JP2023104829A - Rod-shaped cigarette surface defect detection method and device - Google Patents

Abstract

To provide rod-shaped cigarette surface defect detection method and device.SOLUTION: The invention discloses a rod-shaped cigarette surface defect detection method which comprises the following steps of: S1, acquiring images of the front surface and two end surfaces of a rod-shaped cigarette to obtain initial images of the front surface and the two end surfaces of the rod-shaped cigarette; S2, performing preprocessing of format conversion operation, median filtering operation, sobel edge detection operation, OTSU adaptive binarization operation and Hough line detection operation on each collected initial image to obtain preprocessed images of the front surface and the two end surfaces of the rod-shaped cigarette; and S3, performing defect detection on the surface of the rod-shaped cigarette according to the preprocessed images of the front surface and the two end surfaces of the rod-shaped cigarette, wherein the defect detection comprises opening tape missing detection, opening tape offset detection and end surface polyline detection. According to the invention, on the basis of image acquisition and analysis on the surface of the rod-shaped cigarette, defect detection on the surface of the rod-shaped cigarette is achieved, and application in automatic detection is facilitated.SELECTED DRAWING: Figure 1

Description

The present invention relates to tobacco rod detection, and more particularly to a method and apparatus for detecting surface imperfections in tobacco rods.

At present, the level of cigarette manufacturing in our country is basically maturing in cigarette wrapping process technology, accompanied by constant innovation in production technology. It still employs the most traditional manual detection means. The speed of the current cigarette outer packaging machine is very fast, basically reaching the level of 700-800 boxes/minute, and the newly developed cigarette packaging machine can reach as high as 900 boxes/minute. However, in such a fast-paced, high-strength packaging and manufacturing process, due to the nature of human beings, we cannot concentrate for a long period of time. could be connected. Accuracy of detection results in some extreme environments and standardized storage of detection data have also been issues for tobacco companies.

Some domestic tobacco processing and manufacturing companies often adopt a combination of various sensors and manual detection methods to control the inspection failure rate to generally acceptable levels, but defects such as tear tape misalignment is difficult to detect.

The purpose of the present invention is to overcome the deficiencies of the prior art, and to realize the detection of defects on the surface of tobacco rods based on the capture analysis of the surface of tobacco rods, which is also convenient to operate in automated detection. To provide a method and apparatus for tobacco surface defect detection.

The object of the present invention is achieved through the following technical means.

A method for detecting defects on the surface of a tobacco rod has the following configuration. i.e.
S1: Capturing an image of the front face and two end faces of the tobacco stick to obtain an initial image of the front face and two end faces of the tobacco stick;
A method of capturing an image has the following components. i.e.
First, the camera is configured for direct grayscale output, the configuration format is YUV422, the camera captures images of the front and two end faces of the cigarette stick, and then the images in YUV422 format are converted to YUV444 format, Obtain initial images of the front and two end faces of the tobacco rod;
Second, configure the output of the camera as an RGB565 output, capture images of the front and two end faces of the cigarette stick with the camera, obtain three images, and then convert each of the three images from RGB565 to YUV422 format. Convert and then convert the YUV422 format image to YUV444 format to obtain initial images of the front and two end faces of the cigarette stick.

Preferably, the two end faces of the tobacco stick refer to the two faces with the smallest areas of the tobacco stick, and the respective end faces and the front face of the tobacco stick are imaged by a camera.

S2: Each captured initial image is preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a hough line detection operation, and the front and two end faces of the cigarette stick are obtained. obtaining a preprocessed image;
Said step S2 includes the following sub-steps S201-S206. i.e.
Step S201 for converting any of the initial images obtained in step S1 from YUV444 format image data into a dedicated format for the AXI4 protocol Stream video stream;
Step S202 of performing a median filter operation on the converted AXI4-Stream format image data to remove salt and pepper noise in the image to reduce unnecessary interference;
A Sobel edge detection operation is performed on the image data after the median filter. With the Sobel operator as the core, an edge detection operation is performed on the entire image of the image data after the median filter. step S203, discarding interference information such as the background of the
OTSU adaptive binarization is performed on the image data after edge detection, and the entire image is divided into monochrome images (i.e., the grayscale values are only 0 and 255), so that the boundary features can be identified. Step S204 for emphasizing and weakening useless information such as the background to facilitate subsequent judgment;
A Hough line detection operation is performed on the binarized image data to divide straight lines in the entire image. The Hough line detection operation outputs the detected image data and returns a parameter MAX_LINE_NUM at the same time. It is the total number of detected oblique lines, and since the lines in the image after the previous edge detection are not continuous, the Hough line detection further improves the detection accuracy and facilitates the subsequent judgment step S205;
The initial images of the front face and the two end faces of the cigarette stick are preprocessed in steps S201 to S205, respectively, to obtain preprocessed images of the front face and the two end faces of the cigarette stick in step S206.

S3: A step of performing surface defect detection of the cigarette stick, including tear tape missing detection, tear tape deviation detection and end face polyline detection, according to the pre-processed images of the front face and two end faces of the cigarette stick.

The missing tape detection described in step S3 has the following configuration. i.e.
If the cigarette stick has a length of 280 mm and a height of 85 mm, and the surface image of the cigarette stick is captured by a camera, the photographing accuracy is 0.12 mm/pixel (here, Pixel represents a pixel). each row sums 280 mm/(0.12 mm/Pixel) = 2332 Pixels and each column sums 85 mm/(0.12 mm/Pixel) = 708 Pixels;
Scanning to a preprocessed image of the front face of a cigarette stick:
By sequentially scanning the grayscale value of each pixel in the image from left to right and top to bottom, a counter is constructed that is initialized to 0, and during the scanning process the grayscale values of 11 consecutive pixels in a row are obtained. 255, the counter is incremented by 1. After scanning the entire image, the count value of the counter is the sum of the number of lines in the entire image that satisfy the judgment condition, and then make another judgment, namely:
If the value of the counter exceeds 2/3 of the height pixel value of the cigarette stick (i.e., exceeds 708 Pixels x 2/3 = 472 Pixels), it is determined that the tear tape on the cigarette stick is present; It is determined that a missing tape defect has appeared.

