the maximum saliency of the edge outline of the medium plate image, and the output of the Boosting algorithm is the marking position of the medium plate.
The invention has at least the following beneficial effects:
According to the method, the outline in the medium plate is identified, textures on two sides of the outline are analyzed, and the gray value and the long run factor are combined to calculate the external tangent texture contrast on two sides of the outline at the pixel points, so that the degree that the textures on two sides of the outline at each pixel point on the outline are similar to gray distribution is estimated; constructing two-side circumscribed texture contrast fluctuation coefficients of the contour according to the two-side circumscribed texture contrast of the contour at the pixel points on the contour, and estimating the probability that the contour is the true boundary contour of the medium plate, wherein the two-side circumscribed texture contrast fluctuation coefficients are used for estimating the similarity degree of textures at two sides of the whole contour; further, a suspected boundary contour set is obtained through the contrast fluctuation coefficients of the circumscribed textures at the two sides, so that the contours with higher probability can be analyzed in a targeted manner; the method comprises the steps of constructing a corner contour saliency degree based on a gradient value sequence of a suspected boundary contour and two side circumscribed texture contrast fluctuation coefficients, analyzing the saliency degree of the suspected boundary contour, facilitating extraction of a characteristic contour closest to the actual boundary of the medium plate from a suspected boundary contour set, and improving reliability of acquiring the actual boundary contour of the medium plate; the machine learning algorithm is adopted to carry out visual positioning on the medium plate, so that the defect that the surface stripes of the medium plate are easy to be mistakenly identified as the real boundaries in the existing method is overcome, and the accuracy of positioning the medium plate is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of the steps of a visual positioning method for a medium plate lifting appliance clamping process provided by the invention;
Fig. 2 is a schematic diagram of a corner profile saliency acquisition process.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a visual positioning method for the medium plate lifting appliance clamping process according to the invention, which is provided by combining the accompanying drawings and the preferred embodiment, and the specific implementation, structure, characteristics and effects thereof are described in detail below. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The invention provides a visual positioning method for a medium plate lifting appliance clamping process, which is specifically described below with reference to the accompanying drawings.
The invention provides a visual positioning method for a medium plate lifting appliance clamping process, in particular to a visual positioning method for a medium plate lifting appliance clamping process, referring to fig. 1, which comprises the following steps:
and S001, shooting the medium plate by adopting a CMOS camera, and preprocessing the shot image.
The clamping process of the medium plate lifting appliance is as follows: according to the thickness of the hooked medium plate, the pressing plate device is firstly adjusted to be at the same level, and the distance between the surface of the pressing plate and the upper surface of the hook is set to be larger than the thickness of the hooked medium plate; performing visual positioning on the medium plate, and moving the lifting appliance after determining the position; when the lifting appliance slowly falls down and the pressing plate contacts the hooked medium plate, opening the opening and closing device for controlling the hooks, and gradually opening two rows of hooks to the maximum size; lifting the lifting appliance, gradually folding the two rows of hooks and clamping the medium plate; and (3) placing the clamped medium plate at a designated position until the two rows of hooks are opened to the maximum, hanging the padlock of the opening and closing device to a locking state, keeping the two rows of hooks open, and lifting the lifting appliance. In the clamping process, the image of the medium plate is required to be shot for visual positioning, the opening and closing device is positioned at the center position below the lifting appliance, and the CMOS camera is arranged on the side surface of the opening and closing device to collect the image of the medium plate. In the process of measuring the medium plate by utilizing machine vision, the accuracy of a measurement result is closely related to the accuracy of extracting the medium plate image information, and the distortion of the medium plate image needs to be removed before the medium plate image is visually positioned. According to the embodiment, the undistort library function in the OpenCV is called, and the parameters of the calibration camera are combined to perform distortion compensation in the horizontal and vertical directions on the medium plate image, so that the distortion correction of the medium plate image is realized. The parameter acquisition and distortion correction of the calibration camera are known techniques, and the description of this embodiment is omitted. And after the distortion correction is completed, carrying out gray conversion on the medium plate image to obtain a medium plate gray image G.
