CN102914545B - Gear defect detection method and system based on computer vision - Google Patents
Gear defect detection method and system based on computer vision Download PDFInfo
- Publication number
- CN102914545B CN102914545B CN201210460169.6A CN201210460169A CN102914545B CN 102914545 B CN102914545 B CN 102914545B CN 201210460169 A CN201210460169 A CN 201210460169A CN 102914545 B CN102914545 B CN 102914545B
- Authority
- CN
- China
- Prior art keywords
- gear
- image
- radius
- tooth
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image processing and provides a gear defect detection method and system based on computer vision. The method comprises the following steps of: acquiring a source image of a detected gear acquired by an image acquisition card through a peripheral component interconnect (PCI) slot; performing preprocessing operation comprising grey level transformation, filtering and noise reduction and threshold segmentation on the source image of the detected gear, and obtaining gear parameters of the detected gear; and performing morphological parameter analysis on the gear parameters of the detected gear, and acquiring an analysis result of the teeth where the defects possibly exist. The gear defect detection method is high in measurement speed and high in precision, the functional requirements of the conventional gear parameter measurement are met, the production efficiency and production automation degree can be greatly improved in the large-scale industrial production process, the measurement precision requirement in the actual production can be met, the measurement time is only several seconds, the requirement on the measurement speed in industrial measurement can be met, and the method has high theoretical and practical values.
Description
Technical field
The invention belongs to technical field of image processing, relate in particular to a kind of gear defects detection method and system based on computer vision.
Background technology
Gear, as the typical power transmitter part of one, has been brought into play very important effect in the course of industrial development, and the quality of its quality directly affects the performance of engineering goods.Therefore, in the process of Gear Production manufacture, must carry out strict detection and control to product quality.Due to the geometric features of gear itself, for tooth collapse, the measuring process complexity of the defect such as hypodontia, tooth are askew, conventional measuring equipment is as complex structures such as three coordinate measuring machine, Error of Gears somascopes, expensive, also higher to tester's requirement.The development of modern manufacturing industry is had higher requirement to the defects detection of gear product.
At present, the central coordinate of circle that first gear testing method based on computer vision generally need to determine gear is as benchmark, therefore, the establishment of central coordinate of circle directly has influence on the precision of gear tooth parameter measurement, in traditional image method detects, center of circle Spot detection algorithm has the edge that generally needs edge detection method to determine gear, then adopts gravity model appoach, median method and Hough converter technique to determine the center of circle.
First two algorithm requires image distribution more even, otherwise produces larger error; Rear a kind of algorithm needs pointwise ballot, record, and the time used is more, and precision is also not high enough.
Summary of the invention
The object of the present invention is to provide a kind of gear defects detection method based on computer vision, be intended to solve efficiency and the lower problem of precision of the gear defects detection algorithm that prior art provides.
The present invention is achieved in that a kind of gear defects detection method based on computer vision, and described method comprises the steps:
Obtain the source images of the tested gear that image pick-up card collects by PCI slot;
Pretreatment operation to the source images of the tested gear acquiring including greyscale transformation, filtering and noise reduction and Threshold segmentation, obtains the gear parameter of tested gear;
The gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, obtain the analysis result that comprises the number of teeth that may have defect;
Wherein, described to the source images of the tested gear acquiring the pretreatment operation including greyscale transformation, filtering and noise reduction and Threshold segmentation, the step that obtains the gear parameter of tested gear specifically comprises the steps:
24 source color images of the tested gear collecting are carried out to greyscale transform process, obtain 8 gray scale gear graph pictures;
Utilize median filtering algorithm, described 8 gray scale gear graphs are looked like to carry out noise removal;
Utilize Image binarizing algorithm, gray scale gear graph is looked like to carry out Threshold segmentation, obtain the gear and the background image that separate;
Described the gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, the step of obtaining the analysis result that comprises the number of teeth that may have defect specifically comprises the steps:
Utilize filling algorithm, the hole of the gear graph picture that described Binarization methods is obtained is filled;
Binary image after hole is filled, as target, is determined central coordinate of circle, root radius and the radius of addendum of calculating described binary image;
According to the central coordinate of circle of the described binary image calculating, root radius and radius of addendum, tooth top is separated with tooth root, detection of gear gear teeth defect, obtains tooth root image and tooth top image;
By region labeling method, the tooth root image and the tooth top image that obtain are carried out to defect analysis, calculate the number of teeth of gear, determine the defect area of described tooth root and tooth top image.
As a kind of improved plan, described binary image after hole is filled, as target, determines that the step of central coordinate of circle, root radius and radius of addendum of calculating described binary image specifically comprises the steps:
Read in the binary image after hole is filled, record the coordinate on gear;
On described binary image, search the set of distance described picture centre solstics and closest approach set, obtain the coordinate that can comprise respectively the minimal convex polygon of gear;
According to the coordinate of minimal convex polygon, draw out respectively convex polygon;
Described convex polygon is carried out respectively to the Circular curve fitting of least square method, calculate central coordinate of circle, root radius and the radius of addendum of gear.
As a kind of improved plan, central coordinate of circle, root radius and the radius of addendum of the described binary image that described basis calculates, tooth top is separated with tooth root, detection of gear gear teeth defect, the step that obtains tooth root image and tooth top image specifically comprises the steps:
In binary image, draw one taking certain numerical value between root radius and radius of addendum as radius and calculate auxiliary circle;
One illiteracy plate image is set, and described illiteracy plate image size is identical with gear graph picture size, and the radius of described illiteracy plate image equals the half of radius of addendum and root radius sum;
Do difference operation by drawing the gear graph picture and the described illiteracy plate image that calculate after auxiliary circle, obtain tooth top figure image;
Described illiteracy plate image and the gear graph of drawing after calculating auxiliary circle are looked like to do difference operation, obtain tooth root figure image.
As a kind of improved plan, described by region labeling method, the tooth root image and the tooth top image that obtain are carried out to defect analysis, calculate the number of teeth of gear, determine that the step of the defect area of described tooth root and tooth top image specifically comprises:
Respectively the tooth root image and the tooth top image that obtain are carried out to zone marker, obtain tooth root image and tooth top image that pseudo-color mode shows;
The tooth root image that pseudo-color mode is shown and tooth top image carry out respectively number and single area calculates;
Tooth top and tooth root area are added up, asked for respectively the mean value of tooth top area and tooth root area;
By setting the mode of area threshold, determine tooth root and tooth top defect area.
