CN117392113A - Method, device and storage medium for measuring size of deep hole structure workpiece - Google Patents
Method, device and storage medium for measuring size of deep hole structure workpiece Download PDFInfo
- Publication number
- CN117392113A CN117392113A CN202311581010.4A CN202311581010A CN117392113A CN 117392113 A CN117392113 A CN 117392113A CN 202311581010 A CN202311581010 A CN 202311581010A CN 117392113 A CN117392113 A CN 117392113A
- Authority
- CN
- China
- Prior art keywords
- deep hole
- image
- workpiece
- hole structure
- pixel
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000005259 measurement Methods 0.000 claims abstract description 40
- 230000011218 segmentation Effects 0.000 claims abstract description 34
- 238000004364 calculation method Methods 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000000605 extraction Methods 0.000 claims description 23
- 238000005457 optimization Methods 0.000 claims description 12
- 230000004927 fusion Effects 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 5
- 238000013461 design Methods 0.000 claims description 3
- 230000008485 antagonism Effects 0.000 claims description 2
- 238000003708 edge detection Methods 0.000 claims description 2
- 238000000691 measurement method Methods 0.000 abstract description 3
- 238000003384 imaging method Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 239000000047 product Substances 0.000 description 3
- 238000001914 filtration Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000007795 chemical reaction product Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0475—Generative networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4038—Scaling the whole image or part thereof for image mosaicing, i.e. plane images composed of plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
-
- 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
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
Abstract
The invention mainly relates to the technical field of industrial device measurement. The problems of large image processing data volume, high calculation cost and low measurement efficiency caused by the influence of the focal length and the depth of field of an image acquisition system when the size of a workpiece with a deep hole structure is measured by the traditional measurement method are solved; the invention provides a method, a device and a storage medium for measuring the size of a deep hole structure workpiece, which comprise the steps of obtaining an image of the deep hole structure workpiece, performing defocusing and deblurring treatment to obtain a clear image of an edge to be measured; extracting pixel-level semantic segmentation features of the image to be detected after defocusing and deblurring; and obtaining sub-pixel level contour information of the edge to be detected of the deep hole structure workpiece in the image according to the extracted pixel level semantic segmentation characteristics, and designing constraint conditions to obtain size information of the deep hole structure workpiece to be detected.
Description
Technical Field
The invention mainly relates to the technical field of industrial device measurement, in particular to a method, a device and a storage medium for measuring the size of a deep hole structure workpiece.
Background
The size measurement of the deep hole structure workpiece is always a technical problem in the industry due to the restriction of the depth of field and the light source of the imaging system lens. The traditional solution is to shoot a plurality of images through a plurality of focal lengths, and to conduct image registration and depth of field fusion to obtain the size of the deep hole structure workpiece. Meanwhile, a certain jitter can be generated in the motion process of the motion mechanism, so that pictures shot by different focal lengths have fine displacement, the registration of multi-focal-length images has larger challenges due to differences in focal depth and depth of field, and further, the precision errors caused by image registration and depth of field fusion cannot be estimated and eliminated, and the problems of high calculation cost, low measurement efficiency and the like exist due to the fact that the number of images to be processed is large.
The size measurement method based on CCD vision is mostly based on contour point information at pixel level, the detection precision is limited by the imaging precision, and if the imaging precision is further improved, the hardware cost of an imaging system is increased.
The existing traditional subpixel edge extraction algorithm has low robustness under the condition of complex background textures.
Therefore, how to further improve the dimension measurement precision of the deep hole structure workpiece under the existing imaging precision is a problem to be solved urgently.
Disclosure of Invention
The invention aims to solve the technical problems
The method, the device and the storage medium for measuring the size of the workpiece with the deep hole structure are provided, and the problems that the traditional measuring method is influenced by the focal length and the depth of field of an image acquisition system when measuring the size of the workpiece with the deep hole structure, so that the image processing data volume is large, the calculation cost is high and the measuring efficiency is low are solved; and the problem of high hardware cost for measuring the size of the workpiece with the deep hole structure based on CCD vision and the problem of low robustness for measuring the size of the workpiece with the deep hole structure based on a subpixel edge extraction algorithm.
