CN112070738A - Method and system for detecting nose bridge of mask - Google Patents

Abstract

The invention discloses a method and a system for detecting a nose bridge of a mask. According to the method, an initial image of the mask is processed in a binarization processing mode, a minimum external rectangle of a mask outline set is found out, parameter information of the mask to be detected is further obtained, a detection area of a nose bridge is obtained through an outer area rectangle and an inner area rectangle, an adaptive binarization image is generated through filtering and adaptive threshold processing on the image, a connected domain confirmation is carried out on the adaptive binarization image in a detection range, a nose bridge outline set is obtained, the outline size and the center coordinate of the nose bridge are obtained through a minimum external rectangle method, then the length of the nose bridge to be detected is compared with the standard length of the nose bridge, the center coordinate of the mask after affine transformation and the center coordinate of the nose bridge after affine transformation are compared, and therefore the quality of the mask to be detected is effectively judged.

Description

A kind of mask nose bridge detection method and system

technical field

The invention relates to the technical field of production and detection of masks, in particular to a method and system for detecting the bridge of the nose of a mask.

Background technique

With the process of industrialization and urbanization, air pollution has become more and more serious. Mask products are commonly used by people to resist air pollution. In the areas with severe haze and sandstorms in the north, masks have become an indispensable daily dependency. In the mass production process, due to various reasons such as raw materials, production machines and personnel operations, there will be many kinds of defective mask products, such as the bridge of the nose is not installed, etc. If such defective products flow to the market, it will bring users. Many inconveniences will bring economic losses to mask manufacturers.

Since the production of masks has been fully automated, the production requirements and quality requirements are getting higher and higher. Generally, mask manufacturers rely on the human eye to distinguish them one by one, and the efficiency is extremely low. As a result, the current mask quality inspection process is time-consuming and laborious, and it is easy to detect personnel. Cause visual fatigue, make products have hidden quality risks, and cause a large number of product rework and raw material waste.

In the detection of the nose bridge of the mask, one of the detection methods is to use a fixed threshold method to binarize the entire image, and then select the nose bridge according to the area or length in the binarized image or in a limited local position. The disadvantage of this method is that due to the uncertainty of the position of the mask when taking pictures and the consistency problem during the production of the mask, the area or length will be within the required range, but it is not actually the bridge of the nose, which is prone to misjudgment. , and due to the imaging of the nose bridge, the contrast difference between the nose bridge and the surrounding is not obvious, which will also lead to the failure of the nose bridge detection or the large detection error.

SUMMARY OF THE INVENTION

Based on this, it is necessary to provide a method and system for detecting the nose bridge of a mask in view of the shortcomings of the prior art, which can automatically detect the mask, avoid the phenomenon of missed detection and false detection, improve the efficiency of detection and the quality of the mask, and reduce the production of masks. the cost of.

In order to solve the above-mentioned technical problems, the technical scheme adopted in the present invention is: a method for detecting the bridge of the nose of a mask, which comprises the following steps:

(1), collect the initial image Img1 of the mask to be detected;

(2), establish image coordinate system, carry out binarization processing to initial image Img1 according to fixed threshold algorithm, obtain initial binarization image Img2;

(3), construct the circular structural element, carry out morphological closure processing to the initial binarized image Img2, obtain the binarized image Img3 of closed operation;

(4), construct rectangular structural element, carry out morphological opening processing to the binarized image Img3 of closing operation, obtain the binarized image Img4 of opening operation;

(5), carry out connected domain judgment on the binarized image Img4 of the opening operation by the Two-Pass algorithm, obtain all connected domains in the binarized image Img4 of the opening operation, and according to the binary image Img4 of the opening operation. The connected domain finds the contour corresponding to the connected domain in the binarized image Img4, and obtains the initial mask contour set Contour _s ;

计算初始口罩轮廓集合Contour_s中符合要求的轮廓数量，获得口罩轮廓集合Contour_m；其中，预设的口罩周长L_ms为已转化成像素长度的标准口罩周长，CL_m为初始口罩轮廓集合Contour_s中连通域对应的轮廓长度；(6), according to the preset mask circumference L _ms , through the formula

Calculate the number of contours that meet the requirements in the initial mask contour set Contour _s , and obtain the mask contour set Contour _m ; wherein, the preset mask perimeter L _ms is the standard mask perimeter that has been converted into pixel length, and CL _m is the initial mask contour set. Contour length corresponding to the connected domain in Contour _s ;

(7), judge the contour quantity in the mask contour set Contour _m , if the contour quantity in the mask contour set Contour _m is 0 or greater than 1, then go to step (22); If the contour quantity in the mask contour set Contour _m is 1, then go to step (8);

(8), obtain the minimum circumscribed rectangle of the mask contour set Contour _m , obtain the length Ls of the mask to be detected according to the length and width of the minimum circumscribed rectangle, the rectangle center coordinates in the image coordinate system, and the rectangular angle in the image coordinate system , width Ws, the center coordinates (Xs, Ys) in the image coordinate system, the angle As parameter information in the image coordinate system, according to the preset mask size, to determine whether the size of the mask to be detected meets the requirements, if the size of the mask to be detected is If the length and width are within the preset mask size range, then go to step (8); if the length and width of the mask to be detected exceed the preset mask size range, then go to step (22); wherein, in the image coordinate system The rectangle angle in is the angle formed by the long side of the smallest circumscribed rectangle and the X axis in the image coordinate system;