The object of another embodiment of the present invention is to provide a kind of gear defects detection system based on computer vision, and described system comprises:
Source images acquisition module, for obtaining the source images of the tested gear that image pick-up card collects by PCI slot;
Pretreatment operation module, for the pretreatment operation including greyscale transformation, filtering and noise reduction and Threshold segmentation to the source images of the tested gear acquiring, obtains the gear parameter of tested gear;
Morphologic Parameters analysis module, for the analysis result that the gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, obtain comprising the number of teeth that may have defect;
Wherein, described pretreatment operation module specifically comprises:
Greyscale transform process module, for 24 source color images of the tested gear collecting are carried out to greyscale transform process, obtains 8 gray scale gear graph pictures;
Noise removal module, for utilizing median filtering algorithm, looks like to carry out noise removal to described 8 gray scale gear graphs;
Threshold segmentation module, for utilizing Image binarizing algorithm, looks like to carry out Threshold segmentation to gray scale gear graph, obtains the gear and the background image that separate;
Described Morphologic Parameters analysis module specifically comprises:
Hole packing module, for utilizing filling algorithm, the hole of the gear graph picture that described Binarization methods is obtained is filled;
Center of circle radius calculation module, as target, determines central coordinate of circle, root radius and the radius of addendum of calculating described binary image for the binary image after hole is filled;
Tooth root image and tooth top image obtain module, for according to central coordinate of circle, root radius and the radius of addendum of the described binary image calculating, tooth top are separated with tooth root, and detection of gear gear teeth defect, obtains tooth root image and tooth top image;
Defect analysis module, for by region labeling method, carries out defect analysis to the tooth root image and the tooth top image that obtain, calculates the number of teeth of gear, determines the defect area of described tooth root and tooth top image.
As a kind of improved plan, described center of circle radius calculation module specifically comprises:
Coordinate record module, for reading in the binary image after hole is filled, records the coordinate on gear;
Point is searched module, for search the set of distance described picture centre solstics and closest approach set on described binary image, obtains the coordinate that can comprise respectively the minimal convex polygon of gear;
Drafting module, for according to the coordinate of minimal convex polygon, draws out respectively convex polygon;
The Fitting Calculation module, for described convex polygon being carried out respectively to the Circular curve fitting of least square method, calculates central coordinate of circle, root radius and the radius of addendum of gear;
Described tooth root image and tooth top image obtain module and specifically comprise:
Calculate auxiliary circle drafting module, at binary image, draw one taking certain numerical value between root radius and radius of addendum as radius and calculate auxiliary circle;
Cover plate image setting module, for an illiteracy plate image is set, described illiteracy plate image size is identical with gear graph picture size, and the radius of described illiteracy plate image equals the half of radius of addendum and root radius sum;
The first difference operation module, for doing difference operation by drawing the gear graph picture and the described illiteracy plate image that calculate after auxiliary circle, obtains tooth top figure image;
The second difference operation module, for described illiteracy plate image and the gear graph of drawing after calculating auxiliary circle are looked like to do difference operation, obtains tooth root figure image;
Described defect analysis module specifically comprises:
Zone marker module, for respectively the tooth root image and the tooth top image that obtain being carried out to zone marker, obtains tooth root image and tooth top image that pseudo-color mode shows;
Computing module, carries out respectively number and the calculating of single area for tooth root image and tooth top image that pseudo-color mode is shown;
Statistical module, for tooth top and tooth root area are added up, asks for respectively the mean value of tooth top area and tooth root area;
Determination module, for by the mode of setting area threshold, determines tooth root and tooth top defect area.
In embodiments of the present invention, obtain the source images of the tested gear that image pick-up card collects by PCI slot; Pretreatment operation to the source images of the tested gear acquiring including greyscale transformation, filtering and noise reduction and Threshold segmentation, obtains the gear parameter of tested gear; The gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, obtain the analysis result that comprises the number of teeth that may have defect.Not only measuring speed is fast but also precision is high for the embodiment of the present invention, meet the functional requirement that gear conventional parameter is measured, the automaticity that can greatly enhance productivity and produce in industrial processes in enormous quantities, can meet the measuring accuracy requirement in production reality, and Measuring Time is only several seconds, can meet the requirement to measuring speed in commercial measurement, there is larger theory and practical value.
Brief description of the drawings
Fig. 1 is the realization flow figure of the gear defects detection method based on computer vision that provides of the embodiment of the present invention;
Fig. 2 is the pretreatment operation including greyscale transformation, filtering and noise reduction and Threshold segmentation to the source images of the tested gear acquiring that the embodiment of the present invention provides, and obtains the realization flow figure of the gear parameter of tested gear;
Fig. 3 is the schematic diagram of realizing of the noise that brings of removal of images collection that the embodiment of the present invention provides;
Fig. 4 a is that the medium filtering that the embodiment of the present invention provides is processed front gear image schematic diagram;
Fig. 4 b is that the medium filtering that the embodiment of the present invention provides is processed backgear image schematic diagram;
Fig. 5 a is the greyscale transformation histogram that the embodiment of the present invention provides;
Fig. 5 b is the binaryzation gear graph picture that the embodiment of the present invention provides;
Fig. 6 be the embodiment of the present invention provide the gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, obtain the realization flow figure of the analysis result that comprises the number of teeth that may have defect;
Fig. 7 is that the hole that the embodiment of the present invention provides is filled backgear image schematic diagram;
Fig. 8 be the embodiment of the present invention provide hole is filled after binary image as target, determine the realization flow figure of central coordinate of circle, root radius and radius of addendum that calculates described binary image;
Fig. 8 a is the minimum polygon schematic diagram that comprises gear that the embodiment of the present invention provides;
Fig. 8 b be the embodiment of the present invention provide according to the circular schematic diagram of coordinate fitting;
Fig. 9 be the embodiment of the present invention provide according to the central coordinate of circle of the described binary image calculating, root radius and radius of addendum, tooth top is separated with tooth root, detection of gear gear teeth defect, the realization flow figure of acquisition tooth root image and tooth top image;
Figure 10 a is the schematic diagram in the deletion target gear edge region that provides of the embodiment of the present invention;
Figure 10 b is the illiteracy plate image schematic diagram that the embodiment of the present invention provides;
Figure 10 c is that gear graph picture and the described illiteracy plate image that the embodiment of the present invention provides makees difference operation effect schematic diagram;
Figure 10 d is the gear graph picture that provides of the embodiment of the present invention and cover plate image and make pseudo color image schematic diagram after difference operation;
Figure 10 e is that illiteracy plate image and the gear graph that the embodiment of the present invention provides looks like to do difference operation effect schematic diagram;
Figure 10 f is that the illiteracy plate image that provides of the embodiment of the present invention and gear graph look like to do pseudo color image schematic diagram after difference operation;
Figure 11 is the region labeling method of passing through that the embodiment of the present invention provides, and the tooth root image and the tooth top image that obtain are carried out to defect analysis, calculates the number of teeth of gear, determines the specific implementation process flow diagram of the defect area of described tooth root and tooth top image;
Figure 12 a is the tooth top image schematic diagram after the mark that provides of inventive embodiments;
Figure 12 b is the area histogram of each tooth top image of providing of inventive embodiments;
Figure 13 is the structured flowchart of the gear defects detection system based on computer vision that provides of the embodiment of the present invention;
Figure 14 is the structured flowchart of the pretreatment operation module that provides of the embodiment of the present invention;
Figure 15 is the structured flowchart of the Morphologic Parameters analysis module that provides of the embodiment of the present invention;
Figure 16 is the structured flowchart of the center of circle radius calculation module that provides of the embodiment of the present invention;
Figure 17 is the structured flowchart that the tooth root image that provides of the embodiment of the present invention and tooth top image obtain module;
Figure 18 is the structured flowchart of the defect analysis module that provides of the embodiment of the present invention.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein, only in order to explain the present invention, is not intended to limit the present invention.