The deviation detection of the opening tape described in step S3 has the following configuration. i.e.
For the preprocessed image of the front face of the cigarette bar, if the tear tape is present after the determination of the tear tape missing, step A1 of constructing a two-dimensional array:
step A11 of summing the number of pixels with a grayscale value of 255 in each column for the preprocessed image of the front face of the cigarette stick;
Step A12 of constructing a two-dimensional coordinate system: the origin is at the lower right corner of the preprocessed image of the front surface of the cigarette stick, the y-axis and the right boundary line of the preprocessed image of the front surface of the tobacco stick are on the same straight line, The direction of the y-axis is from bottom to top;
In the direction from right to left, for pixels in the i-th column of the preprocessed image of the front face of the cigarette stick, the serial number i of this column is the x-axis coordinate xi, and the total number of pixels whose grayscale value is 255 in this column is the two-dimensional coordinate (xi, yi) of the i-th pixel, where i=1, 2, . represents the number of columns of pixels, N=2332;
step A13 of repeating step A12 for each column of pixels to obtain all the two-dimensional coordinates to form a two-dimensional array (x1, y1), (x2, y2), ..., (xN, yN);
By traversing the two-dimensional array, excluding the left and right borders x1, xN, find the two x-axis coordinates with the largest total number of pixels stored in the array and record them as xk1, xk2, which is the unpacking Step A14, which is the position of the two boundary lines of the tape, and further extracts the intermediate position between xk1 and xk2 to obtain the x-axis coordinate x0 of the center point of the unsealed tape;
The distance between the center point of the tear tape and the right border of the preprocessed image of the front face of the cigarette stick is actually the distance between the center point of the tear tape and the y-axis, i.e. the x of the center point of the tear tape. Corresponding to the coordinate x0 of the coordinate axes, the distance between the center point of the tear tape and the right border of the preprocessed image of the front face of the cigarette stick was recorded as D=x0, and the distance D was compared with the standard distance STD. , to obtain the amount of deviation T = D - STD;
If the absolute value of the deviation amount T exceeds a preset threshold value, step A2, it is assumed that there is a deviation defect of the opening tape.

Preferably, if a tear tape deviation defect is present, the direction of deviation is determined according to the positive or negative of T, i.e., if T>0, the tear tape is shifted to the left, and if T<0, the tear tape is shifted to the right. shift to

The edge face polyline detection described in step S3 has the following configuration. i.e.
Obtaining a preprocessed image of each endface in step S2 and performing endface polyline detection according to the parameter MAX_LINE_NUM returned in step S205: if the MAX_LINE_NUM value is equal to 4, then output 1, the current endface is normal. Otherwise, output 0 to indicate that the current endface is defective.

An apparatus for detecting defects on the surface of a tobacco rod,
a tobacco rod image capture device for capturing an image of the front and two end faces of the tobacco rod in the detection area to obtain an initial image of the front and two end faces of the tobacco rod and transmitting to the control module;
Each initial captured image was preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a hough line detection operation to preprocess the front and two end faces of the cigarette stick. An image is obtained, and defect detection is performed on the surface of the tobacco rod based on the preprocessed images of the front surface and two end surfaces of the tobacco rod, and a defect detection result, which is the presence or absence of a defect and the defect type if there is a defect, is output. A control module for generating rejection commands, if any.

The defect detection device further includes a cigarette stick conveying device, a detection light source, a defective product exclusion device, and a display module,
The tobacco rod conveying device is used to convey the tobacco rod to be detected to the detection area, and after the detection is completed, carry out the tobacco rod to be detected from the detection area,
The detection light source is used to illuminate the detection area and ensure the brightness of the detection area,
The display module is used to receive and display the defect determination result output from the control module,
Said rejection device is used to reject the defective cigarette stick according to the rejection command generated by the control module when the control module detects that the cigarette stick is defective.

The cigarette stick image capturing device comprises three cameras, one camera is used to capture the front image of the cigarette stick, and the other two cameras are used to capture the end image of the cigarette stick. Used.

Based on the image capture analysis of the surface of the tobacco rod, the present invention can realize the detection of defects on the surface of the tobacco rod, which is convenient for the application of automated detection and provides the conditions for rejection during the manufacturing and packaging process. .

3 is a flow chart of the method of the invention; 1 is a schematic diagram illustrating the principle of the device of the invention; FIG.

Hereinafter, the technical means of the present invention will be described in more detail with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following.

As shown in FIG. 1, the method for detecting defects on the surface of a tobacco stick has the following configuration. i.e.
S1: A step of capturing images of the front surface and two end surfaces of the tobacco rod to obtain initial images of the front surface and the two end surfaces of the tobacco rod. The surface is captured by a camera for each end surface and the front surface of the tobacco rod;
A method of capturing an image has the following components. i.e.
First, the camera is configured for direct grayscale output, the configuration format is YUV422, the camera captures images of the front and two end faces of the cigarette stick, and then the images in YUV422 format are converted to YUV444 format, Obtain initial images of the front and two end faces of the tobacco rod;
Second, configure the output of the camera as an RGB565 output, capture images of the front and two end faces of the cigarette stick with the camera, obtain three images, and then convert each of the three images from RGB565 to YUV422 format. Convert and then convert the YUV422 format image to YUV444 format to obtain initial images of the front and two end faces of the cigarette stick.

S2: Each captured initial image is preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a hough line detection operation, and the front and two end faces of the cigarette stick are obtained. Obtaining a preprocessed image Said step S2 comprises the following sub-steps S201-S206. i.e.
Step S201 for converting any of the initial images obtained in step S1 from YUV444 format image data into a dedicated format for the AXI4 protocol Stream video stream;
Step S202 of performing a median filter operation on the converted AXI4-Stream format image data to remove salt and pepper noise in the image to reduce unnecessary interference;
A Sobel edge detection operation is performed on the image data after the median filter. With the Sobel operator as the core, an edge detection operation is performed on the entire image of the image data after the median filter. step S203, discarding interference information such as the background of the
OTSU adaptive binarization is performed on the image data after edge detection, and the entire image is divided into monochrome images (i.e., the grayscale values are only 0 and 255), so that the boundary features can be identified. Step S204 for emphasizing and weakening useless information such as the background to facilitate subsequent judgment;
A Hough line detection operation is performed on the binarized image data to divide straight lines in the entire image. The Hough line detection operation outputs the detected image data and returns a parameter MAX_LINE_NUM at the same time. It is the total number of detected oblique lines, and since the lines in the image after the previous edge detection are not continuous, the Hough line detection further improves the detection accuracy and facilitates the subsequent judgment step S205;
The initial images of the front face and the two end faces of the cigarette stick are preprocessed in steps S201 to S205, respectively, to obtain preprocessed images of the front face and the two end faces of the cigarette stick in step S206.

