CN105424722A - Bottle cap unqualified product marking method based on machine vision - Google Patents

Abstract

The invention relates to a method for marking unqualified bottle caps based on machine vision. The method uses an LED white ring light source to illuminate the surface of the bottle cap to be tested, and the steps are as follows: 1) collecting image information of qualified bottle caps; 2) collecting the bottle cap to be tested Cap image; 3) Binarize the image after median filter preprocessing; 4) Carry out circle positioning, find the edge of the bottle cap as the round edge, and establish a coordinate system with the center of the circle as the origin; 5) According to the qualified bottle cap The number of pixels of the production date inkjet presets the upper and lower limits of the number of pixels of the detection target. Using the circle edge obtained in step 4) and the established coordinate system, set the inside of the circle as the region of interest ROI, and set the production date in the ROI Coding is used as the detection target to detect bottle caps; 6) Mark unqualified products. The algorithm used in the invention is simple, has few steps, realizes rapid detection of images, has high detection efficiency, is less affected by the environment, and has high accuracy.

Description

A method for marking unqualified bottle caps based on machine vision

Technical field

The invention relates to a method for marking unqualified products of bottle cap detection, which belongs to the field of machine vision.

Background technique

In industrial production lines such as beverages and beer, most of them use manual visual inspection to check whether the bottles are qualified. Especially in the production of bottle caps, it is necessary to print the production date on the bottle caps, and there may be situations where the spray code is blurred or missed. Because human eyes are prone to visual and physical fatigue, the efficiency is often very low, which will cause missed or false detections, and the labor cost is also getting higher and higher. The flow of these substandard products into the market will cause consumers to distrust the quality of the products, affect sales, and even cause problems such as drinking expired products by mistake and endangering health.

Aiming at the above-mentioned defects of manual detection, the research on bottle cap detection based on machine vision and the establishment of a detection system based on the platform of optical imaging, image acquisition, digital image processing and analysis not only have theoretical basis, but also have great economic value , can ensure to reduce and eliminate the occurrence of the above phenomena to the greatest extent, and improve product quality.

Machine vision detection has been applied in many detection fields, and there are also application precedents in bottle cap detection, but most of them are for the detection of defects on the inner pad and outer surface of the bottle cap (such as patent CN102095733A, CN103884651A), these are all for the appearance of the bottle cap However, there is little research on the detection of the coding information (such as the date of production) on the bottle cap. In the industrial production line, at least 5 bottles must be detected per second, and the existence, integrity and clarity of the production date coding on the bottle cap need to be detected. If it is unqualified, it must be marked or rejected. This has high requirements for the real-time performance of the detection system and software processing.

Contents of the invention

The purpose of the present invention is to overcome the shortcomings of manual detection and the limitations of bottle cap detection, and provide an online labeling method for unqualified bottle caps based on machine vision. Technical scheme of the present invention is as follows:

A method for marking unqualified bottle caps based on machine vision. The method uses an LED white ring light source to illuminate the surface of the bottle cap to be tested. The steps are as follows:

1) Place the qualified bottle at the station to be tested, collect the image information of the bottle cap, adjust the camera and light source, so that the contrast between the background image of the collected bottle cap and the production date sprayed on it reaches the maximum for subsequent processing;

2) Transfer the bottle to be detected to the station to be tested, collect the cap image, and perform median filter preprocessing on the collected cap image to remove noise interference and enhance the detection target information;

3) Adjust the appropriate threshold according to the actual environment, and perform binarization processing on the image preprocessed by the median filter, so that the binarized image can effectively separate the background and the target;

4) 3) the image that obtains is carried out circular positioning, finds the bottle cap edge as the circular edge, takes the center of circle as the origin to establish a coordinate system, adopts the following steps to carry out circular positioning:

1. Set an annular area so that the edge of the bottle cap is between the inner ring and the outer ring.

2. The search area is set within the ring area, and the search direction is from the inner ring to the outer ring;

3. With the center of the inner ring as the origin, set multiple search lines at fixed intervals between the inner ring and the outer ring;

4. When the polarity of a certain point on the search line changes from gray value 1 to gray value 0, it is considered to be the boundary point to be found;