Step S002, based on the texture similarity degree of the two sides of the outline, constructing the external-cut texture contrast ratio of the two sides of the outline and the external-cut texture contrast fluctuation coefficient of the two sides, further extracting the suspected corner outline, and constructing the corner outline saliency of the suspected corner outline.
In the production process of the medium plate, firstly, the billet raw material is required to be heated, then the billet is pressed into a rough shape through rough rolling equipment, and finally continuous rolling is carried out for a plurality of times through a roller way, so that the required plate thickness is obtained, and the surface smoothness of the medium plate is ensured. During hot rolling of medium plates, "steel sheet scale" formed from oxides and other impurities, also known as steel slag, is produced on the surface. The steel slag is produced because the surface of the steel sheet may react with slag, iron slag on rolls, etc. during the smelting and rolling processes to form oxidized slag skin. In the process of clamping the medium plate lifting appliance, the real boundary of the medium plate needs to be accurately identified, and the formation of steel slag can cause interference to the boundary identification. The texture features on both sides of the steel slag profile in image G are analyzed as follows.
Firstly, smoothing an image G by adopting a Gaussian filter technology to weaken pixel noise, then detecting the edge of the image by using a Canny edge detection algorithm, and marking the edge image obtained by the processing as a Line. And then carrying out corner detection on the image Line by adopting a Shi-Tomasi corner detection algorithm, connecting the corners along the edge of the image, and regarding the connecting Line between the two corners as a contour. The gaussian filtering technique, the Canny edge detection algorithm and the Shi-Tomasi corner detection algorithm are known techniques, and the description of this embodiment is omitted. The two sides of the steel slag profile have larger difference, the internal texture of the steel slag is complex, ravines are more, and the external is relatively smooth. And calculating a normal line at each pixel point in the contour image by using a least square method, wherein the direction of the normal line is tangential to and perpendicular to the contour curve. The method of least squares calculates the normal line as a well-known technique, and this embodiment will not be described in detail.
The method comprises the steps of distinguishing two sides of a contour at each pixel point by utilizing a normal direction at each pixel point, taking a pixel point po on the contour as a tangent point, respectively making circumscribed circles at two sides of the contour, setting a radius of the circle, wherein a radius of the circle is 5 in the embodiment, marking the circumscribed circle at the right side of the contour along the normal direction as a circumscribed circle C1, marking the circumscribed circle at the other side as a circumscribed circle C2, analyzing texture characteristics by calculating a gray scale run matrix, taking the circumscribed circle C1 as an example, firstly calculating a gray scale run matrix at four angles of 0 o、45o、90o、135o in the area of the circumscribed circle C1, then respectively calculating long run factors of the four run matrices, calculating an average value of the long run factors to be L (C1), wherein the long run factors reflect the texture complex characteristics in the area, and the smaller the value is, the more complex the description of texture. The calculation method of the gray scale run matrix and the long run factor is a known technique, and the embodiment will not be described in detail.
Calculating the contrast of circumscribed textures on two sides of the contour, wherein the expression is as follows:
Where denotes the contour both sides circumscribed texture contrast at the pixel point po on the contour,/> denotes the both sides long run factor difference coefficient at the pixel point po on the contour, nc1 denotes the number of pixels in the area of the circumscribed circle C1, denotes the gray value of the pixel point i in the area of the circumscribed circle C1, and μc1 denotes the gray value average of all pixels in the area of the circumscribed circle C1. Nc2 represents the number of pixels in the area of the circumscribed circle C2,/> represents the gray value of the pixel j in the area of the circumscribed circle C2, [ mu ] C2 represents the average value of the gray values of all pixels in the area of the circumscribed circle C2,/> represents an exponential function based on a natural constant, and L (C1) and L (C2) represent the average value of the long run factors in the areas of the circumscribed circle C1 and C2, respectively.