Fig. 1 shows the realization flow figure of the gear defects detection method based on computer vision that the embodiment of the present invention provides, and its concrete step is as described below:
In step S101, obtain the source images of the tested gear that image pick-up card collects by PCI slot.
In embodiments of the present invention, in the PCI of computing machine slot, be inserted with image pick-up card, image pick-up card connects video camera, and video camera is made a video recording to the gear being placed on workbench.
In step S102, the pretreatment operation to the source images of the tested gear acquiring including greyscale transformation, filtering and noise reduction and Threshold segmentation, obtains the gear parameter of tested gear.
In embodiments of the present invention, the gear parameter of this tested gear is gear graph picture and background image.
In step S103, the gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, obtain the analysis result that comprises the number of teeth that may have defect.
Following have detailed enforcement to describe, do not repeat them here, but not in order to limit the present invention.
As an alternative embodiment of the invention, Fig. 2 shows the pretreatment operation including greyscale transformation, filtering and noise reduction and Threshold segmentation to the source images of the tested gear acquiring that the embodiment of the present invention provides, obtain the realization flow figure of the gear parameter of tested gear, its concrete step is as described below:
In step S201,24 source color images of the tested gear collecting are carried out to greyscale transform process, obtain 8 gray scale gear graph pictures.
In this step, because the gear graph of camera acquisition similarly is coloured image, need to carry out gray processing processing when the digitizing in order to improve computing velocity.Susceptibility principle in conjunction with human eye to color, we have adopted method of weighted mean, that is:
Y＝ω
R*R+ω
G*G+ω
B*B
Wherein ω
r, ω
g, ω
bbe respectively color component R, G, the corresponding weight of B, Y is the pixel value of gray-scale map corresponding point.
Because human eye is the highest to green susceptibility, red susceptibility is taken second place, minimum to blue susceptibility, therefore make ω
g> ω
r> ω
b, will obtain more rational gray level image.General conventional parameter is set to ω
r=0.30, ω
g=0.59, ω
b=0.11 o'clock, the pixel value that obtains gray level image was 256 grades.
In step S202, utilize median filtering algorithm, described 8 gray scale gear graphs are looked like to carry out noise removal.
Wherein, for the noise that removal of images collection brings, select to adopt the method for medium filtering herein.It is a kind of nonlinear signal processing technology that can effectively suppress noise based on sequencing statistical theory, its ultimate principle is to adopt a moving window that contains odd number point, replaces the gray-scale value of window center point pixel by the intermediate value of the gray-scale value of each point in window.First determine the window of an odd number pixel, after each pixel is queued up by gray scale size in window, replace the center pixel value of window with the gray-scale value of intermediate position, its process as shown in Figure 3.
Be formulated as: g (x, y)=median{f (i-k, j-l), (k, l) ∈ W}
Wherein, W is selected template, and selected template window size is generally selected the windows such as 3 × 3,5 × 5 or 7 × 7.Medium filtering is processed the image of front and back as shown in Fig. 4 a and Fig. 4 b.
In step S203, utilize Image binarizing algorithm, gray scale gear graph is looked like to carry out Threshold segmentation, obtain the gear and the background image that separate.
In this step, binary conversion treatment is to utilize object and the difference of its background in gamma characteristic that in image, will extract, image is considered as to the combination in the two class regions (target and background) with different grey-scale, its key is to select rational segmentation threshold.
When the gray-scale value of a pixel exceedes this threshold value, just can say that this pixel belongs to the interested target of people, otherwise belong to background parts.
In gear graph picture, only have a target and background, and the intensity profile of target and background is all more even, its histogram is often rendered as bimodal curve so, as shown in Fig. 5 (a).Select two peak-to-peak a certain threshold value T, just the pixel of image can be divided into pixel group (background) two parts that are greater than the pixel group (target) of T and are less than T.In order to improve the adaptive ability of system, take iteration threshold method herein, first select the initial value of an approximate threshold value as estimated value, then continuously improve this estimated value, utilize initial value spanning subgraph picture, and select new threshold value according to the characteristic of subimage, and then utilize new threshold value to cut apart image, this segmentation result is better than the effect of cutting apart image by initial threshold.If original image is f (x, y), after binaryzation, image is g (x, y), and threshold value is T:
Wherein, the part that value is 1 represents target gear, and the part that value is 0 represents background.After binary conversion treatment, image is as shown in Fig. 5 (b).
As an alternative embodiment of the invention, what Fig. 6 showed that the embodiment of the present invention provides carries out Morphologic Parameters analysis to the gear parameter of the tested gear obtaining, obtain the realization flow figure of the analysis result that comprises the number of teeth that may have defect, its concrete step is as described below:
In step S301, utilize filling algorithm, the hole of the gear graph picture that described Binarization methods is obtained is filled.