The invention solves the technical problems
A method of measuring the dimensions of a deep hole structured workpiece, comprising:
acquiring an image of a deep hole structure workpiece;
performing defocusing and deblurring treatment on an image of the deep hole structure workpiece to obtain a clear image of the edge to be detected of the deep hole structure workpiece in the image;
extracting pixel-level semantic segmentation features of the image to be detected after defocusing and deblurring;
acquiring outline information of the edge to be detected of the deep hole structural member in the image according to the extracted pixel-level semantic segmentation features;
and designing constraint conditions according to the contour information of the edge to be detected of the deep hole structural member in the acquired image, and acquiring the size information of the workpiece with the deep hole structure to be detected.
Further, the performing defocus deblurring processing on the image data of the deep hole structure workpiece specifically includes:
based on the generation of the antagonism network design generator, the arbiter, and the loss function; training a generator based on the discriminator and the loss function, and performing deblurring treatment on the acquired image of the deep hole structure workpiece by using the trained generator;
or generating a pixel-level deblurring filter based on spatial deconvolution prediction, and predicting the deblurring characteristics of the out-of-focus image based on the acquired deblurring filter.
Further, extracting pixel-level semantic segmentation features of the out-of-focus deblurred image to be detected specifically includes slicing the deblurred image by using a binary algorithm or a depth semantic segmentation algorithm to obtain a plurality of pixel-level masks of the deep hole structure workpiece image to be detected, carrying out normalization processing, and merging the mask images into a complete image according to a slicing rule.
Further, the extracting the sub-pixel level contour point information of the object to be detected specifically includes obtaining sub-pixel coordinate values of all edges of the image to be detected through a custom Gabor filter and gaussian fitting by using the obtained pixel level segmentation mask.
Further, designing constraint conditions according to the contour information of the edge to be measured of the deep hole structural member in the acquired image, acquiring the size information of the workpiece with the deep hole structure to be measured specifically comprises,
selecting a reference edge, and performing linear fitting on a sub-pixel level outline of the reference edge to obtain a measurement datum line;
selecting sub-pixel level contour points of the edge of a target to be detected of the workpiece with the deep hole structure to be detected according to the designed contour information constraint condition, and performing straight line fitting on the obtained sub-pixel level contour points;
and calculating the distance from the straight line fitting of the sub-pixel level contour points to the measurement datum line, and acquiring the size information to be measured of the deep hole structure workpiece according to the distance from the straight line fitting of the sub-pixel level contour points to the measurement datum line.
Further, the method further comprises the step of selecting a camera, a lens and a light source for acquiring the image of the deep hole structure workpiece according to the depth of field of the deep hole structure when acquiring the image data of the deep hole structure workpiece.
Further, the method is characterized in that if a plurality of deep hole structure workpiece images are obtained in the step 1, defocusing and deblurring are carried out on the plurality of images, and feature point extraction and matching, homography matrix calculation and optimization, splicing sequence optimization, optimal splicing seam searching and image fusion are respectively carried out on each image after defocusing and deblurring, so that the deep hole structure image with clear edges to be detected is obtained.
The invention also provides a device for measuring the size of the deep hole structural workpiece, which is used for realizing the method for measuring the size of the deep hole structural workpiece, and comprises an image acquisition module, a defocusing and deblurring module, an image splicing module, a pixel-level semantic segmentation feature extraction module, a sub-pixel contour point extraction module and a size measurement module;
the image acquisition module is used for acquiring image data of the deep hole structure workpiece;
the defocusing and deblurring module is used for defocusing and deblurring the acquired image data of the deep hole structure workpiece and clearing edges to be detected in the image;
the pixel-level semantic segmentation feature extraction module is used for extracting pixel-level semantic segmentation features of the defocused and defuzzified image;
the sub-pixel contour point extraction module is used for extracting sub-pixel level contour information in the segmented image;
the dimension measurement module is used for obtaining dimension information of the deep hole structure workpiece according to the extracted sub-pixel level profile information and the designed constraint conditions
Further, the device also comprises an image splicing module, wherein the image splicing module is used for carrying out defocusing and deblurring treatment on a plurality of images when the image acquisition module acquires the plurality of deep hole framework workpiece images, and respectively carrying out feature point extraction and matching, homography matrix calculation and optimization, splicing order optimization, optimal splicing seam searching and image fusion on each image subjected to defocusing and deblurring to obtain a deep hole structural member image with a clear edge.