If the gully in the region of the circumscribed circle C1 is more abundant, the gray value of the pixel point in the region of the circumscribed circle C1 is more in order, and the larger the variation of the gray value is, the more the gray value is far from the average value, and the larger is; if the pixel point gray value in the area of the circumscribed circle C2 is smoother, the gray value is closer to the average value, and the gray value is smaller as the gray value is higher than ; if the texture similarity in the areas of the circumscribed circles C1 and C2 is lower, the ratio of the average values of the long run factors in the areas of the circumscribed circles C1 and C2 is far from 1, and the difference coefficient/> of the long run factors on two sides is larger; when the probability that the contour corresponding to the pixel point po belongs to the steel slag part is larger, the circumscribed texture contrast ratio/> at the two sides of the contour at the pixel point po is larger.
In addition to the influence of steel slag on boundary judgment during hot rolling, the heat treatment process may have uneven temperature distribution, which may cause non-uniformity of the sheet. Thus, if the rolling operation is not accurate enough during the rolling process, uneven surfaces of the plate may be caused, and streaks may occur. The two sides of the stripe profile have the characteristic of larger grain difference, but the stripe profile is relatively neat, and the real boundary of the medium plate is in a straight state, so that the bending degree of the profile is further analyzed.
The contrast of the external textures on both sides of the contour at each pixel point on each contour is calculated, then each contour is divided into a plurality of windows according to the length of eta pixels, and the setting implementation of eta can be selected by the user, and eta is 20 in the embodiment. The contour both sides circumscribed texture contrast at all pixel points in each window form the texture contrast sequence of the window, which is marked . According to the texture contrast sequence, constructing a two-sided circumscribed texture contrast fluctuation coefficient, wherein the expression is as follows:
Wherein denotes a two-sided circumscribed texture contrast fluctuation coefficient of the contour l, sum () denotes a summation operation,/> denotes the number of windows in the contour l,/> denotes a local texture similarity of an alpha-th window in the contour l, σ () denotes a calculation variance operation, DTW () denotes a DTW distance,/> and/> denote texture contrast sequences of the alpha-th window and the beta-th window, respectively,/> denotes a local texture similarity sequence composed of local texture similarities of all windows in the contour l in window order, and/> denotes a mean value of all elements in the sequence/> . The calculation of the DTW distance is a known technique, and the embodiment will not be described in detail.
The tortuosity of the contour has randomness, when the tortuosity of the contour l is more irregular, the difference of the texture contrast sequences of the window in the contour l is larger, when the tortuosity of the contour l is more regular, the local texture similarity change is more stable, the texture contrast sequences of the window in the contour l are more similar, the DTW distance of the texture contrast sequences of the window is smaller, so that the smaller the and the/> are, the smaller the two-side circumscribed texture contrast fluctuation coefficient/> of the contour l is, and the probability that the contour l belongs to relatively regular stripes or real boundary contours in a medium plate is larger.
After the hot rolling is finished, quenching treatment is required to be carried out on the medium plate, when the temperatures of different parts are uneven, the medium plate can shrink unevenly, thermal deformation is caused, the geometric shape and the dimensional stability of the medium plate can be affected, and the real boundary of the medium plate can be slightly deformed. The medium plate is standard rectangle, and four angles of medium plate are easier to dispel the heat for the middle part, and the deformation volume is less than other positions, therefore, this embodiment combines the contour feature near four angles to carry out visual localization.