In this step, according to the image pattern analysis of collection in worksite, owing to there is hole in binaryzation backgear image, therefore need filling algorithm that the hole in gear graph picture is eliminated.
Its specific practice is: first read in pretreated bianry image (first negate), initiation parameter.In of polygonal region, point starts, and puts until border from inside to outside with given color picture.If specify with a kind of color on border, seed fill algorithm can be processed until run into border color on individual element ground.Conventional four connected domains of seed fill algorithm and eight connected domain technology are carried out padding.Any point in region, arrives any pixel in region by upper and lower, left and right, upper left, lower-left, upper right and eight of bottom rights direction.So repeatedly, until to entire image been scanned.
Area filling completes, and its image after treatment as shown in Figure 7.
In step S302, the binary image after hole is filled, as target, is determined central coordinate of circle, root radius and the radius of addendum of calculating described binary image.
Wherein, the gear center of circle is as the benchmark of gear tooth parameter measurement, accurately determines that the center of circle of tested gear is directly connected to measuring accuracy, and concrete step is following is described in detail for it.
In step S303, according to the central coordinate of circle of the described binary image calculating, root radius and radius of addendum, tooth top is separated with tooth root, detection of gear gear teeth defect, obtains tooth root image and tooth top image.
Wherein, calculate by above step after center, root radius and the radius of addendum of image middle gear, just can detection of gear gear teeth defect, concrete step is following is described in detail for it.
In step S304, by region labeling method, the tooth root image and the tooth top image that obtain are carried out to defect analysis, calculate the number of teeth of gear, determine the defect area of described tooth root and tooth top image.
Wherein, the tooth top figure and the tooth root image that obtain for above step, after region labeling method marking image, just can obtain the actual geometric parameter of gear, and concrete step is following is described in detail for it.
As a specific embodiment of the present invention, Fig. 8 shows binary image after the embodiment of the present invention provides hole is filled as target, the realization flow figure that determines central coordinate of circle, root radius and the radius of addendum of calculating described binary image, its concrete step is as described below:
In step S401, read in the binary image after hole is filled, record the coordinate on gear.
In step S402, on described binary image, search the set of distance described picture centre solstics and closest approach set, obtain the coordinate that can comprise respectively the minimal convex polygon of gear.
In step S403, according to the coordinate of minimal convex polygon, draw out respectively convex polygon.
In step S404, described convex polygon is carried out respectively to the Circular curve fitting of least square method, calculate central coordinate of circle, root radius and the radius of addendum of gear.
In conjunction with Fig. 8 a and Fig. 8 b, followingly provide its concrete realization:
(1) read in bianry image.
(2) using processing the gear obtaining in binary map as target, record the coordinate on gear.
(3) by finding out on edge image apart from the set in solstics, center, acquisition can comprise the coordinate of the minimal convex polygon of gear.
(4) draw out convex polygon according to coordinate, show.And carry out class roundness measurement.Method is:
(5) utilize above-mentioned coordinate figure to utilize least square method to carry out circular curve matching.
(6) calculate the center of circle (x according to the circle of matching
0, y
0) and radius of addendum r
h.
(7) same reason, is exactly the point on dedendum circle apart from the set of center closest approach, and it is carried out to least square fitting, can obtain root radius r
l.
(8) determine respectively inner circle and the cylindrical after matching.Its formula is:
Wherein, Fig. 8 a is the minimum polygon that can comprise gear, according to the circle of its coordinate fitting as shown in Figure 8 b.
Fig. 9 show that the embodiment of the present invention provides according to the central coordinate of circle of the described binary image calculating, root radius and radius of addendum, tooth top is separated with tooth root, detection of gear gear teeth defect, the realization flow figure that obtains tooth root image and tooth top image, its concrete step is as described below:
In step S501, in binary image, draw one taking certain numerical value between root radius and radius of addendum as radius and calculate auxiliary circle.
In this step, read in image, initiation parameter, for convenience of calculation, deletes target gear edge region, retains pure gear target image in former figure.Result is as shown in Figure 10 a.
In edge image, taking gear center hole as the center of circle, draw one taking certain value between root radius and radius of addendum as radius and calculate auxiliary circle.Suppose that radius of addendum is r
h, root radius is r
l.Auxiliary radius of a circle is:
r
M＝α×r
H+β×r
L
Wherein α and β are respectively the coefficient of point circle and dedendum circle, satisfy condition:
0 < α < 1,0 < β < 1, and alpha+beta=1.
In step S502, an illiteracy plate image is set, described illiteracy plate image size is identical with gear graph picture size, and the radius of described illiteracy plate image equals the half of radius of addendum and root radius sum.
Wherein, a secondary blank image is set, big or small rooted tooth wheel image size is identical, taking central point as the center of circle, with r
hfor circular plate that covers of radius, wherein in circle, pixel value is 1, and an ancient official title's pixel value is 0.Suppose α=0.5, β=0.5, obtainable illiteracy plate image radius size is:
the illiteracy plate image obtaining is as shown in Figure 10 b.
In step S503, do difference operation by drawing the gear graph picture and the described illiteracy plate image that calculate after auxiliary circle, obtain tooth top figure image.
Suppose that it is I that gear graph looks like
gear, masking-out image is I
mask, tooth top figure image I
outbe expressed as:
I
H＝I
gear-I
mask
Operation effect is as shown in Figure 10 c, and wherein white portion represents tooth top, and wherein Figure 10 d is the pseudo color image after its actual computation.
In step S504, described illiteracy plate image and the gear graph of drawing after calculating auxiliary circle are looked like to do difference operation, obtain tooth root figure image.
To cover plate figure image subtraction gear graph picture and obtain tooth root image:
I
L＝I
mask-I
gear
Image operation effect is as shown in Figure 10 e, and wherein Figure 10 f is the pseudo color image after its true calculating.
Figure 11 shows the region labeling method of passing through that the embodiment of the present invention provides, the tooth root image and the tooth top image that obtain are carried out to defect analysis, calculate the number of teeth of gear, determine the specific implementation process flow diagram of the defect area of described tooth root and tooth top image, its concrete step is as described below:
In step S601, respectively the tooth root image and the tooth top image that obtain are carried out to zone marker, obtain tooth root image and tooth top image that pseudo-color mode shows.