The invention also provides a computer readable storage medium, wherein the computer readable storage medium stores program codes, and the program codes are used for realizing the method for measuring the size of the deep hole structure workpiece.
The beneficial effects of the invention are that
The method for measuring the size of the workpiece with the deep hole structure adopts the defocusing and deblurring strategy, and obtains clear imaging of the workpiece with the deep hole structure by adopting the defocusing and deblurring method for the image shot by the lens with larger depth of field, so that the number of times of movement of a movement mechanism along the focal length direction is reduced, the movement is not required in ideal condition, the hardware cost and the number of images to be processed of the traditional method are further reduced, the calculation cost is saved, and the measurement efficiency is improved; meanwhile, the sub-pixel point contour extraction strategy is adopted, sub-pixel point contour information of the object to be measured is extracted according to the pixel-level semantic segmentation characteristics, and the accuracy and the robustness of size measurement can be further improved. The method provided by the invention has the characteristics of high measurement precision, low hardware cost, easiness in implementation and the like, and can effectively solve the problems of the existing deep hole structure workpiece size measurement method.
Drawings
FIG. 1 is a flow chart of a method for measuring the dimension of a workpiece with a deep hole structure according to the present invention;
fig. 2 is a schematic diagram of a device module for measuring the dimension of a workpiece with a deep hole structure according to the present invention.
Detailed Description
As shown in FIG. 1, the method for measuring the size of the deep hole structural workpiece specifically comprises the following steps of.
S1, designing a proper imaging system according to the depth of field required by deep hole structure workpiece measurement, and acquiring image data of the deep hole structure workpiece.
The imaging system specifically comprises an industrial camera, a lens and a light source; some edges of the shot image may be blurred due to defocus, and a lens is selected according to the depth of field of the deep hole structure; the selected light source collimation needs to be compatible with the deep hole structure. Judging whether to acquire images of different areas of a plurality of deep hole structural workpieces according to the field of view of image shooting and the depth of field of a lens
S2, defocusing and deblurring the image obtained in the step S1, so that edges to be detected in the image are clear.
Methods of defocus deblurring include, but are not limited to: adopting a GAN method to design a generator, a discriminator and a loss function, so that the image generated by the generator is similar to the clear image as far as possible to achieve the purpose of deblurring; or based on spatial deconvolution, a deblurring filter at a prediction pixel level is applied to the defocused image to obtain a deblurring feature.
If the number of the images acquired in the step S1 is multiple, performing image stitching on the out-of-focus deblurred images: and inputting defocused and deblurred images of a plurality of areas of the workpiece, and obtaining a complete image of the workpiece with the deep hole structure with clear edges through feature point extraction and matching, homography matrix calculation and optimization, splicing sequence optimization, optimal splicing seam searching and image fusion.
S3, extracting pixel-level semantic segmentation features of the target to be detected;
and (3) carrying out normalization processing on the acquired image slices in the step (S2) by adopting an edge detection, binary algorithm or deep learning semantic segmentation algorithm to obtain pixel level division masks of the targets to be measured of a plurality of slices, and merging the slice masks into a whole graph according to a slicing rule.
S4, extracting sub-pixel level contour point information of the object to be detected according to the pixel level semantic segmentation features of the step S3.