The hot-rolled streaks in the interior of the medium plate are displayed in the image G in a right-angle state, and the identification of the right angle is easy to cause erroneous positioning, so that the identification is required based on the texture on both sides of the square outline and the characteristics of thermal deformation. According to the analysis, the smaller the value is, the more likely the contour belongs to the corner, in order to realize the targeted analysis of the contour with higher probability of belonging to the true boundary of the medium plate, the two sides of all the contours are arranged according to the sequence from small to large, the contour corresponding to the value of ω before extraction is taken as the suspected corner contour, and the value practitioner of ω can select the contour by himself, in this embodiment, ω is 20%. And (3) for each suspected corner contour, acquiring gradient values of each pixel point of the suspected corner contour, and arranging gradient value sequences forming the suspected corner contour according to the sequence of the pixel points, wherein the gradient value sequences are marked as E. Then constructing the edge profile saliency, wherein the expression is:
Wherein represents the edge profile saliency of the suspected edge profile x, i > represents the edge saliency score of the suspected edge profile x relative to the suspected edge profile y, epsilon is a value which is preset to be greater than zero, i > and i > respectively represent the external-tangent texture contrast fluctuation coefficients of the two sides of the suspected edge profile x and the suspected edge profile y, i > and i > respectively represent the gradient value sequences of the suspected edge profile x and the suspected edge profile y, i > is an average taking operation, and tau is a value which is preset to be greater than zero.
If the two suspected corner contours are parallel, the average value of the gradient value sequences of the two suspected corner contours is approximately equal, the value is smaller, and if the two suspected corner contours are perpendicular, the product of the average value of the gradient value sequences of the two suspected corner contours is approximately-1. If a certain suspected corner outline meets the two conditions, the sum of the significant fractions of the corners of the suspected corner outline relative to the rest suspected corner outlines is larger. In the above-mentioned process, the two sides of one corner are regarded as two independent suspected corner profiles, so that the sum of the significant fractions of the corners of the suspected corner profile x is higher when the likelihood that the suspected corner profile x is located at the corner is higher; and the closer the suspected corner profile x is to the two-sided circumscribed texture contrast fluctuation coefficients of the rest of the suspected corner profiles, i.e. the smaller/> , the greater the corner profile saliency/> . The greater the corner profile saliency/> , the greater the probability that the suspected corner profile x belongs to the true boundary of the medium plate. A schematic diagram of the corner profile saliency acquisition process is shown in fig. 2.
And S003, performing visual positioning on the medium plate by adopting a machine learning algorithm.
In this embodiment, a Boosting algorithm is adopted to track the target, in which the loss function is set to be a logarithmic loss function, the learning rate is set to be 0.1, the tree depth is set to be 5, and the Boosting algorithm is a known technology, and no description is given in this embodiment. The input of the algorithm is a medium plate image and the characteristic contour of the medium plate in the image, the characteristic contour is set to be 4 contours with the maximum edge contour saliency, and the output of the algorithm is the marking position of the medium plate, so that the visual positioning of the medium plate in the lifting appliance clamping process is realized.
In this specification, each embodiment is described in a progressive manner, and the same or similar parts of each embodiment are referred to each other, and each embodiment mainly describes differences from other embodiments.
The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; the technical solutions described in the foregoing embodiments are modified or some of the technical features are replaced equivalently, so that the essence of the corresponding technical solutions does not deviate from the scope of the technical solutions of the embodiments of the present application, and all the technical solutions are included in the protection scope of the present application.
Claims (10)
1. The visual positioning method for the medium plate lifting appliance clamping process is characterized by comprising the following steps of:
acquiring a medium plate image and acquiring an edge image of the medium plate image;
Detecting edge image corner points by using a corner point detection algorithm, and connecting lines between the corner points to obtain each contour; for each contour, acquiring the normal direction of each pixel point on the contour, distinguishing two sides of the contour according to the normal direction, and constructing two-side long run factor difference coefficients of each pixel point according to texture features of two sides of the contour of each pixel point; constructing contour two-side circumscribed texture contrast ratios at each pixel point according to the two-side long run factor difference coefficients at each pixel point; dividing each contour into windows according to the preset numerical value pixel length, and constructing a texture contrast sequence of each window according to the circumscribed texture contrast of the two sides of the contour at the pixel point in each window; constructing local texture similarity of each window according to the difference of texture comparison sequences of each window and other windows in the outline of each window; constructing two side circumscribed texture contrast fluctuation coefficients of each contour according to the local texture similarity of the windows in each contour; obtaining suspected corner contours according to the contrast fluctuation coefficients of the circumscribed textures at the two sides of each contour; constructing the edge significant fraction of each suspected edge profile relative to the other suspected edge profiles according to the gradient value sequences of each suspected edge profile and the other suspected edge profiles; constructing the edge profile saliency of each suspected edge profile according to the edge saliency fraction of each suspected edge profile relative to the other suspected edge profiles and the external-tangent texture contrast fluctuation coefficients of the two sides of the suspected edge profile;
and marking the position of the medium plate by utilizing a Boosting algorithm according to the edge profile saliency, and realizing visual positioning in the clamping process of the medium plate lifting tool.