In step S602, the tooth root image that pseudo-color mode is shown and tooth top image carry out respectively number and single area calculates.
In step S603, tooth top and tooth root area are added up, ask for respectively the mean value of tooth top area and tooth root area.
In step S604, by setting the mode of area threshold, determine tooth root and tooth top defect area.
The concrete implementation procedure of above-described embodiment is as described below:
The tooth top image and the tooth root image that obtain for above step, after region labeling method marking image, just can obtain the actual geometric parameter of gear.
The object of zone marker is to enclose identical mark to the pixel linking together, and different coordinator is enclosed different marks.Zone marker occupies very important status in binary Images Processing.By this processing, each different join domain is distinguished, then just the number of the connected region in image is counted, and can analyze the geometric parameter in each region.Its concrete steps are:
(1) scanning bianry image, runs into while not having tagged target picture number (looking like in vain number) P, adds a new mark (label);
(2) give link together (being identical coordinator) with P add identical mark as number;
(3) further give all with label link together as number add identical mark as number;
(4) until link together be all added mark as number.Such coordinator has just been added identical mark;
(5) turn back to the first step, again search the new tagged picture number that do not have, repeat above-mentioned each step.
(6) the tooth top image after mark shows as shown in Figure 12 a by pseudo-colours mode, and wherein, in Figure 12 a, each tooth top figure should be marked with color, to distinguish different tooth tops, unmarked in figure, but not in order to limit the present invention.
(7) according to number and the area of different mark calculating tooth tops;
The maximal value that wherein tooth top number n is mark, the area of each tooth top is the number of connected region pixel, is formulated as:
n＝max(label)
area(i)＝find(I＝＝i)
(8) individual tooth top area is added up, its histogram represents as shown in Figure 12 b, and by finding out in figure, tooth top number n is 20.
(9) ask for the mean value of tooth top area
be formulated as:
(10) judge defect area.Judge defect area by setting threshold, judgment rule is:
(11) output analysis result.
Therefore, can calculate in this way the number of teeth of gear, and defect area is found out.
In this embodiment, utilize the same method can be in the hope of the geometric parameter in tooth root region.
As one embodiment of the present of invention, calculate after radius of addendum and root radius, can determine module, specific as follows:
(1) modulus m' according to a preliminary estimate.Formula is:
m＝(r
H+r
L)/n
Wherein, n is the number of teeth.
(2) preliminary selected and more approaching master gear modulus.Generally may have 2,3 such standard modules and exist, count, m1, m2, m3 also sorts by size, and is assumed to be m1>m2>m3.
(3) modulus and in standard module table compares, and can obtain the actual modulus of gear through judgement.
By above step, the standard module m of gear just can decide.
Followingly provide a concrete example technical scheme of the embodiment of the present invention described:
For the validity of verification method, we test, and the CCD number of pixels that native system adopts is 1280 × 1040, and image-generating unit area is 4.068mm × 3.686mm.Its experimentation comprises the following steps:
(1) K value is demarcated
Adopt a standard gauge block, the measurement to measuring system under the prerequisite that does not change measurement parameter is demarcated than K=L/L0, and the computer picture size that wherein L is this gauge block, represents with pixel number, the standard size that L0 is this gauge block, and unit is millimeter.
(2) measure
Part image to be measured enters into the measuring system of having demarcated, and the picture size that records part is L1 (pixel count).The measurement size of this part is L2=L1K.
(3) judge
Standard of comparison size and measurement size, the size that judges workpiece whether within tolerance permissible range, thereby measured size is carried out to result judgement.
Figure 13 shows the structured flowchart of the gear defects detection system based on computer vision that the embodiment of the present invention provides, and for convenience of explanation, has only provided the part relevant to the embodiment of the present invention in figure.
Source images acquisition module 11 obtains the source images of the tested gear that image pick-up card collects by PCI slot; Pretreatment operation module 12 is the pretreatment operation including greyscale transformation, filtering and noise reduction and Threshold segmentation to the source images of the tested gear acquiring, and obtains the gear parameter of tested gear; Morphologic Parameters analysis module 13 carries out Morphologic Parameters analysis to the gear parameter of the tested gear obtaining, and obtains the analysis result that comprises the number of teeth that may have defect.
As a specific embodiment of the present invention, as shown in figure 14, described pretreatment operation module 12 specifically comprises:
24 source color images of the tested gear collecting are carried out greyscale transform process by greyscale transform process module 121, obtains 8 gray scale gear graph pictures; Noise removal module 122 is utilized median filtering algorithm, and described 8 gray scale gear graphs are looked like to carry out noise removal; Threshold segmentation module 123 is utilized Image binarizing algorithm, and gray scale gear graph is looked like to carry out Threshold segmentation, obtains the gear and the background image that separate.
As a specific embodiment of the present invention, as shown in figure 15, described Morphologic Parameters analysis module 13 specifically comprises:
Hole packing module 131 utilizes filling algorithm, and the hole of the gear graph picture that described Binarization methods is obtained is filled; Binary image after center of circle radius calculation module 132 is filled hole, as target, is determined central coordinate of circle, root radius and the radius of addendum of calculating described binary image; Tooth root image and tooth top image obtain module 133 according to central coordinate of circle, root radius and the radius of addendum of the described binary image calculating, and tooth top is separated with tooth root, and detection of gear gear teeth defect, obtains tooth root image and tooth top image; Defect analysis module 134, by region labeling method, is carried out defect analysis to the tooth root image and the tooth top image that obtain, calculates the number of teeth of gear, determines the defect area of described tooth root and tooth top image.