5. Use the boundary points found on each search line to form a circular edge;

5) Preset the upper and lower limits of the pixel points of the detection target according to the pixel points of the production date on the qualified bottle cap, and use the circle edge obtained in step 4) and the established coordinate system to set the inside of the circle as the area of interest ROI, the production date in the ROI is used as the detection target. According to the actual situation, set the appropriate pixel size of the detection target to filter the influence of irrelevant small pixel black spots caused by light intensity in the image. If the number of pixels of the detection target is within If it is within the specified upper and lower limits, it is qualified, otherwise it is judged as unqualified product;

6) For unqualified products, output the control signal through the serial I/O, open the solenoid valve, and use the nozzle of the marking unit to spray out the marked unqualified bottle with color.

The beneficial effects of the present invention are as follows: firstly, the algorithm used in the method is simple, the steps are few, the rapid detection of the image is realized, and the detection efficiency is high; secondly, the method is less affected by the environment and has a high accuracy rate; Good; the detection method is novel and has a wide range of applications, and is applicable to the detection of industrial glass bottled products; this method can be added to the existing production line to realize on-line detection during production. The invention is suitable for marking unqualified products with the date of production on the surface of bottle caps in automatic production.

Description of drawings

Fig. 1 is used in the schematic diagram of the device for marking unqualified bottle caps based on machine vision of the present invention.

Fig. 2 is a flow chart of the method for detecting the production date on the surface of the bottle cap based on machine vision of the present invention.

Fig. 3 (a) the image of the original bottle cap to be tested by the method of the present invention, (b) the image of the bottle cap collected in step 1.

Fig. 4 (a) the judgment result diagram of qualified bottle cap detected by the method of the present invention, (b) the judgment result diagram of unqualified bottle cap.

detailed description

The present invention will be further described below in conjunction with the accompanying drawings and specific examples.

See Figure 1, a system device for marking unqualified bottle caps based on machine vision, which is divided into six functional modules, namely conveyor belt 1, image acquisition module 2, photoelectric trigger 3, lighting module 4, image processing and recognition 5 and output control device 6. The above-mentioned image acquisition module 2 uses a 1.2 million-pixel CCD camera, the lighting module 4 uses an LED ring light source, and the image processing and recognition module 5 uses an industrial computer. The output control device consists of a nozzle controlled by a solenoid valve. The conveyor belt 1 transports the bottle to be tested to the photographing station, and the photoelectric trigger device 3 is used to detect whether the bottle arrives at the photographing position, so as to generate a pulse signal and deliver it to the industrial computer 5, and the industrial computer sends a signal to trigger the camera 2 to take pictures; the ring light source 4 is The camera provides lighting to improve the image quality when taking pictures. The CCD camera is located directly above the conveyor belt 1 and takes pictures through the middle of the ring light source. The captured images are sent to the industrial computer for processing; the output control device is connected to the industrial computer. When unqualified When the product is produced, the output control signal controls the nozzle 6 to spray paint to mark it.

Referring to Fig. 2, the flow chart of the present invention based on machine vision to detect defective product marks on the bottle cap surface production date. Proceed as follows:

(1) Place the industrial camera at the station to be tested directly above the workpiece conveyor belt of the production line, place qualified bottle cap products on the position to be tested, adjust the exposure compensation of the camera itself and the brightness of the ring light source to obtain the background image of the bottle cap and spray the production date Contrast is maximized.

(2) Start the conveyor belt, the bottle to be tested arrives at the shooting position along with the conveyor belt, trigger the camera to acquire the image of the bottle cap to be tested, and transmit it to the image processing system of the industrial computer through the channel.

(3) Firstly, the image is preprocessed. After many tests, the median filtering effect is the best, and the image target information is enhanced after processing.

(4) Binarize the processed image in step (4), and set an appropriate threshold according to the gray histogram of the image to improve the image recognition rate.