2. The visual positioning method for a medium plate lifting appliance clamping process according to claim 1, wherein the constructing the two-side long run factor difference coefficient at each pixel point according to the texture features at two sides of the contour at each pixel point comprises the following steps:
For each pixel point, taking the pixel point as a tangent point to respectively make circumscribed circles on two sides of the outline, respectively calculating long run factors of gray scale run matrixes under four preset angles in each circumscribed circle region, respectively calculating average values of all long run factors in each circumscribed circle region, calculating a ratio of the average value of the circumscribed circle on the right side of the outline along the normal direction to the average value of the other circumscribed circle, taking the ratio as an index of an exponential function based on a natural constant, calculating a difference absolute value between a calculation result of the exponential function and 1, and taking the difference absolute value as a difference coefficient of the long run factors on two sides of the pixel point.
3. The visual positioning method for the medium plate lifting appliance clamping process according to claim 2, wherein the construction of the contour two-side circumscribed texture contrast ratio at each pixel point according to the two-side long run factor difference coefficient at each pixel point comprises the following steps:
For each pixel point, respectively counting the gray value average value of all the pixel points in each circumscribed circle region of the pixel point, respectively calculating the difference absolute value of the gray value of each pixel point in each circumscribed circle region and the gray value average value of each circumscribed circle, marking the difference absolute value as a first difference absolute value, respectively calculating the ratio of the first difference absolute value of each pixel point in each circumscribed circle region to the total number of the pixel points in each circumscribed circle region, respectively calculating the sum value of the ratio of all the pixel points in each circumscribed circle region, obtaining the difference absolute value of the sum values of the two circumscribed circles, marking the difference absolute value as a second difference absolute value, calculating the product of the difference coefficient of the long run factors at two sides of the pixel point and the second difference absolute value, and taking the product as the contour two-side circumscribed texture contrast at the pixel point.
4. The visual positioning method for a medium plate lifting appliance clamping process according to claim 1, wherein the construction of the texture contrast sequence of each window according to the contour two-sided circumscribed texture contrast at the pixel point in each window comprises the following steps:
And for each window, the two sides of the outline at all pixel points in the window are circumscribed with texture contrast to form a texture contrast sequence of the window.
5. The visual positioning method for the medium plate lifting appliance clamping process according to claim 1, wherein the constructing the local texture similarity of each window according to the difference of the texture contrast sequences of each window and the rest windows in the outline of each window comprises the following steps:
and for each window, calculating the DTW distance between the window and the texture contrast sequences of the other windows in the outline of the window, calculating the sum value of all the DTW distances of the window, and taking the sum value as the local texture similarity of the window.
6. The visual positioning method for a medium plate sling clamping process according to claim 1, wherein the construction of the two-side circumscribed texture contrast fluctuation coefficients of each contour according to the local texture similarity of the window in each contour comprises the following steps:
For each contour, forming a local texture similarity sequence of all windows in the contour according to a window sequence, calculating the sum value of all elements in the local texture similarity sequence, marking the sum value as a first sum value, calculating the ratio of the first sum value to the number of windows in the contour, marking the ratio as a first ratio, calculating the mean value of all elements in the local texture similarity sequence, calculating the difference value of each element in the local texture similarity sequence and the mean value, calculating the variance of all elements in the local texture similarity sequence, calculating the ratio of the difference value of each element in the local texture similarity sequence to the variance, marking the ratio as a second ratio, marking the sum value of the second ratio of all elements in the local texture similarity sequence as a second sum value, and taking the product of the first ratio and the second sum value as a contour two-side circumscribed texture contrast fluctuation coefficient.