(9), under the condition that the center coordinates (Xs, Ys) and the angle As of the mask to be detected obtained in step (8) remain unchanged, take the four vertices of the mask to be detected as the reference points respectively along the length direction of the mask to be detected Subtract the preset length Lr1, and then subtract the preset width Wr1 along the width direction of the mask to be detected to generate the outer area rectangle Ro for nose bridge detection;

(10), under the condition that the center coordinates (Xs, Ys) and the angle As of the mask to be detected obtained in step (8) remain unchanged, take the four vertices of the mask to be detected as the reference points respectively along the length direction of the mask to be detected Subtract the preset length Lr2, and then subtract the preset width Wr2 along the width direction of the mask to be detected to generate the inner area rectangle Ri for nose bridge detection;

(11), perform regional subtraction on the outer area rectangle Ro of nose bridge detection and the inner area rectangle Ri of nose bridge detection to obtain the detection area Rnose of the bridge of the nose, and then use the gray value of the pixel of the image area other than the detection area Rnose of the bridge of the nose. Set to 0 to obtain the image Img5 corresponding to the position of the bridge of the nose;

(12), carry out linear transformation to the image Img5 corresponding to the position of the bridge of the nose, obtain the enhanced image Img6, and obtain the gray value corresponding to the pixel point in the enhanced image Img6;

得到均值滤波图像Img7，其中，src1(x,y)为图像Img6在(x,y)处像素点的灰度值，dst1(x,y)为图像Img7在(x,y)处像素点的灰度值，h(k,l)为滤波核，k，l表示滤波核的大小，k，l均为奇数；(13), carry out mean filter processing to the enhanced image Img6, according to the formula

Obtain the mean filtered image Img7, where src1(x, y) is the gray value of the pixel point at (x, y) of the image Img6, and dst1(x, y) is the pixel point of the image Img7 at (x, y) The gray value of the point Gray value, h(k,l) is the filter kernel, k,l is the size of the filter kernel, k,l are odd numbers;

(14), carry out self-adaptive threshold value processing to mean filter image Img7, obtain self-adaptive binarization image Img8;

(15), constructing a rectangular structural element, and performing a binary morphological preprocessing operation of first eroding and then dilating the adaptive binarized image Img8 to obtain an image Img9;

(16), carry out connected domain judgment on image Img9 by Two-Pass algorithm, obtain all connected domains in the image Img9, and find out the corresponding contour of the connected domain in the image Img9 according to the connected domain in the image Img9, obtain the initial nose bridge outline set Contour _c ;

计算初始鼻梁轮廓集合Contour_c中符合要求的轮廓数量，获得鼻梁轮廓集合Contour_n；(17), according to the preset length of the bridge of the nose, through the formula

Calculate the number of contours that meet the requirements in the initial nose bridge contour set Contour _c , and obtain the nose bridge contour set Contour _n ;

Wherein, the preset nose bridge length L _ns is the standard nose bridge length that has been converted into pixel length, and CL _s is the contour length corresponding to the connected domain in the initial nose bridge contour set Contour _c ;

(18), judge the contour quantity in the nose bridge contour set Contour _n , if the contour quantity in the nose bridge contour set Contour _n is 0 or greater than 1, then go to step (22); If the contour quantity in the nose bridge contour set Contour _m is 1, then go to step (19);

(19), obtain the minimum circumscribed rectangle of the nose bridge contour set Contour _n , obtain the length of the nose bridge to be detected and the center coordinates (Xns, Yns) of the nose bridge to be detected according to the length of the minimum circumscribed rectangle and the center coordinates of the rectangle in the image coordinate system ;

(20), compare the length of the bridge of nose to be detected according to the length of the bridge of the nose to be detected obtained in step (18), judge whether the length of the bridge of the nose to be detected meets the requirements, when the length of the bridge of the nose to be detected deducts the standard length of the bridge of the nose When the absolute value is not greater than the preset nose bridge length tolerance, then go to step (21); When the absolute value of the length of the nose bridge to be detected minus the standard length of the nose bridge is greater than the preset nose bridge length tolerance, then go to step (22) );

(21), construct a transformation matrix M _c , perform affine transformation on the mask center coordinates (Xs, Ys) and the center coordinates (Xns, Yns) of the bridge of the nose to be detected according to the constructed transformation matrix M _c to obtain the mask after affine transformation The center coordinates (Xsc, Ysc) and the center coordinates (Xnsc, Ynsc) of the nose bridge after affine transformation, obtain the absolute value Labs=|Ysc-Ysc|, if the absolute value Labs ≤ the preset nose bridge deviation value, then judge The mask to be detected is a good product, and the mask to be detected is delivered to the product qualified area; if _Labs > the preset nose bridge offset value, then go to step (22);

(22) Determine that the mask to be tested is a defective product, and send the mask to be tested to the unqualified area of the product.