S3: A step of performing surface defect detection of the cigarette stick, including tear tape missing detection, tear tape deviation detection and end face polyline detection, according to the pre-processed images of the front face and two end faces of the cigarette stick.

The missing tape detection described in step S3 has the following configuration. i.e.
If the cigarette stick has a length of 280 mm and a height of 85 mm, and the surface image of the cigarette stick is captured by a camera, the photographing accuracy is 0.12 mm/pixel (here, Pixel represents a pixel). each row sums 280 mm/(0.12 mm/Pixel) = 2332 Pixels and each column sums 85 mm/(0.12 mm/Pixel) = 708 Pixels;
Scanning to a preprocessed image of the front face of a cigarette stick:
By sequentially scanning the grayscale value of each pixel in the image from left to right and top to bottom, a counter is constructed that is initialized to 0, and during the scanning process the grayscale values of 11 consecutive pixels in a row are obtained. 255, the counter is incremented by 1. After scanning the entire image, the count value of the counter is the sum of the number of lines in the entire image that satisfy the judgment condition, and then make another judgment, namely:
If the value of the counter exceeds 2/3 of the cigarette stick height pixel value (i.e., exceeds 708 Pixels x 2/3 = 472 Pixels), the tear tape on said cigarette stick is considered present; A missing tape defect is assumed to have appeared.

The deviation detection of the opening tape described in step S3 has the following configuration. i.e.
For the preprocessed image of the front face of the cigarette bar, if the tear tape is present after the determination of the tear tape missing, step A1 of constructing a two-dimensional array:
step A11 of summing the number of pixels with a grayscale value of 255 in each column for the preprocessed image of the front face of the cigarette stick;
Step A12 of constructing a two-dimensional coordinate system: the origin is at the lower right corner of the preprocessed image of the front surface of the cigarette stick, the y-axis and the right boundary line of the preprocessed image of the front surface of the tobacco stick are on the same straight line, The direction of the y-axis is from bottom to top;
In the direction from right to left, for pixels in the i-th column of the preprocessed image of the front face of the cigarette stick, the serial number i of this column is the x-axis coordinate xi, and the total number of pixels whose grayscale value is 255 in this column is the two-dimensional coordinate (xi, yi) of the i-th pixel, where i=1, 2, . represents the number of columns of pixels, N=2332;
step A13 of repeating step A12 for each column of pixels to obtain all the two-dimensional coordinates to form a two-dimensional array (x1, y1), (x2, y2), ..., (xN, yN);
By traversing the two-dimensional array, excluding the left and right borders x1, xN, find the two x-axis coordinates with the largest total number of pixels stored in the array and record them as xk1, xk2, which is the unpacking Step A14, which is the position of the two boundary lines of the tape, and further extracts the intermediate position between xk1 and xk2 to obtain the x-axis coordinate x0 of the center point of the unsealed tape;
The distance between the center point of the tear tape and the right border of the preprocessed image of the front face of the cigarette stick is actually the distance between the center point of the tear tape and the y-axis, i.e. the x of the center point of the tear tape. Corresponding to the coordinate x0 of the coordinate axes, the distance between the center point of the tear tape and the right border of the preprocessed image of the front face of the cigarette stick was recorded as D=x0, and the distance D was compared with the standard distance STD. , to obtain the amount of deviation T = D - STD;
If the absolute value of the deviation amount T exceeds a preset threshold, it is considered that there is a deviation defect of the tear tape;
Preferably, if a tear tape deviation defect is present, the direction of deviation is determined according to the positive or negative of T, i.e., if T>0, the tear tape is shifted to the left, and if T<0, the tear tape is shifted to the right. step A2.

In some embodiments, if a tear tape misalignment defect is present, the direction of the misalignment can be identified according to the positive or negative of T, i.e., if T>0, the tear tape is displaced to the left; The tear tape shifts to the right.

The edge face polyline detection described in step S3 has the following configuration. i.e.
Obtaining a preprocessed image of each endface in step S2 and performing endface polyline detection according to the parameter MAX_LINE_NUM returned in step S205: if the MAX_LINE_NUM value is equal to 4, then output 1, the current endface is normal. Otherwise, output 0 to indicate that the current endface is defective.

As shown in FIG. 2, an apparatus for detecting defects on the surface of a tobacco rod, comprising:
a tobacco rod image capture device for capturing an image of the front and two end faces of the tobacco rod in the detection area to obtain an initial image of the front and two end faces of the tobacco rod and transmitting to the control module;
Each initial captured image was preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a hough line detection operation to preprocess the front and two end faces of the cigarette stick. An image is obtained, and defect detection is performed on the surface of the tobacco rod based on the preprocessed images of the front surface and two end surfaces of the tobacco rod, and a defect detection result, which is the presence or absence of a defect and the defect type if there is a defect, is output. A control module for generating rejection commands, if any.

The defect detection device further includes a cigarette stick conveying device, a detection light source, a defective product exclusion device, and a display module,
The tobacco rod conveying device is used to convey the tobacco rod to be detected to the detection area, and after the detection is completed, carry out the tobacco rod to be detected from the detection area; , the tobacco stick conveying device can use a conveyor belt to convey the tobacco rod to be detected to the detection area, and continue to convey the tobacco rod to the tobacco rod storage box/case each time detection is completed; Since a conveyor belt is used, it can be detected as if it is flowing, a plurality of rod-shaped cigarettes to be detected can be sequentially conveyed on one conveyor belt, and each rod-shaped tobacco can be detected after it reaches the detection area, and detection is completed. After that, continue detecting when the next cigarette stick arrives;
The detection light source is used to illuminate the detection area and ensure the brightness of the detection area,
The display module is used to receive and display the defect determination result output from the control module,
The rejection device is used to reject the defective cigarette stick according to the rejection command generated by the control module when the control module detects that the cigarette stick is defective; In an example, the reject device may use a manipulator to pick up the defective cigarette stick and place it in a designated defective cigarette stick collection box/case or the like when receiving a reject command from the control module. the reject device can also use a cylinder perpendicular to the co-planar direction of movement and the direction of conveyance of the conveying device, providing a collection box/case of defective cigarette rods/cases in the direction of movement of the piston of the cylinder and controlling Upon receiving the module's eviction command, the cylinder pushes the defective tobacco rod into the defective tobacco rod collection bin/case (e.g., if the tobacco rod transport device uses a conveyor belt, the direction of piston movement of the cylinder is the direction of the conveyor belt). set in a plane, perpendicular to the conveying direction of the conveyor belt, push the defective cigarette stick out of the conveyor belt after receiving a reject command from the control module, and collect the defective tobacco stick directly under the conveyor belt at the push-out position; A box/case can be provided to complete the collection).

In an embodiment of the present application, the control module is composed of a PLC controller and a ZYNQ module, the ZYNQ module implements image preprocessing and defect detection, and transmits the defect detection result to the display module for display; In some cases, the detection result is transmitted to the PLC controller, which generates a rejection command and controls the rejection device to reject the defective cigarette stick.

The cigarette stick image capturing device comprises three cameras, one camera is used to capture the front image of the cigarette stick, and the other two cameras are used to capture the end image of the cigarette stick. Used.

While the above description depicts preferred embodiments of the invention, as noted above, the invention is not limited to the forms disclosed herein and should not be considered to the exclusion of other embodiments. , is capable of use in a wide variety of combinations, modifications and environments, and is capable of modification according to the above teachings or skill or knowledge in the related field within the spirit of the invention described herein. be understood. Modifications and changes made by persons skilled in the art should fall within the protection scope of the appended claims of the present invention without departing from the spirit and scope of the present invention.

The present invention relates to cigarette bar detection, and more particularly to a method and apparatus for bar surface defect detection.

At present, the level of tobacco production in our country is accompanied by constant innovation in production technology, and the technology of the cigarette wrapping process has basically matured. still employ the most traditional manual detection means. The speed of the current cigarette outer packaging machine is very fast, basically reaching the level of 700-800 boxes/minute, and the newly developed cigarette packaging machine can reach as high as 900 boxes/minute. However, in such a fast-paced, high-strength packaging and manufacturing process, due to the nature of human beings, we cannot concentrate for a long period of time , so quality cannot be guaranteed. could lead to Accuracy of detection results in some extreme environments and standardized storage of detection data have also been issues for tobacco companies.

Some domestic tobacco processing and manufacturing companies often adopt a combination of various sensors and manual detection methods to control the inspection failure rate to generally acceptable levels, but defects such as tear tape misalignment is difficult to detect.

The purpose of the present invention is to overcome the shortcomings of the prior art, and based on the capture analysis of the surface of the cigarette stick , it is possible to realize the detection of defects on the surface of the cigarette stick , and it is convenient to operate in the automated detection. It is an object of the present invention to provide a method and apparatus for detecting surface defects in a stick-shaped cigarette pack .

The object of the present invention is achieved through the following technical means.

A method for detecting defects on the surface of a bar-shaped cigarette pack has the following configuration. i.e.
S1: Capturing images of the front face and two end faces of the cigarette stick to obtain initial images of the front face and two end faces of the cigarette stick ;
A method of capturing an image has any of the following configurations. i.e.
First, configure the output of the camera as YUV422 format, capture images of the front face and two end faces of the cigarette stick with the camera, then convert the YUV422 format images to YUV444 format, and convert the images in YUV422 format to YUV444 format. configuration to obtain initial images of two end faces ; After acquisition, each of the three images is converted from RGB565 to YUV422 format, and then the YUV422 format images are converted to YUV444 format to obtain initial images of the front face and two end faces of the cigarette stick .

Preferably, the two end faces of the cigarette stick refer to the two faces of the cigarette stick whose area is the smallest, and the camera captures images of each of the end faces and the front face of the cigarette stick .

S2: Each captured initial image is preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation and a Hough line detection operation, and the front and two end faces of the cigarette stick are obtained. obtaining a preprocessed image of
Said step S2 includes the following sub-steps S201-S206. i.e.
Step S201 for converting any of the initial images obtained in step S1 from YUV444 format image data to AXI4 - Stream video stream format ;
Step S202 of performing a median filter operation on the converted AXI4-Stream format image data to remove salt and pepper noise in the image to reduce unnecessary interference;
A Sobel edge detection operation is performed on the image data after the median filter. With the Sobel operator as the core, an edge detection operation is performed on the entire image of the image data after the median filter . discarding interference information such as box background, preserving all borders across the image, and reducing unnecessary interference, thus saving resources and time for post-processing step S203;
OTSU adaptive binarization is performed on the image data after edge detection, and the entire image is divided into monochrome images (i.e., the grayscale values are only 0 and 255), so that the boundary features can be identified. Step S204 for emphasizing and weakening useless information such as the background to facilitate subsequent judgment;
The Hough line detection operation is performed on the binarized image data to divide the straight lines of the entire image. The Hough line detection operation outputs the detected image data and returns a parameter ( MAX_LINE_NUM ) at the same time. is the sum of the number of oblique lines detected in step S205, and since the lines in the image after the previous edge detection are not continuous, the Hough line detection further improves the detection accuracy and facilitates the subsequent judgment step S205;
The initial images of the front face and the two end faces of the cigarette stick are pre-processed in steps S201 to S205, respectively, and preprocessed images of the front face and the two end faces of the cigarette stick are obtained in step S206.

S3: A step of performing surface defect detection of the cigarette stick , including tear tape missing detection, tear tape shift detection and end face polyline detection, according to the pre-processed images of the front face and two end faces of the cigarette stick .

The missing tape detection described in step S3 has the following configuration. i.e.
If the cigarette stick has a length of 280 mm and a height of 85 mm, and the surface image of the cigarette stick is captured by a camera, the photographing accuracy is 0.12 mm/pixel (where Pixel represents a pixel). The sum of each row in is 280 mm/(0.12 mm/Pixel) = 2332 Pixels, and the sum of each column is 85 mm/(0.12 mm/Pixel) = 708 Pixels;
Scanning to a preprocessed image of the front face of a cigarette stick :
By sequentially scanning the grayscale value of each pixel in the image from left to right and top to bottom, a counter is constructed that is initialized to 0, and during the scanning process the grayscale values of 11 consecutive pixels in a row are obtained. 255, the counter is incremented by 1. After scanning the entire image, the count value of the counter is the sum of the number of lines in the entire image that satisfy the judgment condition, and then make another judgment, namely:
If the value of the counter exceeds 2/3 of the pixel value of the cigarette stick height (i.e., exceeds 708 Pixels × 2/3 = 472 Pixels), it is determined that there is an opening tape of the cigarette stick , otherwise , it is judged that the defect of missing tape has appeared.

The deviation detection of the opening tape described in step S3 has the following configuration. i.e.
With respect to the preprocessed image of the front surface of the stick-shaped cigarette pack , if there is a tear-opening tape after it is determined that the tear-tape is missing, the following steps A11 to A14 are performed to construct a two-dimensional array : step A11 of summing the number of pixels with a grayscale value of 255 in each column for the foreground preprocessed image;
Step A12 of constructing a two-dimensional coordinate system: The origin is at the lower right corner of the preprocessed image of the front surface of the cigarette stick , and the y-axis and the right boundary line of the preprocessed image of the front surface of the cigarette stick are collinear. Yes, the direction of the y- axis is from bottom to top;
In the direction from right to left, for the i-th pixel of the preprocessed image of the front face of the cigarette stick , the serial number i of this column is the x-axis coordinate xi, and the grayscale value of this column is the number of pixels of 255. The two-dimensional coordinates (xi, yi) of the i-th pixel, where sum is the y-axis coordinate yi, where i=1, 2, . represents the number of columns of pixels in the image, N=2332;
step A13 of repeating step A12 for each column of pixels to obtain all the two-dimensional coordinates to form a two-dimensional array (x1, y1), (x2, y2), ..., (xN, yN);
By traversing the two-dimensional array, excluding the left and right borders x1, xN, find the two x-axis coordinates with the largest total number of pixels stored in the array and record them as xk1, xk2, which is the unpacking Step A14, which is the position of the two boundary lines of the tape, and further extracts the intermediate position between xk1 and xk2 to obtain the x-axis coordinate x0 of the center point of the unsealed tape;
The distance between the center point of the tear tape and the right boundary line of the preprocessed image of the front face of the cigarette stick is actually the distance between the center point of the tear tape and the y-axis, that is, the distance between the center point of the tear tape and the y-axis. Corresponding to the coordinate x0 of the x-coordinate axis, the distance between the center point of the tear tape and the right border of the preprocessed image of the front surface of the cigarette stick was recorded as D=x0, and the distance D was compared with the standard distance STD. to obtain the amount of deviation T = D - STD;
If the absolute value of the deviation amount T exceeds a preset threshold value, step A2, it is assumed that there is a deviation defect of the opening tape.

Preferably, if a tear tape deviation defect is present, the direction of deviation is determined according to the positive or negative of T, i.e., if T>0, the tear tape is shifted to the left, and if T<0, the tear tape is shifted to the right. shift to

The edge face polyline detection described in step S3 has the following configuration. i.e.
Obtaining the preprocessed image of each endface in step S2 and performing endface polyline detection according to the parameter MAX_LINE_NUM returned in step S205: if the MAX_LINE_NUM value is equal to 4, then output 1, the current endface is normal. Otherwise, output 0 to indicate that the current endface is defective.

A defect detection device for the surface of a bar-shaped cigarette pack ,
a cigarette stick image capture device for capturing an image of the front face and two end faces of the cigarette stick in the detection area, obtaining an initial image of the front face and two end faces of the cigarette stick and transmitting to the control module;
Each captured initial image was preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a Hough line detection operation to obtain the front and two end faces of the cigarette stick. A processed image is obtained, and based on the preprocessed images of the front surface and two end faces of the cigarette stick, defect detection is performed on the surface of the cigarette stick , and defect detection results indicating the presence or absence of defects and defect types when there are defects are output. , and a control module for generating a reject command if there is a defect.

The defect detection device further includes a stick-shaped cigarette pack conveying device, a detection light source, a defective product exclusion device, and a display module,
The rod-shaped cigarette pack conveying device is used to convey the rod-shaped cigarette pack to be detected to the detection area, and after the detection is completed, carry out the rod-shaped cigarette pack to be detected from the detection area,
The detection light source is used to illuminate the detection area and ensure the brightness of the detection area,
The display module is used to receive and display the defect determination result output from the control module,
Said rejection device is used to reject the defective cigarette stick according to the rejection command generated by the control module when the control module detects that the cigarette stick is defective.

The cigarette stick image capture device comprises three cameras, one camera is used to capture the front image of the cigarette stick , and the other two cameras are used to capture end images of the cigarette stick. used to

Based on the image capture analysis of the surface of the cigarette stick , the present invention can realize the detection of defects on the surface of the cigarette stick. offer.

3 is a flow chart of the method of the invention; 1 is a schematic diagram illustrating the principle of the device of the invention; FIG.

Hereinafter, the technical means of the present invention will be described in more detail with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following.

As shown in FIG. 1, the method for detecting defects on the surface of a stick-shaped cigarette pack has the following configuration. i.e.
S1: A step of capturing images of the front surface and two end surfaces of the rod-shaped cigarette packet to obtain initial images of the front surface and two end surfaces of the rod-shaped cigarette packet . The minimum two faces, each of which captures an image with a camera for each end face and the front face of the cigarette stick ;
A method of capturing an image has the following components. i.e.
First, the camera was configured for direct grayscale output, the configuration format was YUV422, the camera captured images of the front and two end faces of the cigarette stick , and then the YUV422 format images were converted to YUV444 format. , obtaining initial images of the front face and two end faces of the cigarette stick ;
Second, the output of the camera is configured as an RGB565 output, and the camera captures images of the front and two end faces of the cigarette stick to obtain three images, each of which is in RGB565 to YUV422 format. , and then the YUV422 format image is converted to YUV444 format to obtain initial images of the front and two end faces of the cigarette stick .

S2: Each captured initial image is preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation and a Hough line detection operation, and the front and two end faces of the cigarette stick are obtained. Obtaining a preprocessed image of The step S2 includes the following substeps S201-S206. i.e.
Step S201 for converting any of the initial images obtained in step S1 from YUV444 format image data to AXI4 - Stream video stream format ;
Step S202 of performing a median filter operation on the converted AXI4-Stream format image data to remove salt and pepper noise in the image to reduce unnecessary interference;
A Sobel edge detection operation is performed on the image data after the median filter. With the Sobel operator as the core, an edge detection operation is performed on the entire image of the image data after the median filter . discarding interference information such as box background, preserving all borders across the image, and reducing unnecessary interference, thus saving resources and time for post-processing step S203;
OTSU adaptive binarization is performed on the image data after edge detection, and the entire image is divided into monochrome images (i.e., the grayscale values are only 0 and 255), so that the boundary features can be identified. Step S204 for emphasizing and weakening useless information such as the background to facilitate subsequent judgment;
The Hough line detection operation is performed on the binarized image data to divide the straight line of the entire image. The Hough line detection operation outputs the detected image data and returns the parameter MAX_LINE_NUM at the same time. It is the total number of detected oblique lines, and since the lines in the image after the previous edge detection are not continuous, the Hough line detection further improves the detection accuracy and facilitates the subsequent judgment step S205;
The initial images of the front face and the two end faces of the cigarette stick are pre-processed in steps S201 to S205, respectively, and preprocessed images of the front face and the two end faces of the cigarette stick are obtained in step S206.

S3: A step of performing surface defect detection of the cigarette stick , including tear tape missing detection, tear tape shift detection and end face polyline detection, according to the pre-processed images of the front face and two end faces of the cigarette stick .

The missing tape detection described in step S3 has the following configuration. i.e.
If the cigarette stick has a length of 280 mm and a height of 85 mm, and the surface image of the cigarette stick is captured by a camera, the photographing accuracy is 0.12 mm/pixel (where Pixel represents a pixel). The sum of each row in is 280 mm/(0.12 mm/Pixel) = 2332 Pixels, and the sum of each column is 85 mm/(0.12 mm/Pixel) = 708 Pixels;
Scanning to a preprocessed image of the front face of a cigarette stick :
By sequentially scanning the grayscale value of each pixel in the image from left to right and top to bottom, a counter is constructed that is initialized to 0, and during the scanning process the grayscale values of 11 consecutive pixels in a row are obtained. 255, the counter is incremented by 1. After scanning the entire image, the count value of the counter is the sum of the number of lines in the entire image that satisfy the judgment condition, and then make another judgment, namely:
If the value of the counter exceeds 2/3 of the pixel value of the cigarette stick height (i.e., exceeds 708 Pixels x 2/3 = 472 Pixels), the tear tape of said cigarette stick is considered present; , is deemed to have exhibited the defect of missing tear tape.

The deviation detection of the opening tape described in step S3 has the following configuration. i.e.
With respect to the preprocessed image of the front surface of the bar-shaped cigarette packet , if the tear tape is present after it has been determined that the tear tape is missing, step A1 of constructing a two-dimensional array by the following processes A11 to A14 , that is,
step A11 of totaling the number of pixels with a grayscale value of 255 in each column for the preprocessed image of the front face of the cigarette stick ;
Step A12 of constructing a two-dimensional coordinate system: The origin is at the lower right corner of the preprocessed image of the front surface of the cigarette stick , and the y-axis and the right boundary line of the preprocessed image of the front surface of the cigarette stick are collinear. Yes, the direction of the y- axis is from bottom to top;
In the direction from right to left, for the i-th pixel of the preprocessed image of the front face of the cigarette stick , the serial number i of this column is the x-axis coordinate xi, and the grayscale value of this column is the number of pixels of 255. The two-dimensional coordinates (xi, yi) of the i-th pixel, where sum is the y-axis coordinate yi, where i=1, 2, . represents the number of columns of pixels in the image, N=2332;
step A13 of repeating step A12 for each column of pixels to obtain all the two-dimensional coordinates to form a two-dimensional array (x1, y1), (x2, y2), ..., (xN, yN);
By traversing the two-dimensional array, excluding the left and right borders x1, xN, find the two x-axis coordinates with the largest total number of pixels stored in the array and record them as xk1, xk2, which is the unpacking Step A14, which is the position of the two boundary lines of the tape, and further extracts the intermediate position between xk1 and xk2 to obtain the x-axis coordinate x0 of the center point of the unsealed tape;
The distance between the center point of the tear tape and the right boundary line of the preprocessed image of the front face of the cigarette stick is actually the distance between the center point of the tear tape and the y-axis, that is, the distance between the center point of the tear tape and the y-axis. Corresponding to the coordinate x0 of the x-coordinate axis, the distance between the center point of the tear tape and the right border of the preprocessed image of the front surface of the cigarette stick was recorded as D=x0, and the distance D was compared with the standard distance STD. to obtain the amount of deviation T = D - STD;
If the absolute value of the deviation amount T exceeds a preset threshold, it is considered that there is a deviation defect of the tear tape;
Preferably, if a tear tape deviation defect is present, the direction of deviation is determined according to the positive or negative of T, i.e., if T>0, the tear tape is shifted to the left, and if T<0, the tear tape is shifted to the right. step A2.

In some embodiments, if a tear tape misalignment defect is present, the direction of the misalignment can be identified according to the positive or negative of T, i.e., if T>0, the tear tape is displaced to the left; The tear tape shifts to the right.

The edge face polyline detection described in step S3 has the following configuration. i.e.
Obtaining the preprocessed image of each endface in step S2 and performing endface polyline detection according to the parameter MAX_LINE_NUM returned in step S205: if the MAX_LINE_NUM value is equal to 4, then output 1, the current endface is normal. Otherwise, output 0 to indicate that the current endface is defective.

As shown in FIG. 2, an apparatus for detecting defects on the surface of a stick-shaped cigarette pack ,
a cigarette stick image capture device for capturing an image of the front face and two end faces of the cigarette stick in the detection area, obtaining an initial image of the front face and two end faces of the cigarette stick and transmitting to the control module;
Each captured initial image was preprocessed with a format conversion operation, a median filter operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a Hough line detection operation to obtain the front and two end faces of the cigarette stick. A processed image is obtained, and based on the preprocessed images of the front surface and two end faces of the cigarette stick, defect detection is performed on the surface of the cigarette stick , and defect detection results indicating the presence or absence of defects and defect types when there are defects are output. , and a control module for generating a reject command if there is a defect.

The defect detection device further includes a stick-shaped cigarette pack conveying device, a detection light source, a defective product exclusion device, and a display module,
The stick-shaped cigarette pack conveying device is used to carry the stick-shaped cigarette pack to be detected to the detection area, and after the detection is completed, carry out the stick-shaped cigarette pack to be detected from the detection area; In an embodiment, the stick-shaped cigarette pack conveying device can use a conveyor belt to carry the stick-shaped cigarette pack to be detected to the detection area, and every time detection is completed, the stick-shaped cigarette pack is transferred to the storage box of the stick-shaped cigarette pack. /Continue conveying to the case; Since a conveyor belt is used, it can be detected as if it is flowing, and a plurality of rod-shaped cigarette packs to be detected can be sequentially conveyed on one conveyor belt, and each rod-shaped cigarette pack reaches the detection area. After the detection is completed, the detection is continued when the next cigarette pack arrives;
The detection light source is used to illuminate the detection area and ensure the brightness of the detection area,
The display module is used to receive and display the defect determination result output from the control module,
The reject device is used to reject the defective cigarette stick according to the reject command generated by the control module when the control module detects that the cigarette stick is defective; In the embodiment of , the defective product rejection device can use a manipulator to pick up the defective cigarette stick and put it into the designated defective cigarette stick collection box/ put into a case, etc.; the reject device can also use a cylinder perpendicular to the co-planar movement direction and the conveying direction of the conveying device, and remove the stick-shaped cigarette packet collection box / A case is provided, and upon receipt of an eject command from the control module, a cylinder pushes the defective cigarette bar into the defective cigarette bar collection box/case (e.g., if the cigarette bar conveyor uses a conveyor belt, The piston movement direction of the cylinder is set in the plane where the conveyor belt is located, perpendicular to the conveying direction of the conveyor belt, and after receiving the control module's rejection command, pushes the defective cigarette stick out of the conveyor belt, and pushes the conveyor belt in the pushing position. collection can be completed by placing a defective cigarette stick collection box/case directly under the .

In an embodiment of the present application, the control module is composed of a PLC controller and a ZYNQ module, the ZYNQ module implements image preprocessing and defect detection, and transmits the defect detection result to the display module for display; In some cases, the detection result is transmitted to the PLC controller, and the PLC controller generates a rejection command and controls the rejection device to reject the defective cigarette stick .

The cigarette stick image capture device comprises three cameras, one camera is used to capture the front image of the cigarette stick , and the other two cameras are used to capture end images of the cigarette stick. used to

While the above description depicts preferred embodiments of the invention, as noted above, the invention is not limited to the forms disclosed herein and should not be considered to the exclusion of other embodiments. , is capable of use in a wide variety of combinations, modifications and environments, and is capable of modification according to the above teachings or skill or knowledge in the related field within the spirit of the invention described herein. be understood. Modifications and changes made by persons skilled in the art should fall within the protection scope of the appended claims of the present invention without departing from the spirit and scope of the present invention.

Claims (10)

A method for detecting defects on the surface of a tobacco rod, comprising:
a step S1 of capturing an image of the front face and two end faces of the cigarette stick with a camera to obtain an initial image of the front face and the two end faces of the cigarette stick;
Each of the initial images captured is preprocessed with a format conversion operation, a median filtering operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a Hough line detection operation to obtain the front and two end faces of the cigarette stick. a step S2 of obtaining a processed image;
a step S3 of performing surface defect detection of the cigarette stick, including detection of missing tape, deviation of the tear tape, and polyline detection of the end face, according to the preprocessed images of the front face and the two end faces of the cigarette bar;
A method for detecting defects on the surface of a tobacco stick, comprising: 2. The two end faces of the tobacco stick refer to the two faces with the smallest areas of the tobacco stick, and the images of each of the end faces and the front face of the tobacco stick are respectively captured by a camera. 3. The method for detecting defects on the surface of a tobacco rod according to . The method for capturing an image in step S1 includes the following components:
First, the camera is configured for direct grayscale output, the configuration format is YUV422, the camera captures images of the front and two end faces of the cigarette bar, and then the YUV422 format images are converted to YUV444 format. to obtain initial images of the front and two ends of the tobacco rod;
Secondly, the output of the camera is configured as an RGB565 output, and the camera captures images of the front and two end faces of the cigarette rod to obtain three images, and then to each of the three images. converting from RGB565 to the YUV422 format and then converting the YUV422 format image to the YUV444 format to obtain initial images of the front and two end faces of the tobacco rod;
The method for detecting defects on the surface of a tobacco rod according to claim 1, characterized by: Said step S2 follows the next sub-steps S201-S206, i.e., a step S201 of converting any of said initial images obtained in said step S1 from said YUV444 format image data into a dedicated format of an AXI4 protocol Stream video stream. ,
Step S202 of performing a median filter operation on the converted AXI4-Stream format image data to remove salt and pepper noise in the image;
A Sobel edge detection operation is performed on the image data after the median filter. With the Sobel operator as the core, an edge detection operation is performed on the entire image of the image data after the median filter. step S203 of discarding interference information such as the background of and retaining all borders of the entire image;
By performing OTSU adaptive binarization operation on the image data after the edge detection and dividing the entire image into monochrome images (i.e., the grayscale values are only 0 and 255), the boundary line features and weakening useless information such as the background at step S204;
Perform a Hough line detection operation on the binarized image data, divide the straight line of the entire image, and output the image data detected through the Hough line detection operation. Step S205, which is the sum of the number of diagonal lines detected in
The initial images of the front surface and the two end surfaces of the tobacco rod are subjected to the preprocessing of steps S201 to S205, respectively, to obtain preprocessed images of the front surface and the two end surfaces of the tobacco rod in step S206.
4. The method for detecting defects on the surface of a tobacco stick according to claim 3, characterized by comprising: The unsealing tape missing detection described in step S3 has the following configuration, that is,
If the cigarette stick packet has a length of 280 mm and a height of 85 mm, and the surface image of the cigarette stick is captured by the camera, the photographing accuracy is 0.12 mm/pixel (where Pixel represents a pixel). The sum of each row in is 280 mm/(0.12 mm/Pixel) = 2332 Pixels, and the sum of each column is 85 mm/(0.12 mm/Pixel) = 708 Pixels;
Scanning into the preprocessed image of the front face of the cigarette stick: The grayscale value of each pixel of the image is sequentially scanned from left to right and top to bottom, constructing a counter initialized to 0 and scanning one row during the scanning process. If the grayscale value of 11 consecutive pixels is 255, the counter is incremented by 1, and after scanning the entire image, the count value of the counter is the sum of the number of rows satisfying the criterion in the entire image, and this When the decision is made again, i.e.,
If the value of the counter exceeds 2/3 of the height pixel value of the cigarette stick (i.e., exceeds 708 Pixels x 2/3 = 472 Pixels), it is determined that the tear tape of the cigarette stick is present; It is determined that the opening tape missing defect has appeared,
The method for detecting defects on the surface of a tobacco rod according to claim 1, characterized by: The detection of deviation of the opening tape described in step S3 has the following configuration, that is,
Step A1 of constructing a two-dimensional array of the preprocessed image of the front surface of the cigarette bar, if the tear tape is present after the determination of the tear tape missing:
step A11 of totaling the number of pixels with a grayscale value of 255 in each column for the preprocessed image of the front face of the cigarette stick;
Step A12 of constructing a two-dimensional coordinate system: The origin is at the lower right corner of the preprocessed image of the front surface of the tobacco stick, and the y-axis and the right boundary line of the preprocessed image of the front surface of the tobacco rod are aligned on the same straight line. Yes, the direction of the y-axis is from bottom to top; the x-axis and the lower boundary line of the preprocessed image of the front surface of the tobacco rod are collinear, and the direction of the x-axis is from right to left;
In the direction from right to left, for the i-th row of pixels of the preprocessed image of the front face of said cigarette stick, the serial number i of this row is the x-axis coordinate xi, and the grayscale value of this row is the number of pixels of 255. The two-dimensional coordinates (xi, yi) of the pixel in the i-th column, where sum is the y-axis coordinate yi, where i=1, 2, . represents the number of columns of pixels in the processed image, N=2332;
step A13 of repeating said step A12 for each column of pixels, obtaining all the two-dimensional coordinates to form a two-dimensional array (x1, y1), (x2, y2), ..., (xN, yN);
By traversing the two-dimensional array, excluding the left and right borders x1, xN, find the two x-axis coordinates with the largest total number of pixels stored in the array and record them as xk1, xk2, which is A step A14, which is the position of the two boundary lines of the tear tape and extracts the intermediate position between xk1 and xk2 to obtain the x-axis coordinate x0 of the center point of the tear tape;
The distance between the center point of the tear tape and the right boundary line of the preprocessed image of the front face of the cigarette bar is actually the distance between the center point of the tear tape and the y-axis, i.e. the distance between the center point of the tear tape and the y-axis. The distance between the center point of the opening tape and the right border of the preprocessed image of the front surface of the cigarette stick is recorded as D=x0, and the distance D is Compare with the standard distance STD to get the amount of deviation T=D−STD;
If the absolute value of the deviation amount T exceeds a preset threshold value, step A2 in which it is considered that there is a deviation defect of the opening tape.
The method for detecting defects on the surface of a tobacco rod according to claim 1, characterized by: The edge face polyline detection described in step S3 is configured as follows:
obtaining a pre-processed image of each said edge in said step S2 and performing edge polyline detection according to said parameter MAX_LINE_NUM returned in said step S205: if said MAX_LINE_NUM value is equal to 4, output 1; indicates that the end face is normal, otherwise outputs 0, indicating that the current end face is defective;
The method for detecting defects on the surface of a tobacco rod according to claim 1, characterized by: A tobacco rod surface defect detection apparatus for performing defect detection of a tobacco rod surface by the method according to any one of claims 1 to 7,
a tobacco rod image capture device for capturing an image of the front and two end faces of the tobacco rod in the detection area to obtain an initial image of the front and two end faces of the tobacco rod and transmitting to the control module;
Each of the initial images captured is preprocessed with a format conversion operation, a median filtering operation, a sobel edge detection operation, an OTSU adaptive binarization operation, and a Hough line detection operation to obtain the front and two end faces of the cigarette stick. Obtaining a processed image, performing defect detection on the surface of the tobacco rod based on the preprocessed images of the front surface and two end surfaces of the tobacco rod, and outputting defect detection results indicating the presence or absence of a defect and a defect type if there is a defect; a control module for generating a rejection command if there is a defect;
A defect detection device for a stick-shaped tobacco surface, comprising: further comprising a stick-shaped tobacco conveying device, a detection light source, a defective product rejection device, and a display module;
The tobacco rod conveying device is used for conveying a tobacco rod to be detected to the detection area, and carrying out the tobacco rod to be detected from the detection area after detection is completed,
The detection light source is used to illuminate the detection area and ensure the brightness of the detection area;
The display module is used to receive and display the defect determination results output from the control module,
The rejection device is used to reject the defective cigarette stick according to the rejection command generated by the control module when the control module detects that the cigarette stick is defective.
9. The apparatus for detecting defects on the surface of a stick-shaped tobacco according to claim 8, wherein: The cigarette stick image capture device includes three cameras, one of which is used to capture the front image of the cigarette stick, and the other two of which are used to capture the end image of the cigarette stick. 9. The device for detecting defects on the surface of tobacco rods according to claim 8, which is used for trapping.

Images