CN105954301A - Bottleneck quality detection method based on machine vision - Google Patents

Abstract

The invention discloses a bottleneck quality detection method based on machine vision. The bottleneck quality detection method comprises the following steps: collecting an image of a to-be-detected bottleneck, and converting the image into a grey-scale map; calculating the gradient vector of each pixel point of the grey-scale map, so as to obtain an edge image of the grey-scale map of the to-be-detected bottleneck; cutting edges according to a grey-scale threshold; cutting an inner ring and an outer ring of the to-be-detected bottleneck by taking the area as a character; respectively calculating circle center coordinates and radiuses of the inner ring and the outer ring, averaging the circle center coordinates to obtain the circle center coordinate of the to-be-detected bottleneck, and setting a radius value range according to the radiuses of the inner ring and the outer ring; acquiring a circle parameter equation according to the circle center coordinate and the radius value range of the to-be-detected bottleneck, carrying out circular scanning on a circular ring according to the circle parameter equation, calculating an average gray value, and drawing an average gray value curve; and analyzing the average gray value curve, and determining the annular ring is not damaged when the variation range of the annular ring is in a certain range. According to the bottleneck quality detection method, the detection efficiency of the bottleneck quality is improved.

Description

A kind of bottleneck quality detection method based on machine vision

Technical field

The present invention relates to technical field of quality detection, be specifically related to the inspection of a kind of bottleneck quality based on machine vision Survey method.

Background technology

The manufacturings such as drinks, beverage, medicine, food have employed filling production lines the most in a large number, and And mostly use vial as the packaging of product.But vial is due to unavoidable in production, transportation It is contaminated and damages, especially needing to use callable vial as industries such as medicated beer, thus vial Have to pass through the operations such as cleaning, detection, fill operation could be entered.In order to overcome foreign body and damage to be brought Harm, it is necessary to the vial before fill is carried out careful detection, industry is referred to as the detection of real bottle.This Detection is typically in darkroom under light by manually carrying out.

Visual detection robot, mainly by the theory and technology of machine vision, comes filling production lines overhead Bottle quality detects.Machine vision, as a comprehensive front subject, obtains people extensive in recent years Paying close attention to, be one of study hotspot, research and application to it are the most active.

Existing empty bottle inspection can only manually detect, and not only efficiency is low, and precision is low, has a strong impact on life Produce line efficiency.

Summary of the invention

In view of this, it is an object of the invention to overcome the drawbacks described above of prior art, it is provided that a kind of based on machine The bottleneck quality detection method of device vision, solves manual detection precision and the inefficient problem of in the past relying on, Improve the technology content of domestic manufacturing industry detection.

The present invention solves the problems referred to above by techniques below means:

A kind of bottleneck quality detection method based on machine vision, comprises the steps:

S1, gather the image of bottleneck to be detected, and be converted into gray-scale map；

S2, Sobel edge edge detective operators is utilized to calculate the gradient vector of each pixel of gray-scale map, Edge image to bottleneck gray-scale map to be detected；

S3, setting gray value threshold range, according to gray value threshold range edge segmentation out；

S4, according to the region after segmentation, utilize area as feature to split the internal ring of bottleneck to be detected and outer Ring；

S5, centroid method is utilized to calculate internal ring and outer shroud central coordinate of circle and radius respectively, outer shroud and the circle of internal ring Heart coordinate is averaged, and obtains the central coordinate of circle of bottleneck to be detected, sets radius according to outer shroud and internal ring radius Span；

S6, according to bottleneck central coordinate of circle to be detected and radius span obtain circle parametric equation, according to The parametric equation of circle carries out circular scanning to annulus, calculates average gray value, draws average gray value curve；

S7, being analyzed average gray value curve, the excursion of annulus within the specific limits, then judges There is not breakage in annulus, otherwise, it is determined that annulus exists breakage, is classified as product of failing.

Further, in step S1, illumination detection system is utilized to gather the image of bottleneck to be detected, described photograph Bright detecting system includes bottleneck to be detected, LED light source, baffle plate, CCD industrial camera, and described baffle plate corresponds to Bottle mouth position to be detected has an opening, and LED light source light shines bottleneck to be detected, is reflected and is entered by opening CCD industrial camera, obtains the image of bottleneck to be detected.

Further, in step S2, Sobel edge edge detective operators is utilized to calculate each pixel of gray-scale map The gradient vector concrete grammar of point is as follows:

Sobel edge edge detective operators comprises the matrix of two groups of 3*3, the most horizontal warp factor And longitudinal direction warp factor

Representing original image with A, Gx and Gy represents the image intensity value through transverse direction and longitudinal direction rim detection respectively, Its formula is as follows:

$G_x=\left[\begin{array}{ccc}-1& 0& +1\\ -2& 0& +2\\ -1& 0& +1\end{array}\right]*A$

$G_y=\left[\begin{array}{ccc}+1& +2& +1\\ 0& 0& 0\\ -1& -2& -1\end{array}\right]*A$

The Grad size of each pixel of image is calculated by below equation:

Wherein, the approximation not extracted square root to improve efficiency to use:

| G |=| G_x|+| G_y|,

Gradient direction is calculated by below equation:

$\mathrm{\θ}=arctan\frac{G_y}{G_x}$

If above angle, θ is equal to zero, i.e. representative image has longitudinal edge at this, relatively right, left Secretly.