A mask nose bridge detection system, comprising

The image acquisition unit is used to collect the initial image Img1 of the mask to be detected;

The image processing unit is used to obtain an initial binarized image Img2 through a fixed threshold algorithm, obtain a closed-operation binarized image Img3 through morphological closing processing, obtain an open-operation binarized image Img4 through morphological opening processing, and obtain an open-operation binarized image Img4 through morphological closing processing. Transform to obtain an enhanced image Img6, obtain a mean value filtered image Img7 through mean filtering, obtain an adaptive binarized image Img8 through adaptive threshold processing, and obtain an image Img9 through a binary morphological preprocessing operation;

The image labeling unit is used to judge the connected domain of the binarized image Img4 of the split operation by the Two-Pass algorithm to obtain the mask contour set Contour _m and to judge the connected domain of the image Img9 by the Two-Pass algorithm to obtain the initial nose bridge contour set Contour _c ;

The analysis and calculation unit is used to obtain the length Ls, width Ws of the mask to be detected, the center coordinates (Xs, Ys) of the mask to be detected in the image coordinate system, the angle As of the mask to be detected in the image coordinate system, the length of the bridge of the nose to be detected, The center coordinates of the bridge of the nose to be detected (Xns, Yns);

The image determination unit is used to determine whether the size of the mask to be detected meets the requirements, whether the length of the bridge of the nose to be detected meets the requirements, and the determination is based on the center coordinates (Xsc, Ysc) of the mask after affine transformation and the bridge of the nose after affine transformation. Whether the absolute value Labs=|Ysc-Ysc| obtained by the center coordinates (Xnsc, Ynsc) meets the requirements.

To sum up, a method and system for detecting the bridge of the nose of a mask of the present invention processes the initial image of the mask by means of binarization, and then finds the minimum circumscribed rectangle of the mask contour set Contour _m , and then obtains the length Ls of the mask to be detected. , the width Ws, the center coordinates (Xs, Ys) in the image coordinate system, the angle As parameter information in the image coordinate system, and then obtain the detection area Rnose of the nose bridge through the outer area rectangle Ro and the inner area rectangle Ri, and then determine the nose bridge. In the detection area _Rnose , the adaptive binarized image is generated by filtering the image, and the adaptive binarized image is generated by adaptive thresholding. The external dimensions and center coordinates of the bridge of the nose are obtained by the minimum circumscribed rectangle method, and then the length of the bridge to be detected is compared with the standard length of the bridge of the nose, and the center coordinates (Xsc, Ysc) of the mask after affine transformation and the affine transformation are compared. The center coordinates (Xnsc, Ynsc) of the posterior bridge of the nose are compared, so as to effectively determine the length of the bridge of the nose and the quality of the location of the bridge of the nose of the mask to be detected.

Description of drawings

Fig. 1 is the structural representation of a kind of mask nose bridge detection system of the present invention.

Detailed ways

In order to further understand the features, technical means, and specific goals and functions of the present invention, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

A method for detecting the bridge of the nose of a mask of the present invention comprises the following steps:

(1), the mask to be detected on the conveyor belt is photographed, and the initial image Img1 of the mask to be detected is collected;

(2), establish the image coordinate system, carry out binarization processing to the initial image Img1 according to the fixed threshold algorithm, obtain the initial binarized image Img2, wherein, the fixed threshold algorithm formula is:

Among them, src(x, y) is the gray value of the pixel at (x, y) of the initial image Img1, and dst(x, y) is the gray value of the pixel at (x, y) of the initial binarized image Img2 degree value, thresh and maxval are the preset grayscale values of mask extraction, in this embodiment, thresh is set to 210, and maxval is set to 255, that is, when the initial image Img1 is binarized according to the fixed threshold algorithm, When src(x,y)>thresh, dst(x,y)=0 in the initial binarized image Img2, and when src(x,y)≤thresh, dst(x,y) in the initial binarized image Img2 )=maxval; wherein, in the initial binarized image Img2, the gray value of the pixel at (x, y) is only 0 or 255.

(3), construct a circular structural element with a radius of 9 pixels in length, the center of the circular structural element is located at the center of the circular structural element, and perform morphological closing processing on the initial binarized image Img2 for filling the initial binarization The small holes and bright interference points at the solder joints on the image Img2 obtain a closed operation binarized image Img3; wherein, the morphological closed processing method is to first dilate and then corrode the initial binarized image Img2. The circular structuring elements and the corroded circular structuring elements are processed on the initial binarized image Img2, the corroded circular structuring elements and the dilated circular structuring elements are all set to the shape of a solid circle, and the radius of the solid circle is set to 9 pixels long.

(4), construct a rectangular structural element with a size of 25 pixels*25 pixels, the center of the rectangular structural element is located in the center of the rectangular structural element, perform morphological opening processing on the binarized image Img3 of the closing operation, and obtain the binarization of the opening operation. The image Img4 is used to remove the infection effect of the ear belt and the background belt in the binarized image Img3 of the closed operation; wherein, the morphological opening processing method is to first erode the closed operation of the binarized image Img3, and then dilate the operation method , the eroded rectangular structuring elements and the expanded rectangular structuring elements are processed on the closed-operation binarized image Img3, the eroded rectangular structuring elements and the expanded rectangular structuring elements are both set as solid rectangular shapes, and the length of the solid rectangular shape, The width is set to 25 pixels long and 25 pixels long, respectively.

(5), carry out connected domain judgment on the binarized image Img4 of the opening operation by the Two-Pass algorithm, obtain all connected domains in the binarized image Img4 of the opening operation, and according to the binary image Img4 of the opening operation. The connected domain finds the contour corresponding to the connected domain in the binarized image Img4, obtains the initial mask contour set Contour _s , and the contour corresponding to the connected domain in the binarized image Img4 of the opening operation refers to a curve composed of multiple pixel points, The length of the contour corresponding to the connected domain is obtained according to the number of pixels on the curve. For example, the length of the contour corresponding to the connected domain is 100, which means that the contour corresponding to the connected domain is composed of 100 pixels. Among them, the Two-pass algorithm is a known technology, and it is unnecessary to describe it here.