Inputting the sub-pixel coordinates of all edges obtained by the custom Gabor filter and Gaussian fitting, wherein the mask is divided by the pixel level obtained in the step S3. Inputting the pixel-level semantic segmentation mask obtained in the step S4, customizing a kernel of a Gabor filter, extracting gradient amplitude and angle of the edge of the semantic segmentation mask based on the Gabor filter, obtaining a preliminary sub-pixel contour point through Gaussian fitting, finally performing non-maximum suppression, setting a threshold value, and filtering the sub-pixel contour point according to the edge length to obtain final sub-pixel-level contour point information of the target to be detected.
S5, designing constraint conditions according to the sub-pixel level contour point information of the step S4 to obtain information of each size of the deep hole structure workpiece, and judging whether the workpiece is qualified or not;
firstly, designing constraint conditions according to measurement requirements, selecting a reference edge according to the constraint conditions, and obtaining a measurement reference line y=k by performing linear fitting on sub-pixel level contour points of the reference edge 1 x+b 1 ；
Selecting sub-pixel level contour points of the target edge according to constraint conditions such as position, size and the like, and performing straight line fitting to obtain sub-pixel level of the target edgeContour point and fitting straight line y=k 2 x+b 2 ；
Calculating the edge of the object to a measurement datum y=k 1 x+b 1 The calculation method includes, but is not limited to, the average distance from the point to the straight line, the nearest distance between the two straight lines, the farthest distance between the two straight lines, etc.
Example 1:
the deep hole structure workpiece to be measured is the female end of the connector, the size of the opening of the female end of the connector is required to be measured, and the depth of field of the measured size is about 5mm.
According to the deep hole structure workpiece to be measured, a camera and a lens which meet the measurement precision are selected, such as a 2000-ten thousand camera, a lens with a depth of field of about 3mm, a coaxial light source with good collimation, a field of view can cover a complete connector female end product, the number of pictures shot is 1, and part of edges to be measured are blurred due to defocusing.
GAN-based methods, such as DeblurGAN, are input as out-of-focus blurred images. Because the video memory occupied by the image with 2000 ten thousand resolution is larger, the video memory of the deployed computer is insufficient. The acquired image is sliced, for example 500x500, and finally deblurred and spliced into a whole map of 2000 ten thousand in a slicing mode.
And (3) slicing the 2000 ten thousand deblurred images obtained in the step (S2), wherein the slice size is 256x256, adopting a semantic segmentation model such as unet, using labeling tools such as labelme to label edges of the parts to be measured at a pixel level, converting labeling information into mask images, and training the semantic segmentation model until the semantic segmentation model is in a convergence state according to the preprocessed 256x256 images and mask images of the slices. And reasoning the test pictures by adopting the trained model, normalizing the test pictures to 0-255, and obtaining pixel-level semantic segmentation characteristics of the target to be tested.
And (3) customizing a kernel of the Gabor filter, extracting gradient amplitude and angle of a semantic segmentation mask edge based on the Gabor filter, obtaining a preliminary sub-pixel contour point through Gaussian fitting, finally performing non-maximum suppression, setting a threshold value, and filtering the sub-pixel contour point according to the edge length to obtain final sub-pixel contour point information of the target to be detected.
According to the extracted sub-pixel level contour point information, designing constraint conditions, for example, selecting sub-pixel level contour points of the target edge according to the constraint conditions such as position, size and the like, and performing straight line fitting to obtain the sub-pixel level contour points of the target edge and a fitting straight line;
the distance from the edge of the target to the measurement datum is calculated by means of, but not limited to, the average distance from the point to the straight line, the nearest distance between the two straight lines, the farthest distance between the two straight lines and the like.
Selecting sub-pixel level contour points of the reference edge according to the position and the length of the reference edge to perform straight line fitting to obtain a measurement reference line y=k 1 x+b 1 The method comprises the steps of carrying out a first treatment on the surface of the Similarly, selecting sub-pixel level contour points of the target edge according to the geometric constraint relation, and calculating all the sub-pixel level contour points of the target edge to a datum line y=k 1 x+b 1 And obtaining the high-precision measurement size. If the measurement size is 0.15mm, the measurement standard is 0.12mm plus or minus 0.04mm, the tolerance range is met, and the product is qualified.