Further, in step S3, described gray value threshold range is (25,255).

Further, in step S4, set a threshold value, if area more than threshold value be considered as bottle to be detected There is crackle in mouth, is directly judged as failing.

Further, described threshold value is 60000.

Further, in step S5, detailed process is as follows:

Calculate the external minimum rectangle of outer shroud and obtain the coordinate of four points in outer shroud upper and lower, left and right, then pass through Centroid method calculating midpoint i.e. outer shroud central coordinate of circle:

Outer shroud radius calculation formula:

Wherein, x₂For the abscissa of point, x on the right of outer shroud₁For the abscissa of outer shroud left side point, y₂Following for outer shroud The vertical coordinate of point, y₁For the vertical coordinate of edge point on outer shroud；

The most in like manner can obtain internal ring central coordinate of circle: (center_{X (interior)}, center_{Y (interior)}) and interior Ring radius R_In；

The central coordinate of circle of outer shroud and internal ring is averaged, obtains the central coordinate of circle of bottleneck to be detected: (center_X, center_Y)；

Radius span is set as R according to outer shroud and internal ring radius.

Further, in step S6, the parametric equation of circle is:

$\left\{\begin{array}{c}x={\mathrm{center}}_X+R*\cos \mathrm{\ψ}\\ y={\mathrm{center}}_Y-R*\sin \mathrm{\ψ}\end{array}\right.$

Wherein, R is [R_In–3,R_Outward+ 3], central angle ψ is [0,360 °]；

I (x, y) be annulus (x, y) gray value that coordinate points is corresponding, then:

mean_IFor average gray value.

Further, in step S7, the excursion of annulus within 40 pixels, then judges annulus There is not breakage.

The bottleneck quality detection method based on machine vision of the present invention solves and relies on manual detection precision in the past With inefficient problem, improve the technology content of domestic manufacturing industry detection.Meanwhile, the successful reality of the method Industrial robot hand-eye system, logistics transportation industry, packing business, optical detection are had by row with processing and other fields Well application prospect.

Accompanying drawing explanation

Fig. 1 is the flow chart of present invention bottleneck quality based on machine vision detection method；

The illumination detection system that the bottleneck quality detection method based on machine vision that Fig. 2 is the present invention is used Structural representation；

The bottleneck gray-scale map that Fig. 3 is gathered by using Fig. 2 in embodiments of the invention；

Fig. 4 is the average gray curve chart that in Fig. 3, every width bottleneck gray-scale map is corresponding.

Detailed description of the invention

Understandable, below in conjunction with accompanying drawing for enabling the above-mentioned purpose of the present invention, feature and advantage to become apparent from With specific embodiment, technical scheme is described in detail.It is pointed out that described Embodiment is only a part of embodiment of the present invention rather than whole embodiments, based on the reality in the present invention Execute example, the every other reality that those of ordinary skill in the art are obtained under not making creative work premise Execute example, broadly fall into the scope of protection of the invention.

As it is shown in figure 1, a kind of bottleneck quality detection method based on machine vision, comprise the steps:

S1, gather the image of bottleneck to be detected, and be converted into gray-scale map；

As in figure 2 it is shown, utilize illumination detection system to gather the image of bottleneck to be detected, described illumination detection is System includes bottleneck to be detected, LED light source, baffle plate, CCD industrial camera, and described baffle plate corresponds to bottle to be detected Having an opening at Kou, LED light source light shines bottleneck to be detected, is reflected and enters CCD industry by opening Camera, obtains the image of bottleneck to be detected, and is converted into gray-scale map, as shown in Figure 3.

S2, Sobel edge edge detective operators is utilized to calculate the gradient vector of each pixel of gray-scale map, Edge image to bottleneck gray-scale map to be detected；

Technically, it is a discreteness difference operator, for the approximation of the gray scale of arithmograph image brightness function Value.This operator is used, it will produce corresponding gray scale vector or its law vector in any point of image；

Sobel edge edge detective operators comprises the matrix of two groups of 3*3, the most horizontal warp factor And longitudinal direction warp factorIt and image are made planar convolution, can draw respectively laterally and Longitudinal brightness difference approximation；

If representing original image with A, Gx and Gy represents the image ash through transverse direction and longitudinal direction rim detection respectively Angle value, its formula is as follows:

$G_x=\left[\begin{array}{ccc}-1& 0& +1\\ -2& 0& +2\\ -1& 0& +1\end{array}\right]*A$

$G_y=\left[\begin{array}{ccc}+1& +2& +1\\ 0& 0& 0\\ -1& -2& -1\end{array}\right]*A$

The Grad of each pixel of image is calculated by below equation: Generally, the approximation not extracted square root to improve efficiency to use:

| G |=| G_x+G_y|,

Then gradient direction can be calculated by below equation:

$\mathrm{\θ}=arctan\frac{G_y}{G_x}$

If above angle, θ is equal to zero, i.e. representative image has longitudinal edge at this, and left is dark compared with right.