计算初始口罩轮廓集合Contour_s中符合要求的轮廓数量，获得口罩轮廓集合Contour_m；其中，预设的口罩周长L_ms为已转化成像素长度的标准口罩周长，CL_m为初始口罩轮廓集合Contour_s中连通域对应的轮廓长度；在本实施例中，以毫米为单位，会先将预设的口罩周长转换成像素长度，即像素点数量，转化公式为：(6), according to the preset mask circumference L _ms , through the formula

Calculate the number of contours that meet the requirements in the initial mask contour set Contour _s , and obtain the mask contour set Contour _m ; wherein, the preset mask perimeter L _ms is the standard mask perimeter that has been converted into pixel length, and CL _m is the initial mask contour set. The contour length corresponding to the connected domain in the Contour _s ; in this embodiment, in millimeters, the preset mask perimeter will first be converted into a pixel length, that is, the number of pixels, and the conversion formula is:

Lpixel=Lreal/Rptr

Among them, Lpixel represents the number of pixels, Lreal represents the preset mask perimeter in millimeters, Rptr is the actual ratio of pixels, that is, a pixel represents the actual length in millimeters, and the value of the actual pixel ratio Rptr is calibrated by the camera Obtained, in this embodiment, the value of the actual pixel ratio Rptr is obtained by calibrating the camera by placing the calibration plate printed with two black dots under the camera to take pictures to obtain a calibration image, wherein the radius of the black dots on the calibration plate is 6mm , The distance between the two black dots is 80mm, and the actual pixel ratio Rptr is calculated according to the calibration image.

(7), judge the contour quantity in the mask contour set Contour _m , if the contour quantity in the mask contour set Contour _m is 0 or greater than 1, then go to step (22); If the contour quantity in the mask contour set Contour _m is 1, then go to step (8).

(8) According to the minimum circumscribed rectangle algorithm, obtain the minimum circumscribed rectangle of the mask contour set Contour _m , and obtain according to the length and width of the minimum circumscribed rectangle, the coordinates of the center of the rectangle in the image coordinate system, and the angle of the rectangle in the image coordinate system. Parameter information such as the length Ls, width Ws of the mask to be detected, the center coordinates (Xs, Ys) in the image coordinate system, and the angle As in the image coordinate system, according to the preset mask size, to determine whether the size of the mask to be detected conforms to Requirements, if the length and width of the mask to be detected are within the preset mask size range, then go to step (8); if the length and width of the mask to be detected exceed the preset mask size range, then go to step (22) Wherein, the rectangle angle in the image coordinate system is the angle formed by the long side of the minimum circumscribed rectangle and the X-axis in the image coordinate system, and the preset size range of the mask includes the preset length of the mask and the preset width of the mask , the preset length of the mask is ±2mm of the length of the standard mask, and the preset width of the mask is ±2mm of the width of the standard mask. The length and width of the standard mask can be determined by the length and width of the masks sold in the market or according to The length and width of masks commonly used in the art are determined, and the minimum circumscribed rectangle algorithm is a known technology, and it is unnecessary to repeat them here.

(9), under the condition that the center coordinates (Xs, Ys) and the angle As of the mask to be detected obtained in step (8) remain unchanged, take the four vertices of the mask to be detected as the reference points respectively along the length direction of the mask to be detected The preset length Lr1 is subtracted, and then the preset width Wr1 is subtracted along the width direction of the mask to be detected to generate the outer area rectangle Ro for nose bridge detection.

(10), under the condition that the center coordinates (Xs, Ys) and the angle As of the mask to be detected obtained in step (8) remain unchanged, take the four vertices of the mask to be detected as the reference points respectively along the length direction of the mask to be detected Subtract the preset length Lr2, and then subtract the preset width Wr2 along the width direction of the mask to be detected to generate the inner area rectangle Ri for nose bridge detection, where Lr1>Lr2, Wr1<Wr2, usually the value range of Lr1 is 16-26mm , the value range of Lr2 is 15-25mm, the value range of Wr1 is 0-5mm, and the value range of Wr2 is 15-25mm.

(11), perform regional subtraction on the outer area rectangle Ro of nose bridge detection and the inner area rectangle Ri of nose bridge detection to obtain the detection area Rnose of the bridge of the nose, and then use the gray value of the pixel of the image area other than the detection area Rnose of the bridge of the nose. Set to 0 to obtain the image Img5 corresponding to the position of the bridge of the nose;

(12), linear transformation is carried out to the image Img5 corresponding to the position of the bridge of the nose, to enhance the contrast, obtain the enhanced image Img6, and obtain the gray value corresponding to the pixel point in the enhanced image Img6; The image Img5 is linearly transformed to obtain the enhanced image Img6, and the method for obtaining the gray value corresponding to the pixel point in the enhanced image Img6 includes the following steps:

The pixels in the Img5 image are traversed by the linear transformation formula Pixel _n =1.5*Pixel _o +30, wherein Pixel _o is the gray value of the pixel in the image Img5 corresponding to the position of the bridge of the nose, and Pixel _n is the enhanced image Img6 The gray value corresponding to the middle pixel, 30 is the offset;

Obtain the grayscale value Pixeln corresponding to the pixel in the enhanced image _Img6 . If _Pixeln is greater than 255, then _Pixeln =255; if _{Pixeln is} less than 0, then _Pixeln =0; if Pixeln _is between 0 and 255, Then Pixel _n is equal to 1.5*Pixel _o +30.