7. The visual positioning method for a medium plate sling clamping process according to claim 1, wherein the obtaining the suspected corner profile according to the external tangent texture contrast fluctuation coefficient at two sides of each profile comprises:
And arranging the contours from small to large according to the fluctuation coefficients of the external-cut texture contrast ratios at the two sides, and taking the contours corresponding to the fluctuation coefficients of the external-cut texture contrast ratios at the two sides of the preset proportion as suspected corner contours.
8. The visual positioning method for a medium plate lifting appliance clamping process according to claim 1, wherein the constructing the significant fraction of each suspected corner profile relative to the other suspected corner profiles according to the gradient value sequence of each suspected corner profile and the other suspected corner profiles comprises:
For each suspected corner contour, acquiring gradient values of each pixel point of the suspected corner contour, and arranging the gradient values according to the sequence of the pixel points to form a gradient value sequence of the suspected corner contour to acquire the average value of all elements in the gradient value sequence of the suspected corner contour; calculating the absolute value of the difference value of the average value of the suspected corner outline and the rest suspected corner outlines, and when the absolute value of the difference value of the suspected corner outlines is smaller than or equal to 0.1, the significant fraction of the suspected corner outlines relative to the corners of the suspected corner outlines is 1;
Calculating the product of the average value of the suspected corner profile and the rest of the suspected corner profiles, calculating the difference value between-1 and a preset value larger than zero, and calculating the sum value of-1 and the preset value larger than zero, wherein when the product of the suspected corner profiles is larger than the difference value and smaller than the sum value, the significant fraction of the suspected corner profile relative to the corners of the suspected corner profiles is 1;
Otherwise, the significant fraction of the suspected corner profile relative to the corners of the remaining suspected corner profiles is 0.1.
9. The visual positioning method for a medium plate lifting appliance clamping process according to claim 1, wherein the constructing the edge profile saliency of each suspected edge profile according to the edge saliency score of each suspected edge profile relative to the rest of each suspected edge profile and the two-sided circumscribed texture contrast fluctuation coefficients of the suspected edge profile comprises:
And calculating the sum value of the edge significance scores of the suspected edge contours relative to the other suspected edge contours, marking the sum value as a first sum value, calculating the absolute value of the difference value of the external-tangent texture contrast fluctuation coefficients of the two sides of the suspected edge contours and the other suspected edge contours, calculating the sum value of all the absolute values of the difference values of the suspected edge contours, marking the sum value as a second sum value, calculating the sum value of the second sum value and a preset value larger than zero, marking the sum value as a third sum value, and taking the ratio of the first sum value to the third sum value as the edge contour significance of the suspected edge contours.