S3, setting gray value threshold range, according to gray value threshold range edge segmentation out；

Described gray value threshold range is (25,255).

S4, according to the region after segmentation, utilize area as feature to split the internal ring of bottleneck to be detected and outer Ring；

According to the region after segmentation, utilize area as feature to split internal ring and the outer shroud of bottleneck.General next Saying, with a collection of bottle under identical illumination detection system, annulus area has fixed range, i.e. edge Area also have fixed range.When crackle occurs in bottleneck, internal ring and outer shroud after edge segmentation can connect Together, the b from Fig. 3 can be seen that；Here a threshold decision can be set, if area is more than The situation being considered as crackle occur of threshold value, can directly be judged as failing, according to experimental data, set here Determining area threshold is 60000.

S5, centroid method is utilized to calculate internal ring and outer shroud central coordinate of circle and radius respectively, outer shroud and the circle of internal ring Heart coordinate is averaged, and obtains the central coordinate of circle of bottleneck to be detected, sets radius according to outer shroud and internal ring radius Span；

After extracting annular edge, calculate the external minimum rectangle of outer shroud and obtain four, outer shroud upper and lower, left and right point Coordinate, then by centroid method calculate midpoint i.e. outer shroud central coordinate of circle:

Outer shroud radius calculation formula:

Wherein, x₂For the abscissa of point, x on the right of outer shroud₁For the abscissa of outer shroud left side point, y₂Following for outer shroud The vertical coordinate of point, y₁For the vertical coordinate of edge point on outer shroud；

The most in like manner can obtain internal ring central coordinate of circle: (center_{X (interior)}, center_{Y (interior)}) and interior Ring radius R_In；

The central coordinate of circle of outer shroud and internal ring is averaged, obtains the central coordinate of circle of bottleneck to be detected: (center_X, center_Y)；

Radius span is set as R according to outer shroud and internal ring radius.

S6, according to bottleneck central coordinate of circle to be detected and radius span obtain circle parametric equation, according to The parametric equation of circle carries out circular scanning to annulus, calculates average gray value, draws average gray value curve；

Owing to bottleneck image can substantially regard the annulus of a standard as, according to round parametric equation to this annulus Carry out circular scanning, calculate average gray value, draw average gray value curve, as shown in Figure 4,

The parametric equation of circle is:

$\left\{\begin{array}{c}x={\mathrm{center}}_X+R*\cos \mathrm{\ψ}\\ y={\mathrm{center}}_Y-R*\sin \mathrm{\ψ}\end{array}\right.$

Wherein, R is (R_In–3,R_Outward+ 3), central angle ψ is (0,360 °)；

I (x, y) be annulus (x, y) gray value that coordinate points is corresponding, then:

mean_IFor average gray value.

S7, being analyzed average gray value curve, the excursion of annulus within the specific limits, then judges There is not breakage in annulus, otherwise, it is determined that annulus exists breakage, is classified as product of failing.

The excursion of annulus within 40 pixels, then judges that annulus does not exist breakage.

A in Fig. 3 is the doughnut of standard, a from its average gray curve chart Fig. 4 it can be seen that Its excursion is 16 pixels, for qualifying product；And the excursion of remaining 4 width doughnut is the most relatively Greatly, it is known that the annulus of those a few width figures all exists breakage, therefore corresponding bottleneck can be classified as product of failing.

The bottleneck quality detection method based on machine vision of the present invention solves and relies on manual detection precision in the past With inefficient problem, improve the technology content of domestic manufacturing industry detection.Meanwhile, the successful reality of the method Industrial robot hand-eye system, logistics transportation industry, packing business, optical detection are had by row with processing and other fields Well application prospect.

Embodiment described above only have expressed the several embodiments of the present invention, and it describes more concrete and detailed, But therefore can not be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that, for this area Those of ordinary skill for, without departing from the inventive concept of the premise, it is also possible to make some deformation and Improving, these broadly fall into protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be with appended Claim is as the criterion.