As a specific embodiment of the present invention, as shown in figure 16, described center of circle radius calculation module 132 specifically comprises:
Coordinate record module 1321 is read in the binary image after hole is filled, and records the coordinate on gear; Point is searched module 1322 and on described binary image, is searched the set of distance described picture centre solstics and closest approach set, obtains the coordinate that can comprise respectively the minimal convex polygon of gear; Drafting module 1323, according to the coordinate of minimal convex polygon, is drawn out respectively convex polygon; The Fitting Calculation module 1324 is carried out respectively the Circular curve fitting of least square method to described convex polygon, calculate central coordinate of circle, root radius and the radius of addendum of gear;
As a specific embodiment of the present invention, as shown in figure 17, described tooth root image and tooth top image obtain module 133 and specifically comprise:
Calculate auxiliary circle drafting module 1331 in binary image, draw one taking certain numerical value between root radius and radius of addendum as radius and calculate auxiliary circle; Cover plate image setting module 1332 an illiteracy plate image is set, described illiteracy plate image size is identical with gear graph picture size, and the radius of described illiteracy plate image equals the half of radius of addendum and root radius sum; The first difference operation module 1333 is done difference operation by drawing the gear graph picture and the described illiteracy plate image that calculate after auxiliary circle, obtains tooth top figure image; The second difference operation module 1334 looks like to do difference operation by described illiteracy plate image and the gear graph of drawing after calculating auxiliary circle, obtains tooth root figure image;
As a specific embodiment of the present invention, as shown in figure 18, described defect analysis module 134 specifically comprises:
Zone marker module 1341 is carried out zone marker to the tooth root image and the tooth top image that obtain respectively, obtains tooth root image and tooth top image that pseudo-color mode shows; The tooth root image that computing module 1342 shows pseudo-color mode and tooth top image carry out respectively number and single area calculates; Statistical module 1343 is added up tooth top and tooth root area, asks for respectively the mean value of tooth top area and tooth root area; Determination module 1344, by setting the mode of area threshold, is determined tooth root and tooth top defect area.
Above are only system embodiment of the present invention, the function of its each module realizes as described in above-mentioned embodiment of the method, do not repeat them here, but not in order to limit the present invention.
The gear tooth parameter measurement system of the embodiment of the present invention based on machine vision, it is by camera collection image, pass in computing machine through capture card, in conjunction with the powerful data-handling capacity of computing machine, utilize image processing techniques to realize the non-cpntact measurement of gear parameter.Not only measuring speed is fast but also precision is high for the embodiment of the present invention, meet the functional requirement that gear conventional parameter is measured, the automaticity that can greatly enhance productivity and produce in industrial processes in enormous quantities, can meet the measuring accuracy requirement in production reality, and Measuring Time is only several seconds, can meet the requirement to measuring speed in commercial measurement, there is larger theory and practical value.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any amendments of doing within the spirit and principles in the present invention, be equal to and replace and improvement etc., within all should being included in protection scope of the present invention.
Claims (6)
1. the gear defects detection method based on computer vision, is characterized in that, described method comprises the steps:
Obtain the source images of the tested gear that image pick-up card collects by PCI slot;
Pretreatment operation to the source images of the tested gear acquiring including greyscale transformation, filtering and noise reduction and Threshold segmentation, obtains the gear parameter of tested gear;
The gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, obtain the analysis result that comprises the number of teeth that may have defect;
Wherein, described to the source images of the tested gear acquiring the pretreatment operation including greyscale transformation, filtering and noise reduction and Threshold segmentation, the step that obtains the gear parameter of tested gear specifically comprises the steps:
24 source color images of the tested gear collecting are carried out to greyscale transform process, obtain 8 gray scale gear graph pictures;
Utilize median filtering algorithm, described 8 gray scale gear graphs are looked like to carry out noise removal;
Utilize Image binarizing algorithm, gray scale gear graph is looked like to carry out Threshold segmentation, obtain the gear and the background image that separate;
Described the gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, the step of obtaining the analysis result that comprises the number of teeth that may have defect specifically comprises the steps:
Utilize filling algorithm, the hole of the gear graph picture that described Binarization methods is obtained is filled;
Binary image after hole is filled, as target, is determined central coordinate of circle, root radius and the radius of addendum of calculating described binary image;
According to the central coordinate of circle of the described binary image calculating, root radius and radius of addendum, tooth top is separated with tooth root, detection of gear gear teeth defect, obtains tooth root image and tooth top image;
By region labeling method, the tooth root image and the tooth top image that obtain are carried out to defect analysis, calculate the number of teeth of gear, determine the defect area of described tooth root and tooth top image.
2. the gear defects detection method based on computer vision according to claim 1, it is characterized in that, described binary image after hole is filled, as target, determines that the step of central coordinate of circle, root radius and radius of addendum of calculating described binary image specifically comprises the steps:
Read in the binary image after hole is filled, record the coordinate on gear;
On described binary image, search the set of distance described picture centre solstics and closest approach set, obtain the coordinate that can comprise respectively the minimal convex polygon of gear;
According to the coordinate of minimal convex polygon, draw out respectively convex polygon;
Described convex polygon is carried out respectively to the Circular curve fitting of least square method, calculate central coordinate of circle, root radius and the radius of addendum of gear.
3. the gear defects detection method based on computer vision according to claim 1, it is characterized in that, central coordinate of circle, root radius and the radius of addendum of the described binary image that described basis calculates, tooth top is separated with tooth root, detection of gear gear teeth defect, the step that obtains tooth root image and tooth top image specifically comprises the steps:
In binary image, draw one and calculate auxiliary circle, form and cover plate image, described illiteracy plate image size is identical with gear graph picture size, and the radius of described illiteracy plate image equals the half of radius of addendum and root radius sum;
Do difference operation by drawing the gear graph picture and the described illiteracy plate image that calculate after auxiliary circle, obtain tooth top figure image;
Described illiteracy plate image and the gear graph of drawing after calculating auxiliary circle are looked like to do difference operation, obtain tooth root figure image.
4. the gear defects detection method based on computer vision according to claim 1, it is characterized in that, described by region labeling method, the tooth root image and the tooth top image that obtain are carried out to defect analysis, calculate the number of teeth of gear, determine that the step of the defect area of described tooth root and tooth top image specifically comprises:
Respectively the tooth root image and the tooth top image that obtain are carried out to zone marker, obtain tooth root image and tooth top image that pseudo-color mode shows;
The tooth root image that pseudo-color mode is shown and tooth top image carry out respectively number and single area calculates;
Tooth top and tooth root area are added up, asked for respectively the mean value of tooth top area and tooth root area;
By setting the mode of area threshold, determine tooth root and tooth top defect area.