得到均值滤波图像Img7，其中，src1(x,y)为图像Img6在(x,y)处像素点的灰度值，dst1(x,y)为图像Img7在(x,y)处像素点的灰度值，h(k,l)为滤波核，k，l表示滤波核的大小，k，l均为奇数，本实施例中，k，l取值均为15。(13), carry out mean filter processing to the enhanced image Img6, according to the formula

Obtain the mean filtered image Img7, where src1(x, y) is the gray value of the pixel point at (x, y) of the image Img6, and dst1(x, y) is the pixel point of the image Img7 at (x, y) The gray value of the point Gray value, h(k, l) is the filter kernel, k, l represent the size of the filter kernel, k, l are odd numbers, in this embodiment, k, l are both 15.

(14) Perform adaptive threshold processing on the mean value filtered image Img7 to obtain an adaptive binarized image Img8, wherein the adaptive threshold processing is a known technology, and it is unnecessary to describe it here.

(15), construct a rectangular structural element with a size of 25 pixels*25 pixels, perform a binary morphological preprocessing operation of first eroding and then expanding on the adaptive binarized image Img8 to obtain an image Img9, and combine the eroded rectangular structural element and the expanded image Img9. The rectangular structuring element processes the adaptive binarized image Img8, and the corroded rectangular structuring elements and the dilated rectangular structuring elements are both set as solid rectangle shapes, and the length and width of the solid rectangle shape are set to 25 pixels long and 25 pixels respectively. pixel length.

(16), carry out connected domain judgment on image Img9 by Two-Pass algorithm, obtain all connected domains in the image Img9, and find out the corresponding contour of the connected domain in the image Img9 according to the connected domain in the image Img9, obtain the initial nose bridge outline set Contour _c .

计算初始鼻梁轮廓集合Contour_c中符合要求的轮廓数量，获得鼻梁轮廓集合Contour_n；(17), according to the preset length of the bridge of the nose, through the formula

Calculate the number of contours that meet the requirements in the initial nose bridge contour set Contour _c , and obtain the nose bridge contour set Contour _n ;

Wherein, the preset nose bridge length L _ns is the standard nose bridge length that has been converted into pixel length, and CL _s is the contour length corresponding to the connected domain in the initial nose bridge contour set Contour _c ; Converts preset nose bridge lengths to pixel lengths.

(18), judge the contour quantity in the nose bridge contour set Contour _n , if the contour quantity in the nose bridge contour set Contour _n is 0 or greater than 1, then go to step (22); If the contour quantity in the nose bridge contour set Contour _m is 1, then go to step (19).

(19), obtain the minimum circumscribed rectangle of the nose bridge contour set Contour _n , obtain the length of the nose bridge to be detected and the center coordinates (Xns, Yns) of the nose bridge to be detected according to the length of the minimum circumscribed rectangle and the center coordinates of the rectangle in the image coordinate system .

(20), compare the length of the bridge of nose to be detected according to the length of the bridge of the nose to be detected obtained in step (18), judge whether the length of the bridge of the nose to be detected meets the requirements, when the length of the bridge of the nose to be detected deducts the standard length of the bridge of the nose When the absolute value is not greater than the preset nose bridge length tolerance, then go to step (21); When the absolute value of the length of the nose bridge to be detected minus the standard length of the nose bridge is greater than the preset nose bridge length tolerance, then go to step (22) ), wherein, the standard length of the bridge of the nose can be determined by the length of the bridge of the nose in the masks sold in the market or determined according to the commonly used length of the bridge of the nose in the art.

(21), construct a transformation matrix M _{c with coordinates (Xs, Ys) as the center and As as the rotation angle, and obtain the mask center coordinates in the image coordinate system according to the constructed transformation matrix M c in} step (8) (Xs, Ys) and the center coordinates (Xns, Yns) of the nose bridge to be detected in the image coordinate system obtained in step (19) are subjected to affine transformation to obtain the center coordinates (Xsc, Ysc) of the mask after affine transformation ) and the center coordinates (Xnsc, Ynsc) of the bridge of the nose after affine transformation, to obtain the absolute value Labs=|Ysc-Ysc|, if the absolute value Labs≤the preset nose bridge offset value, then the mask to be tested is judged to be a good product, so The mask to be tested is transported to the product qualified area; if _Labs > the preset nose bridge offset value, then go to step (22).

Among them, the preset nose bridge offset value is set to 5mm, and it is a known technology to construct a transformation matrix and perform _affine transformation.