Claims (9)

1. a bottleneck quality detection method based on machine vision, it is characterised in that comprise the steps:

S1, gather the image of bottleneck to be detected, and be converted into gray-scale map；

S2, Sobel edge edge detective operators is utilized to calculate the gradient vector of each pixel of gray-scale map, Edge image to bottleneck gray-scale map to be detected；

S3, setting gray value threshold range, according to gray value threshold range edge segmentation out；

S4, according to the region after segmentation, utilize area as feature to split the internal ring of bottleneck to be detected and outer Ring；

S5, centroid method is utilized to calculate internal ring and outer shroud central coordinate of circle and radius respectively, outer shroud and the circle of internal ring Heart coordinate is averaged, and obtains the central coordinate of circle of bottleneck to be detected, sets radius according to outer shroud and internal ring radius Span；

S6, according to bottleneck central coordinate of circle to be detected and radius span obtain circle parametric equation, according to The parametric equation of circle carries out circular scanning to annulus, calculates average gray value, draws average gray value curve；

S7, being analyzed average gray value curve, the excursion of annulus within the specific limits, then judges There is not breakage in annulus, otherwise, it is determined that annulus exists breakage, is classified as product of failing.

Bottleneck quality detection method based on machine vision the most according to claim 1, it is characterised in that In step S1, utilizing illumination detection system to gather the image of bottleneck to be detected, described illumination detection system includes Bottleneck to be detected, LED light source, baffle plate, CCD industrial camera, described baffle plate has corresponding to bottle mouth position to be detected One opening, LED light source light shines bottleneck to be detected, is reflected and enters CCD industrial camera by opening, Obtain the image of bottleneck to be detected.

Bottleneck quality detection method based on machine vision the most according to claim 2, its feature exists In, in step S2, utilize the gradient of each pixel of Sobel edge edge detective operators calculating gray-scale map to vow Measurer body method is as follows:

Sobel edge edge detective operators comprises the matrix of two groups of 3*3, the most horizontal warp factor And longitudinal direction warp factor

Representing original image with A, Gx and Gy represents the image intensity value through transverse direction and longitudinal direction rim detection respectively, Its formula is as follows:

$G_x=\left[\begin{array}{ccc}-1& 0& +1\\ -2& 0& +2\\ -1& 0& +1\end{array}\right]*A$

$G_y=\left[\begin{array}{ccc}+1& +2& +1\\ 0& 0& 0\\ -1& -2& -1\end{array}\right]*A$

The gray value of each pixel of image is calculated by below equation:

Wherein, the approximation not extracted square root to improve efficiency to use:

| G |=| G_x+G_y|,

Gradient direction is calculated by below equation:

$\mathrm{\θ}=\arctan \frac{G_y}{G_x}$

If above angle, θ is equal to zero, i.e. representative image has longitudinal edge at this, and left is dark compared with right.

Bottleneck quality detection method based on machine vision the most according to claim 3, it is characterised in that In step S3, described gray value threshold range is (25,255).

Bottleneck quality detection method based on machine vision the most according to claim 4, it is characterised in that In step S4, set a threshold value, if area crackle occurs, directly more than the bottleneck to be detected that is considered as of threshold value Connect and be judged as failing.

Bottleneck quality detection method based on machine vision the most according to claim 5, it is characterised in that Described threshold value is 60000.

Bottleneck quality detection method based on machine vision the most according to claim 6, it is characterised in that In step S5, detailed process is as follows:

Calculate the external minimum rectangle of outer shroud and obtain the coordinate of four points in outer shroud upper and lower, left and right, then pass through Centroid method calculating midpoint i.e. outer shroud central coordinate of circle:

Outer shroud radius calculation formula:

Wherein, x₂For the abscissa of point, x on the right of outer shroud₁For the abscissa of outer shroud left side point, y₂Following for outer shroud The vertical coordinate of point, y₁For the vertical coordinate of edge point on outer shroud；

The most in like manner can obtain internal ring central coordinate of circle: (center_{X (interior)}, center_{Y (interior)}) and interior Ring radius R_In；

The central coordinate of circle of outer shroud and internal ring is averaged, obtains the central coordinate of circle of bottleneck to be detected: (center_X, center_Y)；

Radius span is set as R according to outer shroud and internal ring radius.

Bottleneck quality detection method based on machine vision the most according to claim 7, it is characterised in that In step S6, the parametric equation of circle is:

$\left\{\begin{array}{c}x={\mathrm{center}}_X+R*cos\mathrm{\ψ}\\ y={\mathrm{center}}_Y-R*sin\mathrm{\ψ}\end{array}\right.$

Wherein, R is (R_In–3,R_Outward+ 3), central angle ψ is (0,360 °)；

I (x, y) be annulus (x, y) gray value that coordinate points is corresponding, then:

mean_IFor average gray value.

Bottleneck quality detection method based on machine vision the most according to claim 8, it is characterised in that In step S7, the excursion of annulus within 40 pixels, then judges that annulus does not exist breakage.