Example 2:
as shown in fig. 2, the device for measuring the size of the workpiece with the deep hole structure comprises an image acquisition module, a defocusing and deblurring module, an image splicing module, a pixel-level semantic segmentation feature extraction module, a sub-pixel contour point extraction module and a size measurement module;
the image acquisition module is used for acquiring image data of the deep hole structure workpiece;
the defocusing and deblurring module is used for defocusing and deblurring the acquired image data of the deep hole structure workpiece and clearing edges to be detected in the image;
the image splicing module is used for carrying out defocusing and deblurring treatment on the plurality of images when the image acquisition module acquires the plurality of deep hole structure workpiece images, and respectively carrying out feature point extraction and matching, homography matrix calculation and optimization, splicing sequence optimization, optimal splicing seam searching and image fusion on each image subjected to defocusing and deblurring to obtain a deep hole structure image with a clear edge;
the pixel-level semantic segmentation feature extraction module is used for extracting pixel-level semantic segmentation features of the defocused and defuzzified image;
the sub-pixel contour point extraction module is used for extracting sub-pixel level contour information in the segmented image;
and the dimension measurement module is used for obtaining dimension information of the deep hole structure workpiece according to the extracted subpixel level profile information and the designed constraint conditions.
The deep hole structure work piece that awaits measuring is the screw, and the size information of the deep hole structure work piece that awaits measuring is the distance of screw to screw top edge, and the depth of field of measurement size is about 4mm, selects camera, the camera that satisfies measurement accuracy, camera, like 500 ten thousand cameras, the camera of about 2mm of depth of field, the better pointolite of collimation, because of the precision requirement is higher, the camera field of vision can not cover complete product, and the picture quantity of shooing is 2, and overlap area is 100 pixels, and the edge of partly waiting to measure can be blurred because of defocusing.
The defocusing and deblurring module predicts a pixel-level deblurring filter based on a spatial deconvolution method and is applied to a defocused image to obtain a deblurring characteristic. The input is an out-of-focus blurred image, a specific method such as IFAN. Because the image with 500 ten thousand resolution is input, the occupied video memory is larger, and the deployed computer video memory is insufficient. The acquired image is sliced, for example 400x400, and finally deblurred and spliced into a 500-ten thousand whole image in a slicing mode.
Because the acquired images are 2, respectively performing defocusing and deblurring on the 2 images, and then performing image stitching; and carrying out feature point matching on 2 area images by adopting SIFT features, selecting feature points in an overlapped area, calculating a homography matrix, carrying out rotation and translation on the images according to the homography matrix, and carrying out image fusion by weighted average to obtain a deblurred image with the whole image of about 800 ten thousand.
Selecting a unet++ network, and setting a training pixel-level semantic segmentation feature extraction module such as a loss function, an optimizer, a learning strategy and the like by constructing training samples. And reasoning the test picture by adopting the trained pixel-level semantic segmentation feature extraction module, and normalizing the test picture to 0-255 to obtain the pixel-level semantic segmentation feature of the target to be tested.
The dimension measuring module selects the reference edge according to the position and the length of the reference edgeStraight line fitting is carried out on the sub-pixel level contour points of (2) to obtain a measurement datum line y 1 ＝k 1 x+b 1 The method comprises the steps of carrying out a first treatment on the surface of the Similarly, according to the geometric constraint relation, sub-pixel level contour points of the target edge are selected for straight line fitting, and a target edge line y is obtained 2 ＝k 2 x+b 2 . Calculating the center distance of the two straight lines to obtain high-precision measurement size, wherein the measurement size is 0.2mm, the measurement standard is 0.15mm plus or minus 0.03mm, the tolerance range is not satisfied, and the high-precision measurement size is a defective product.