(5) Find the edge of the bottle cap in the binarized image as the round edge, and establish a coordinate system with the center of the circle as the origin. The method is as follows: In the first step, based on the bottle cap, set a circular area so that the edge of the bottle cap is between the inner ring and the outer ring. In the second step, the circular area is used as the search circle area, and the search direction is from the inner ring to the outer ring; in the third step, multiple rays are set with the center of the inner ring as the starting point, and the adjacent interval between the inner ring and the outer ring is intercepted as 10px line segments for search lines. In the fourth step, when the polarity of a certain point on the line segment changes from gray value 1 to gray value 0, it is considered as the boundary point to be found. The fifth step is to set the minimum border strength to 20, the width of the search line to 3px, and the circle edge pixels to 3px. According to the above conditions, a circle edge composed of points on the search line as boundary points can be obtained.

(6) Find the inside of the round edge as the processing area, and print the production date as the detection target. According to the actual situation, set the appropriate pixel size of the detection target. When the number of detection targets is within the set threshold, it is judged as qualified, otherwise it is unqualified.

(7) The unqualified signal is output by the industrial computer to control the opening of the solenoid valve, and the nozzle connected to the solenoid valve sprays the paint to mark the unqualified bottle. After the marking is completed, continue to detect the next bottle cap to be tested. The specific operation steps are as follows:

The beverage bottle to be tested passes through the photoelectric trigger, CCD camera and marking unit once on the conveyor belt. It is known that the distance between the photoelectric trigger and the camera is M, the distance between the camera and the nozzle is N, and the speed of the conveyor belt is v. When the bottle passes the trigger, the sensor sends a signal to the industrial computer to record the current time as t, which is used as a reference time. When the camera passes the M/V time, the camera starts to take pictures and collect images. After the industrial computer detects and judges, the qualified product does not send out a signal. The unqualified product will output the control signal through N/V, open the solenoid valve, and make the nozzle spray out the red pigment to mark the bottle.

As shown in Figure 3, the image on the left is the original image captured by the camera, and the image on the right is the image collected after adjusting the exposure rate of the camera and the light source. This adjusted image makes subsequent processing and judgment easier.

See Figure 4, (a) is the judgment result of qualified bottle caps. It can be seen from the figure that the edge of the bottle is the found round edge. When detecting the target, only the inkjet code inside the circle is detected. (b) is the test result when the production date on the bottle cap is incomplete. If the production date is missing or unclear, it will be judged as unqualified.

Claims (1)

1. A method for marking unqualified bottle caps based on machine vision, the method uses LED white ring light source to illuminate the surface of the bottle cap to be tested, and the steps are as follows: 1) Place the qualified bottle at the station to be tested, collect the image information of the bottle cap, adjust the camera and light source, so that the contrast between the background image of the collected bottle cap and the production date sprayed on it reaches the maximum for subsequent processing; 2) Transfer the bottle to be detected to the station to be tested, collect its cap image, and perform median filter preprocessing on the collected cap image to remove noise interference and enhance the detection target information; 3) Adjust the appropriate threshold according to the actual environment, and perform binarization processing on the image preprocessed by the median filter, so that the binarized image can effectively separate the background and the target; 4) 3) the image that obtains is carried out circular positioning, finds the bottle cap edge as the circular edge, takes the center of circle as the origin to establish a coordinate system, adopts the following steps to carry out circular positioning: 1. Set an annular area so that the edge of the bottle cap is between the inner ring and the outer ring. 2. The search area is set within the ring area, and the search direction is from the inner ring to the outer ring; 3. With the center of the inner ring as the origin, set multiple search lines at fixed intervals between the inner ring and the outer ring; 4. When the polarity of a certain point on the search line changes from gray value 1 to gray value 0, it is considered to be the boundary point to be found; 5. Use the boundary points found on each search line to form a circular edge. 5) Preset the upper and lower limits of the pixel points of the detection target according to the pixel points of the production date on the qualified bottle cap, and use the circle edge obtained in step 4) and the established coordinate system to set the inside of the circle as the area of interest ROI, the production date in the ROI is used as the detection target. According to the actual situation, set the appropriate pixel size of the detection target to filter the influence of irrelevant small pixel black spots caused by light intensity in the image. If the number of pixels of the detection target is within If it is within the specified upper and lower limits, it is qualified, otherwise it is judged as unqualified product; 6) For unqualified products, output the control signal through the serial I/O, open the solenoid valve, and use the nozzle of the marking unit to spray out the marked unqualified bottle with color.