仿射变换公式为

(X,Y)为仿射变换前的图像像素坐标，(X′,Y′)为仿射变换后的图像像素坐标，本实施例中当(X,Y)＝(Xs,Ys)，则(X′,Y′)＝(Xsc,Ysc)；当(X,Y)＝(Xns,Yns)，则(X′,Y′)＝(Xnsc,Ynsc)。(t _x , _ty ) represents the amount of translation, a ₁ , a ₂ , a ₃ , and a ₄ reflect changes such as image rotation, scaling, etc. In this embodiment, a ₁ =cos(As), a ₂ =-sin( As), a ₃ =sin(As), a ₄ =cos(As), t _x =Xs, _ty =Ys, that is, the affine transformation matrix M _c =

The affine transformation formula is

(X, Y) are the pixel coordinates of the image before affine transformation, and (X', Y') are the pixel coordinates of the image after affine transformation. In this embodiment, when (X, Y)=(Xs, Ys), then (X', Y')=(Xsc, Ysc); when (X, Y)=(Xns, Yns), then (X', Y')=(Xnsc, Ynsc).

(22) Determine that the mask to be tested is a defective product, and send the mask to be tested to the unqualified area of the product.

As shown in Figure 1, according to a kind of mask nose bridge detection method of the above-mentioned present invention, the invention provides a kind of mask nose bridge detection system, first through the binarization processing mode, the initial image of the mask is processed, and then find out the mask outline collection Contour The minimum circumscribed rectangle of _m , and then obtain the length Ls, width Ws of the mask to be detected, the center coordinates (Xs, Ys) in the image coordinate system, and the angle As parameter information in the image coordinate system, and then pass the outer area rectangle Ro and inner area. The rectangle Ri obtains the detection area Rnose of the bridge of the nose, and then in the detection area Rnose of the bridge of the nose, the image is filtered, and the adaptive binarization image is generated through adaptive threshold processing, and the adaptive binarization within the detection range is The connected domain of the image is confirmed, and the contour set of the nose bridge Contour _n is obtained, and then the outline size and center coordinates of the nose bridge are obtained by the minimum circumscribed rectangle method, and then the length of the nose bridge to be detected is compared with the standard length of the nose bridge, and the affine transformation is carried out. The center coordinates (Xsc, Ysc) of the mask are compared with the center coordinates (Xnsc, Ynsc) of the nose bridge after affine transformation, so as to effectively determine the length of the nose bridge and the quality of the nose bridge position of the mask to be detected.

The present invention is a mask nose bridge detection system, comprising:

The image acquisition unit is used to collect the initial image Img1 of the mask to be detected;

The image processing unit is used to obtain an initial binarized image Img2 through a fixed threshold algorithm, obtain a closed-operation binarized image Img3 through morphological closing processing, obtain an open-operation binarized image Img4 through morphological opening processing, and obtain an open-operation binarized image Img4 through morphological closing processing. Transform to obtain an enhanced image Img6, obtain a mean value filtered image Img7 through mean filtering, obtain an adaptive binarized image Img8 through adaptive threshold processing, and obtain an image Img9 through a binary morphological preprocessing operation;

The image labeling unit is used to judge the connected domain of the binarized image Img4 of the split operation by the Two-Pass algorithm to obtain the mask contour set Contour _m and to judge the connected domain of the image Img9 by the Two-Pass algorithm to obtain the initial nose bridge contour set Contour _c ;

The analysis and calculation unit is used to obtain the length Ls, width Ws of the mask to be detected, the center coordinates (Xs, Ys) of the mask to be detected in the image coordinate system, the angle As of the mask to be detected in the image coordinate system, the length of the bridge of the nose to be detected, The center coordinates of the bridge of the nose to be detected (Xns, Yns);

The image determination unit is used to determine whether the size of the mask to be detected meets the requirements, whether the length of the bridge of the nose to be detected meets the requirements, and the determination is based on the center coordinates (Xsc, Ysc) of the mask after affine transformation and the bridge of the nose after affine transformation. Whether the absolute value Labs=|Ysc-Ysc| obtained by the center coordinates (Xnsc, Ynsc) meets the requirements.

To sum up, a method and system for detecting the bridge of the nose of a mask of the present invention processes the initial image of the mask by means of binarization, and then finds the minimum circumscribed rectangle of the mask contour set Contour _m , and then obtains the length Ls of the mask to be detected. , the width Ws, the center coordinates (Xs, Ys) in the image coordinate system, the angle As parameter information in the image coordinate system, and then obtain the detection area Rnose of the nose bridge through the outer area rectangle Ro and the inner area rectangle Ri, and then determine the nose bridge. In the detection area _Rnose , the adaptive binarized image is generated by filtering the image, and the adaptive binarized image is generated by adaptive thresholding. The external dimensions and center coordinates of the bridge of the nose are obtained by the minimum circumscribed rectangle method, and then the length of the bridge to be detected is compared with the standard length of the bridge of the nose, and the center coordinates (Xsc, Ysc) of the mask after affine transformation and the affine transformation are compared. The center coordinates (Xnsc, Ynsc) of the posterior bridge of the nose are compared, so as to effectively determine the length of the bridge of the nose and the quality of the location of the bridge of the nose of the mask to be detected.