10. The visual positioning method for the medium plate lifting appliance clamping process according to claim 1, wherein the marking of the position of the medium plate by using Boosting algorithm according to the edge profile saliency, the visual positioning in the medium plate lifting appliance clamping process is realized, and the visual positioning method comprises the following steps:
The input of the Boosting algorithm is the 4 outlines with the maximum saliency of the edge outline of the medium plate image, and the output of the Boosting algorithm is the marking position of the medium plate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410295596.6A CN117893612B (en) | 2024-03-15 | 2024-03-15 | Visual positioning method for medium plate lifting appliance clamping process |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410295596.6A CN117893612B (en) | 2024-03-15 | 2024-03-15 | Visual positioning method for medium plate lifting appliance clamping process |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117893612A true CN117893612A (en) | 2024-04-16 |
CN117893612B CN117893612B (en) | 2024-05-24 |
Family
ID=90650995
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410295596.6A Active CN117893612B (en) | 2024-03-15 | 2024-03-15 | Visual positioning method for medium plate lifting appliance clamping process |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117893612B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1996032691A1 (en) * | 1995-04-10 | 1996-10-17 | Rebus Technology, Inc. | System and method for automatic page registration and automatic zone detection during forms processing |
CN108921821A (en) * | 2018-06-01 | 2018-11-30 | 中国人民解放军战略支援部队信息工程大学 | Method of discrimination based on the LASSO mammary cancer armpit lymph gland transfering state returned |
CN109072285A (en) * | 2016-02-25 | 2018-12-21 | 麻省理工学院 | Extend the method for clinical tissue sample |
CN117541578A (en) * | 2024-01-04 | 2024-02-09 | 深圳市鑫显光电科技有限公司 | High-performance full-view angle liquid crystal display screen detection method and system |
-
2024
- 2024-03-15 CN CN202410295596.6A patent/CN117893612B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1996032691A1 (en) * | 1995-04-10 | 1996-10-17 | Rebus Technology, Inc. | System and method for automatic page registration and automatic zone detection during forms processing |
CN109072285A (en) * | 2016-02-25 | 2018-12-21 | 麻省理工学院 | Extend the method for clinical tissue sample |
CN108921821A (en) * | 2018-06-01 | 2018-11-30 | 中国人民解放军战略支援部队信息工程大学 | Method of discrimination based on the LASSO mammary cancer armpit lymph gland transfering state returned |
CN117541578A (en) * | 2024-01-04 | 2024-02-09 | 深圳市鑫显光电科技有限公司 | High-performance full-view angle liquid crystal display screen detection method and system |
Non-Patent Citations (1)
Title |
---|
王田;邹子龙;乔美娜;: "基于图像特征分析的物体轮廓提取", 北京航空航天大学学报, no. 08, 19 November 2015 (2015-11-19) * |
Also Published As
Publication number | Publication date |
---|---|
CN117893612B (en) | 2024-05-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN117173189B (en) | Visual inspection system for polishing effect of aluminum alloy surface | |
CN115063423B (en) | Self-adaptive identification method for cold and hot cracks of mechanical castings based on computer vision | |
CN116645367B (en) | Steel plate cutting quality detection method for high-end manufacturing | |
CN115049656A (en) | Method for identifying and classifying defects in silicon steel rolling process | |
CN116612470B (en) | Bread detection method and system based on visual characteristics | |
CN114998198A (en) | Injection molding surface defect identification method | |
CN104568956B (en) | The detection method of the steel strip surface defect based on machine vision | |
CN115082472B (en) | Quality detection method and system for hub mold casting molding product | |
CN107999548B (en) | Method and device for identifying local high points of hot-rolled strip steel profile | |
CN112884746A (en) | Character defect intelligent detection algorithm based on edge shape matching | |
CN115096206A (en) | Part size high-precision measurement method based on machine vision | |
CN110717900A (en) | Pantograph abrasion detection method based on improved Canny edge detection algorithm | |
CN117893612B (en) | Visual positioning method for medium plate lifting appliance clamping process | |
CN109308707B (en) | Non-contact type online measuring method for thickness of aluminum ingot | |
CN107657620A (en) | A kind of method and system of textured metal freezing region recognition | |
CN117474891B (en) | Gear heat treatment defect detection method | |
CN116823820B (en) | Aluminum-carbon integral stopper rod contour bending detection method | |
CN109622926B (en) | Ingot block edge detection method for ingot casting process | |
CN115100510B (en) | Tire wear degree identification method | |
CN109615626B (en) | Quality evaluation method for chip surface printing symbol structure defects | |
CN114971064A (en) | Hot-rolled strip steel surface defect prediction method based on NGboost algorithm | |
CN114565563B (en) | Color steel plate surface abnormity detection method based on artificial intelligence | |
CN117974663B (en) | Glass fiber sleeve quality detection method based on image characteristics | |
CN111209872A (en) | Real-time rolling fingerprint splicing method based on dynamic programming and multi-objective optimization | |
CN116563290B (en) | Method for detecting polishing degree of frame surface of showcase door |
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 |