Claims (10)
1. A method for measuring the size of a workpiece with a deep hole structure, comprising the steps of:
acquiring a deep hole structure workpiece image;
performing defocusing and deblurring treatment on the image of the deep hole structure workpiece to obtain a clear image of the edge to be detected of the deep hole structure workpiece in the image;
extracting pixel-level semantic segmentation features of the image to be detected after defocusing and deblurring;
acquiring outline information of the edge to be detected of the deep hole structure workpiece in the image according to the extracted pixel-level semantic segmentation features;
and designing constraint conditions according to the contour information of the edge to be measured of the workpiece with the deep hole structure in the acquired image, and acquiring the information of the size to be measured of the workpiece with the deep hole structure to be measured.
2. The method for measuring the size of the deep hole workpiece according to claim 1, wherein the defocusing and deblurring processing of the image data of the deep hole workpiece specifically comprises:
based on the generation of the antagonism network design generator, the arbiter, and the loss function; training a generator based on the discriminator and the loss function, and performing deblurring treatment on the acquired deep hole structure workpiece image by using the trained generator;
or generating a pixel-level deblurring filter based on spatial deconvolution prediction, and predicting the deblurring characteristics of the out-of-focus image with the deblurring filter.
3. The method for measuring the size of the deep hole structure workpiece according to claim 1, wherein the extracting of the pixel-level semantic segmentation features of the image to be measured after the defocus deblurring treatment specifically comprises slicing the image after deblurring by using edge detection, a binary algorithm or a depth semantic segmentation algorithm to obtain a plurality of pixel-level mask images of the image to be measured of the deep hole structure workpiece, normalizing the pixel-level mask images, and merging the pixel-level mask images into a complete image according to slicing rules.
4. The method for measuring the size of the deep hole structural workpiece according to claim 3, wherein the extracting of the sub-pixel level contour point information of the object to be detected comprises the steps of dividing the obtained pixel level into mask patterns, and obtaining sub-pixel coordinate values of all edges of the image to be detected through a custom Gabor filter and Gaussian fitting.
5. The method for measuring the dimension of a deep hole workpiece according to claim 1, wherein the step of obtaining dimension information of the deep hole workpiece to be measured comprises the steps of,
selecting a reference edge serving as a measurement reference, and performing linear fitting on a sub-pixel level outline of the reference edge to obtain the measurement reference;
selecting sub-pixel level contour points of the edge of a target to be detected of the workpiece with the deep hole structure to be detected according to the designed contour information constraint condition, and performing straight line fitting on the obtained sub-pixel level contour points;
and calculating the distance from the sub-pixel level contour point fitting straight line of the edge of the target to be measured to the measurement datum line, and obtaining the dimension information of the workpiece to be measured with the deep hole structure according to the distance from the sub-pixel level contour point fitting straight line of the edge of the target to be measured to the measurement datum line.
6. The method for measuring the size of the workpiece with the deep hole structure according to claim 1, wherein the method further comprises selecting a camera, a lens and a light source for acquiring the image of the workpiece with the deep hole structure according to the depth of field of the workpiece with the deep hole structure when acquiring the image data of the workpiece with the deep hole structure.
7. The method for measuring the size of the deep hole structural workpiece according to any one of claims 1 to 6, wherein if a plurality of deep hole structural workpiece images are obtained from the deep hole structural workpiece images, defocus deblurring is carried out on the plurality of images, and feature point extraction and matching, homography matrix calculation and optimization, splicing order optimization, optimal splice joint searching and image fusion are respectively carried out on each image after defocus deblurring, so that the deep hole structural workpiece image with clear edges to be measured is obtained.
8. A device for measuring the size of a workpiece with a deep hole structure, which is used for realizing the method for measuring the size of the workpiece with the deep hole structure, and is characterized by comprising an image acquisition module, a defocusing and deblurring module, an image stitching module, a pixel-level semantic segmentation feature extraction module, a sub-pixel contour point extraction module and a size measurement module;
the image acquisition module is used for acquiring image data of the deep hole structure workpiece;
the defocusing and deblurring module is used for defocusing and deblurring the acquired image data of the deep hole structure workpiece and clearing edges to be detected in the image;
the pixel-level semantic segmentation feature extraction module is used for extracting pixel-level semantic segmentation features of the defocused and defuzzified image;
the sub-pixel contour point extraction module is used for extracting sub-pixel level contour information in the segmented image;
and the dimension measurement module is used for obtaining dimension information of the deep hole structure workpiece according to the extracted subpixel level profile information and the designed constraint conditions.
9. The device for measuring the size of the deep hole structural workpiece according to claim 8, further comprising an image stitching module, wherein the image stitching module is used for performing defocusing and deblurring processing on a plurality of images when the image acquisition module acquires the plurality of deep hole structural workpiece images, and respectively performing feature point extraction and matching, homography matrix calculation and optimization, stitching sequence optimization, optimal stitching seam searching and image fusion on each image after defocusing and deblurring to obtain the deep hole structural workpiece image with a clear edge.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a program code for implementing a method of deep hole structured workpiece dimension measurement according to any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311581010.4A CN117392113A (en) | 2023-11-24 | 2023-11-24 | Method, device and storage medium for measuring size of deep hole structure workpiece |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311581010.4A CN117392113A (en) | 2023-11-24 | 2023-11-24 | Method, device and storage medium for measuring size of deep hole structure workpiece |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117392113A true CN117392113A (en) | 2024-01-12 |
Family
ID=89437519
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311581010.4A Pending CN117392113A (en) | 2023-11-24 | 2023-11-24 | Method, device and storage medium for measuring size of deep hole structure workpiece |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117392113A (en) |
-
2023
- 2023-11-24 CN CN202311581010.4A patent/CN117392113A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10699476B2 (en) | Generating a merged, fused three-dimensional point cloud based on captured images of a scene | |
| CN106204572B (en) | Road target depth estimation method based on scene depth mapping | |
| US8385630B2 (en) | System and method of processing stereo images | |
| CN111652085B (en) | Object identification method based on combination of 2D and 3D features | |
| CN107369159B (en) | Threshold segmentation method based on multi-factor two-dimensional gray level histogram | |
| CN111612765B (en) | Method for identifying and positioning round transparent lens | |
| JP6899189B2 (en) | Systems and methods for efficiently scoring probes in images with a vision system | |
| CN113324478A (en) | Center extraction method of line structured light and three-dimensional measurement method of forge piece | |
| CN105718931B (en) | System and method for determining clutter in acquired images | |
| Cvišić et al. | Recalibrating the KITTI dataset camera setup for improved odometry accuracy | |
| CN111354047B (en) | Computer vision-based camera module positioning method and system | |
| CN114252449B (en) | Aluminum alloy weld joint surface quality detection system and method based on line structured light | |
| CN103824275A (en) | System and method for finding saddle point-like structures in an image and determining information from the same | |
| EP4066162A1 (en) | System and method for correspondence map determination | |
| CN113686314A (en) | Monocular water surface target segmentation and monocular distance measurement method of shipborne camera | |
| CN115953550A (en) | Point cloud outlier rejection system and method for line structured light scanning | |
| CN112200790A (en) | Cloth defect detection method, device and medium | |
| Kruger et al. | In-factory calibration of multiocular camera systems | |
| CN104966283A (en) | Imaging layered registering method | |
| CN113688846A (en) | Object size recognition method, readable storage medium, and object size recognition system | |
| KR101733028B1 (en) | Method For Estimating Edge Displacement Againt Brightness | |
| CN116596987A (en) | Workpiece three-dimensional size high-precision measurement method based on binocular vision | |
| CN115830018A (en) | Carbon block detection method and system based on deep learning and binocular vision | |
| CN117392113A (en) | Method, device and storage medium for measuring size of deep hole structure workpiece | |
| CN113970560B (en) | Defect three-dimensional detection method based on multi-sensor fusion |
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 |