The above-mentioned embodiments only represent several embodiments of the present invention, and the descriptions thereof are specific and detailed, but should not be construed as limiting the scope of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can also be made, which all belong to the protection scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims (9)

1. A method for detecting the bridge of the nose of a mask is characterized by comprising the following steps:

(1) acquiring an initial image Img1 of the mask to be detected;

(2) establishing an image coordinate system, and carrying out binarization processing on the initial image Img1 according to a fixed threshold algorithm to obtain an initial binarized image Img 2;

(3) constructing a circular structural element, and performing morphological closing processing on the initial binary image Img2 to obtain a closed-operation binary image Img 3;

(4) constructing a rectangular structural element, and performing morphological opening processing on the binary image Img3 subjected to the closing operation to obtain an opening operation binary image Img 4;

(5) judging the connected domains of the binary image Img4 subjected to the open operation by a Two-Pass algorithm to obtain all the connected domains in the binary image Img4 subjected to the open operation, finding out the corresponding outline of the connected domains in the binary image Img4 according to the connected domains in the binary image Img4 subjected to the open operation, and obtaining an initial mask outline set Contour_s；

(6) According to the preset perimeter L of the mask_msBy the formula

Calculating initial mask Contour set Contour_sThe number of contours meeting the requirements in the mask Contour set is obtained_m(ii) a Wherein, the preset perimeter L of the mask_msCL, the standard mask perimeter that has been converted to pixel length_mSet Contour for initial mask Contour_sThe contour length corresponding to the middle connected domain;

(7) and determining mask Contour set Contour_mNumber of contours in, if mask wheelContour set Contour_mIf the number of the contours in the sequence is 0 or more than 1, the step (22) is carried out; if mask Contour set Contour_mIf the number of the contours in the step (5) is 1, then the step (8) is carried out;

(8) acquiring mask Contour set Contour_mAccording to the length and the width of the minimum circumscribed rectangle, the length Ls and the width Ws of the mask to be detected, the center coordinates (Xs, Ys) in the image coordinate system and the angle As parameter information in the image coordinate system are obtained according to the length and the width of the minimum circumscribed rectangle, the center coordinates (Xs, Ys) in the image coordinate system and the angle As parameter information in the image coordinate system, whether the size of the mask to be detected meets the requirements or not is judged according to the preset size of the mask, and if the length and the width of the mask to be detected are within the preset size range of the mask, the step (8) is carried out; if the length and the width of the mask to be detected exceed the preset mask size range, turning to the step (22); wherein the rectangle angle in the image coordinate system is an angle formed by the long side of the minimum circumscribed rectangle and the X axis in the image coordinate system;

(9) respectively subtracting a preset length Lr1 from the four vertexes of the mask to be detected along the length direction of the mask to be detected and subtracting a preset width Wr1 from the four vertexes of the mask to be detected along the width direction of the mask to be detected by taking the four vertexes of the mask to be detected As datum points under the condition that the central coordinates (Xs, Ys) and the angle As of the mask to be detected obtained in the step (8) are not changed, and generating an outer region rectangle Ro for nose bridge detection;

(10) respectively subtracting a preset length Lr2 along the length direction of the mask to be detected and subtracting a preset width Wr2 along the width direction of the mask to be detected by taking four vertexes of the mask to be detected As datum points under the condition that the central coordinates (Xs, Ys) and the angle As of the mask to be detected acquired in the step (8) are not changed, and generating an inner area rectangle Ri for nose bridge detection;

(11) carrying out region subtraction on the outer region rectangle Ro detected by the nose bridge and the inner region rectangle Ri detected by the nose bridge to obtain a detection region Rnosie of the nose bridge, and then setting the gray value of pixel points of an image region except the detection region Rnosie of the nose bridge as 0 to obtain an image Img5 corresponding to the position of the nose bridge;

(12) performing linear transformation on the image Img5 corresponding to the position of the nose bridge to obtain an enhanced image Img6, and obtaining a gray value corresponding to a pixel point in the enhanced image Img 6;

(13) and carrying out mean value filtering processing on the enhanced image Img6 according to a formula

Obtaining a mean value filtering image Img7, wherein src1(x, y) is the gray value of a pixel point of the image Img6 at (x, y), dst1(x, y) is the gray value of a pixel point of the image Img7 at (x, y), h (k, l) is a filtering kernel, k and l represent the size of the filtering kernel, and k and l are both odd numbers;

(14) carrying out self-adaptive threshold processing on the mean value filtering image Img7 to obtain a self-adaptive binary image Img 8;

(15) constructing a rectangular structural element, and carrying out binary morphological preprocessing operation of firstly corroding and then expanding on the self-adaptive binary image Img8 to obtain an image Img 9;

(16) judging the connected domains of the image Img9 through a Two-Pass algorithm, acquiring all the connected domains in the image Img9, finding out the corresponding contours of the connected domains in the image Img9 according to the connected domains in the image Img9, and acquiring an initial nose bridge Contour set Contour_c；

(17) According to the preset length of the nose bridge, by formula

Calculating initial nose bridge Contour set Contour_cObtaining the nose bridge Contour set Contour according to the required Contour number_n；

Wherein the preset nose bridge length L_nsFor a standard nose bridge length, CL, which has been converted into a pixel length_sSet Contour for initial nose bridge Contour_cThe contour length corresponding to the middle connected domain;

(18) judging the nose bridge Contour set Contour_nIf nose bridge Contour set Contour_nIf the number of the contours in the sequence is 0 or more than 1, the step (22) is carried out; if nose bridge Contour set Contour_mIf the number of the contours in the step (2) is 1, then the step (19) is carried out;

(19) obtaining a nose bridge Contour set Contour_nThe length of the nose bridge to be detected and the center coordinates of the nose bridge to be detected are obtained according to the length of the minimum circumscribed rectangle and the center coordinates of the rectangle in the image coordinate system (Xns, Yns);

(20) comparing the preset nose bridge length with the length of the nose bridge to be detected obtained in the step (18), judging whether the length of the nose bridge to be detected meets the requirement, and turning to the step (21) when the absolute value of the length of the nose bridge to be detected minus the standard length of the nose bridge is not larger than the preset nose bridge length tolerance; when the absolute value of the length of the nose bridge to be detected minus the standard length of the nose bridge is larger than the preset tolerance of the length of the nose bridge, turning to the step (22);

(21) constructing a transformation matrix M_cAccording to the constructed transformation matrix M_cAffine transformation is carried out on the center coordinates (Xs, Ys) of the mask and the center coordinates (Xns, Yns) of the nose bridge to be detected, so that center coordinates (Xsc, Ysc) of the mask after affine transformation and center coordinates (Xnsc, Ynsc) of the nose bridge after affine transformation are obtained, an absolute value Labs is | Ysc-Ysc | is obtained, if the absolute value Labs is smaller than or equal to a preset nose bridge deviation value, the mask to be detected is judged to be good, and the mask to be detected is conveyed to a product qualified area; if L is_absIf the nose bridge deviation value is larger than the preset nose bridge deviation value, the step (22) is carried out;

(22) and judging that the mask to be detected is a defective product, and conveying the mask to be detected to a product failing area.

2. The method of detecting bridge of the nose of a mask according to claim 1, wherein the fixed threshold algorithm is formulated as:

wherein src (x, y) is a gray value of a pixel point of the initial binary image Img1 at (x, y), dst (x, y) is a gray value of a pixel point of the initial binary image Img2 at (x, y), thresh and maxval are preset mask extraction gray values, and when the initial image Img1 is binarized according to a fixed threshold algorithm, when src (x, y) > thresh, dst (x, y) in the initial binary image Img2 is 0, and when src (x, y) ≦ thresh, dst (x, y) in the initial binary image Img2 is maxval.

3. The method of detecting bridge of the nose of a mask of claim 1, wherein: the predetermined gauze mask size range includes the length of predetermined gauze mask and the width of predetermined gauze mask, and the length of predetermined gauze mask is the length 2mm of standard gauze mask, and the width of predetermined gauze mask is the width 2mm of standard gauze mask.

4. The method for detecting the bridge of the nose of a mask according to claim 1, wherein in the step (12), the image Img5 corresponding to the position of the bridge of the nose is linearly transformed to obtain the enhanced image Img6, and the method for obtaining the gray value corresponding to the pixel point in the enhanced image Img6 includes the following steps:

by linear transformation of the formula Pixel_n＝1.5*Pixel_o+30 traversal of Pixel points in Img5 image, where Pixel_oThe gray value, Pixel, of the Pixel point in the image Img5 corresponding to the position of the nose bridge_n30 is an offset value for the gray value corresponding to the pixel point in the enhanced image Img 6;

obtaining gray value Pixel corresponding to Pixel point in the enhanced image Img6_nIf Pixel_nIf greater than 255, Pixel_n255; if Pixel_nLess than 0, Pixel_n0; if Pixel_nBetween 0 and 255, then Pixel_nEqual to 1.5 pixels_o+30。

5. The method of detecting bridge of the nose of a mask of claim 1, wherein: the radius of the circular structural element in the step (3) is 9 pixels in length.

6. The method of detecting bridge of the nose of a mask of claim 1, wherein: the size of the rectangular structural element in the step (4) is 25 pixels by 25 pixels.

7. The method of detecting bridge of the nose of a mask of claim 1, wherein: the size of the rectangular structural element in the step (15) is 25 pixels by 25 pixels.

8. The method of detecting bridge of the nose of a mask of claim 1, wherein: the preset nose bridge deviation value is set to be 5 mm.

9. A mask nose bridge detection system which characterized in that: comprises that

The image acquisition unit is used for acquiring an initial image Img1 of the mask to be detected;

an image processing unit for obtaining an initial binarized image Img2 through a fixed threshold algorithm, a binarized image Img3 for closed operation through morphological closed processing, a binarized image Img4 for open operation through morphological open processing, an enhanced image Img6 through linear transformation, a mean filtered image Img7 through mean filtering processing, an adaptive binarized image Img8 through adaptive threshold processing, and an image Img9 through binary morphological pre-processing operation;

an image marking unit for judging the connected domain of the binary image Img4 subjected to the binary operation by the Two-Pass algorithm to obtain the mask Contour set Contour_mAnd judging a connected domain of the image Img9 through a Two-Pass algorithm to obtain an initial nose bridge Contour set Contour_c；

The analysis and calculation unit is used for obtaining the length Ls and the width Ws of the mask to be detected, the central coordinates (Xs, Ys) of the mask to be detected in an image coordinate system, the angle As of the mask to be detected in the image coordinate system, the length of the nose bridge to be detected and the central coordinates (Xns, Yns) of the nose bridge to be detected;

and the image judging unit is used for judging whether the size of the mask to be detected meets the requirement or not, whether the length of the nose bridge to be detected meets the requirement or not, and whether an absolute value Labs ═ Ysc-Ysc | obtained according to the center coordinates (Xsc, Ysc) of the mask subjected to affine transformation and the center coordinates (Xnsc, Ynsc) of the nose bridge subjected to affine transformation meets the requirement or not.

Images