5. the gear defects detection system based on computer vision, is characterized in that, described system comprises:
Source images acquisition module, for obtaining the source images of the tested gear that image pick-up card collects by PCI slot;
Pretreatment operation module, for the pretreatment operation including greyscale transformation, filtering and noise reduction and Threshold segmentation to the source images of the tested gear acquiring, obtains the gear parameter of tested gear;
Morphologic Parameters analysis module, for the analysis result that the gear parameter of the tested gear obtaining is carried out to Morphologic Parameters analysis, obtain comprising the number of teeth that may have defect;
Wherein, described pretreatment operation module specifically comprises:
Greyscale transform process module, for 24 source color images of the tested gear collecting are carried out to greyscale transform process, obtains 8 gray scale gear graph pictures;
Noise removal module, for utilizing median filtering algorithm, looks like to carry out noise removal to described 8 gray scale gear graphs;
Threshold segmentation module, for utilizing Image binarizing algorithm, looks like to carry out Threshold segmentation to gray scale gear graph, obtains the gear and the background image that separate;
Described Morphologic Parameters analysis module specifically comprises:
Hole packing module, for utilizing filling algorithm, the hole of the gear graph picture that described Binarization methods is obtained is filled;
Center of circle radius calculation module, as target, determines central coordinate of circle, root radius and the radius of addendum of calculating described binary image for the binary image after hole is filled;
Tooth root image and tooth top image obtain module, for according to central coordinate of circle, root radius and the radius of addendum of the described binary image calculating, tooth top are separated with tooth root, and detection of gear gear teeth defect, obtains tooth root image and tooth top image;
Defect analysis module, for by region labeling method, carries out defect analysis to the tooth root image and the tooth top image that obtain, calculates the number of teeth of gear, determines the defect area of described tooth root and tooth top image.
6. the gear defects detection system based on computer vision according to claim 5, is characterized in that, described center of circle radius calculation module specifically comprises:
Coordinate record module, for reading in the binary image after hole is filled, records the coordinate on gear;
Point is searched module, for search the set of distance described picture centre solstics and closest approach set on described binary image, obtains the coordinate that can comprise respectively the minimal convex polygon of gear;
Drafting module, for according to the coordinate of minimal convex polygon, draws out respectively convex polygon;
The Fitting Calculation module, for described convex polygon being carried out respectively to the Circular curve fitting of least square method, calculates central coordinate of circle, root radius and the radius of addendum of gear;
Described tooth root image and tooth top image obtain module and specifically comprise:
Cover plate image setting module, at binary image, draw one and calculate auxiliary circle, form and cover plate image, described illiteracy plate image size is identical with gear graph picture size, and the radius of described illiteracy plate image equals the half of radius of addendum and root radius sum;
The first difference operation module, for doing difference operation by drawing the gear graph picture and the described illiteracy plate image that calculate after auxiliary circle, obtains tooth top figure image;
The second difference operation module, for described illiteracy plate image and the gear graph of drawing after calculating auxiliary circle are looked like to do difference operation, obtains tooth root figure image;
Described defect analysis module specifically comprises:
Zone marker module, for respectively the tooth root image and the tooth top image that obtain being carried out to zone marker, obtains tooth root image and tooth top image that pseudo-color mode shows;
Computing module, carries out respectively number and the calculating of single area for tooth root image and tooth top image that pseudo-color mode is shown;
Statistical module, for tooth top and tooth root area are added up, asks for respectively the mean value of tooth top area and tooth root area;
Determination module, for by the mode of setting area threshold, determines tooth root and tooth top defect area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210460169.6A CN102914545B (en) | 2012-11-15 | 2012-11-15 | Gear defect detection method and system based on computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210460169.6A CN102914545B (en) | 2012-11-15 | 2012-11-15 | Gear defect detection method and system based on computer vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102914545A CN102914545A (en) | 2013-02-06 |
| CN102914545B true CN102914545B (en) | 2014-10-01 |
Family
ID=47613007
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210460169.6A Active CN102914545B (en) | 2012-11-15 | 2012-11-15 | Gear defect detection method and system based on computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102914545B (en) |
Families Citing this family (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103528549B (en) * | 2013-10-14 | 2016-05-18 | 广西壮族自治区质量技术监督局珍珠产品质量监督检验站 | A kind of lossless detection method of pearl pearl layer thickness |
| CN103528550B (en) * | 2013-10-14 | 2016-05-18 | 广西壮族自治区质量技术监督局珍珠产品质量监督检验站 | A kind of lossless detection method of pearl pearl layer thickness |
| CN104198496A (en) * | 2014-09-04 | 2014-12-10 | 安徽财经大学 | Automatic cotton defect detection device |
| CN105043722A (en) * | 2015-07-28 | 2015-11-11 | 哈尔滨工程大学 | Reflector reflectivity measuring method |
| CN105069790B (en) * | 2015-08-06 | 2018-09-14 | 潍坊学院 | A kind of gear open defect fast image detection method |
| CN106053485A (en) * | 2016-08-01 | 2016-10-26 | 苏州宙点自动化设备有限公司 | Machine vision-based novel algorithm of intelligent circular inspection of steel ball surface defects |
| CN106524934B (en) * | 2016-11-11 | 2019-11-15 | 北京工业大学 | A kind of high-precision fine module gear vision measuring method |
| CN106442549A (en) * | 2016-12-03 | 2017-02-22 | 河池学院 | Gear defect detection system |
| CN106824806B (en) * | 2016-12-27 | 2018-12-18 | 聊城大学 | The detection method of low module plastic gear based on machine vision |
| CN107036542A (en) * | 2017-05-19 | 2017-08-11 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of ring gear internal-and external diameter appearance detecting method and device |
| CN107610104A (en) * | 2017-08-31 | 2018-01-19 | 南通兴华达高实业有限公司 | Crack detecting method at a kind of elevator compensation chain R based on machine vision |
| CN107796826B (en) * | 2017-09-27 | 2019-12-31 | 浙江大学 | Micro duplicate gear broken tooth defect detection method based on tooth center distance curve analysis |
| CN108051203A (en) * | 2018-02-09 | 2018-05-18 | 华北理工大学 | Gear Fault Diagnosis control system and its diagnostic method based on Labview |
| CN108662977A (en) * | 2018-03-14 | 2018-10-16 | 浙江大学山东工业技术研究院 | A kind of refractory brick geometric dimension measurement method |
| CN108731606A (en) * | 2018-05-17 | 2018-11-02 | 北京化工大学 | A kind of gear parameter rapid detection method based on convexity defect key point |
| CN108593710B (en) * | 2018-06-14 | 2020-02-11 | 湖南大学 | Thermal imaging detection system and method for surface defects of high-reflectivity material |
| CN108844961A (en) * | 2018-08-01 | 2018-11-20 | 佛山科学技术学院 | A kind of temperature controller case vision detection system and method |
| CN109064463B (en) * | 2018-08-06 | 2020-10-20 | 中车青岛四方机车车辆股份有限公司 | Method and system for measuring tooth surface pitting area rate |
| CN109632007B (en) * | 2019-01-17 | 2020-12-04 | 北京理工大学 | Edge point extraction method and gear high-precision vision measurement system |
| TWI702493B (en) * | 2019-09-09 | 2020-08-21 | 英業達股份有限公司 | Testing method and system of socket |
| CN110866916A (en) * | 2019-11-29 | 2020-03-06 | 广州大学 | Machine vision-based photovoltaic cell black-core black-corner detection method, device and equipment |
| CN111047574B (en) * | 2019-12-12 | 2024-06-18 | 苏州韦士肯检测科技有限公司 | Visual inspection defect identification method for cylindrical workpiece |
| CN111353981B (en) * | 2020-02-28 | 2023-03-14 | 广州大学 | Gear detection method and system based on machine vision and storage medium |
| CN111815571B (en) * | 2020-06-16 | 2024-04-02 | 理光软件研究所（北京）有限公司 | Gear-shaped object size detection method, gear-shaped object size detection device, electronic equipment and storage medium |
| CN111754524B (en) * | 2020-06-22 | 2024-01-16 | 厦门大学 | Remote sensing image rapid sea and land segmentation method based on shoreline data |
| CN111882537B (en) * | 2020-07-28 | 2023-12-15 | 研祥智能科技股份有限公司 | Visual detection method and system |
| CN114789743B (en) * | 2022-06-22 | 2022-09-16 | 成都铁安科技有限责任公司 | Method and system for monitoring abnormal running of train wheels |
| CN115035107B (en) * | 2022-08-10 | 2022-11-08 | 山东正阳机械股份有限公司 | Axle gear working error detection method based on image processing |
| CN118196025A (en) * | 2024-03-13 | 2024-06-14 | 深圳个元科技有限公司 | Visual detection method and system for gear appearance defects |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102183528A (en) * | 2011-01-10 | 2011-09-14 | 宁波工程学院 | Defect detection device and method of duplicate gear |
| CN102374996A (en) * | 2011-09-23 | 2012-03-14 | 西安交通大学 | Multicast detection device and method for full-depth tooth side face defects of bevel gear |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100480627C (en) * | 2007-10-26 | 2009-04-22 | 北京航空航天大学 | Steel rail wearing integrative parameter vehicle-mounted dynamic measuring device and method |
| JP2010043941A (en) * | 2008-08-12 | 2010-02-25 | Tokyo Seimitsu Co Ltd | Image inspection apparatus and image inspection method |
-
2012
- 2012-11-15 CN CN201210460169.6A patent/CN102914545B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102183528A (en) * | 2011-01-10 | 2011-09-14 | 宁波工程学院 | Defect detection device and method of duplicate gear |
| CN102374996A (en) * | 2011-09-23 | 2012-03-14 | 西安交通大学 | Multicast detection device and method for full-depth tooth side face defects of bevel gear |
Non-Patent Citations (5)
| Title |
|---|
| JP特开2010-43941A 2010.02.25 |
| 基于机器视觉的齿轮参数测量系统设计;王文成;《机械传动》;20110215;第35卷(第2期);第42页左栏第6-8行，第2节 * |
| 基于机器视觉的齿轮缺陷快速检测;王剑平等;《起重运输机械》;20040520(第5期);41页第1.2节，第2节第1段，43页第3节 * |
| 王剑平等.基于机器视觉的齿轮缺陷快速检测.《起重运输机械》.2004,(第5期), |
| 王文成.基于机器视觉的齿轮参数测量系统设计.《机械传动》.2011,第35卷(第2期), |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102914545A (en) | 2013-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102914545B (en) | Gear defect detection method and system based on computer vision | |
| CN103424409B (en) | Vision detecting system based on DSP | |
| CN103759758B (en) | A kind of method for detecting position of the automobile meter pointer based on mechanical angle and scale identification | |
| CN102999886B (en) | Image Edge Detector and scale grating grid precision detection system | |
| CN103439348B (en) | Remote controller key defect detection method based on difference image method | |
| CN103345755B (en) | A kind of Chessboard angular point sub-pixel extraction based on Harris operator | |
| CN103207987B (en) | A kind of registration recognition methods of pointer instrument | |
| CN103164692B (en) | A kind of special vehicle instrument automatic identification system based on computer vision and method | |
| Zheng et al. | HLU 2-Net: a residual U-structure embedded U-Net with hybrid loss for tire defect inspection | |
| CN103886612B (en) | Automatic water level extraction method and system based on reservoir monitoring camera | |
| CN107014819A (en) | A kind of solar panel surface defects detection system and method | |
| CN105574161B (en) | A kind of brand logo key element recognition methods, device and system | |
| CN101593277A (en) | A kind of complicated color image Chinese version zone automatic positioning method and device | |
| CN112149667A (en) | Method for automatically reading pointer type instrument based on deep learning | |
| CN105973148A (en) | Pen point appearance size standardization detection system based on machine vision | |
| CN102157024B (en) | System and method for on-line secondary detection checking of checking data of large-sheet checking machine | |
| CN102799867A (en) | Meter pointer angle identification method based on image processing | |
| CN110473184A (en) | A kind of pcb board defect inspection method | |
| CN109978872B (en) | White matter microstructure characteristic screening system and method based on white matter fiber tracts | |
| Wang et al. | A computer vision based machine learning approach for fatigue crack initiation sites recognition | |
| CN111932511A (en) | Electronic component quality detection method and system based on deep learning | |
| CN109993154A (en) | The lithium sulfur type instrument intelligent identification Method of substation's simple pointer formula | |
| CN110765963A (en) | Vehicle brake detection method, device, equipment and computer readable storage medium | |
| CN113657339A (en) | Instrument pointer counting and reading method and medium based on machine vision | |
| CN115035251B (en) | Bridge deck vehicle real-time tracking method based on field enhanced synthetic data set |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |