A plastic film defect detection system

Abstract

The application discloses a plastic film defect detection system and relates to the technical field of plastic film production. The system comprises a light source supplementing module for uniformly irradiating the surface of the plastic film to supplement imaging brightness, an image collecting module for collecting and transmitting image data of the surface of the plastic film, and an analysis and recognition unit of a control module for processing the collected image. The collected image data is preprocessed, the data that does not need to be detected is filtered, the analysis and recognition pressure of subsequent data is reduced, and the accuracy and speed of subsequent recognition are improved at the same time. Then, the color distribution and the connected domain shape structure are combined for analysis by using a defect judgment model, so that the problem that the prior art cannot distinguish and judge different kinds of defects and has a large pressure for processing a large amount of data is solved.

Description

Technical Field

[0001] This invention relates to the field of plastic film production technology, specifically to a plastic film defect detection system. Background Technology

[0002] Plastic film is a common packaging material widely used in food, pharmaceutical, and chemical industries. The quality of the plastic film is crucial to its effectiveness during use. Defects such as bubbles, cracks, and dirt can cause the film to break during use, affecting its strength.

[0003] In the inspection of plastic films, due to the large surface area of ​​plastic films, a large amount of image data needs to be analyzed and processed. Existing inspection systems can basically only detect defects in plastic films or only identify single defects, and cannot distinguish between multiple defects on plastic films, thus failing to provide a basis for subsequent production adjustments. Therefore, we provide a plastic film defect detection system. Summary of the Invention

[0004] The purpose of this invention is to provide a plastic film defect detection system.

[0005] The technical problem solved by this invention is: how to analyze the color distribution and connected component shape structure by preprocessing the acquired image data and using a defect judgment model, thereby solving the problem that existing technologies cannot distinguish between different types of defects and have high processing pressure on large amounts of data.

[0006] This invention can be achieved through the following technical solution: a plastic film defect detection system, comprising: a light source supplement module, used to uniformly irradiate the surface of the plastic film to supplement the imaging brightness;

[0007] The image acquisition module is used to acquire and transmit image data from the surface of the plastic film.

[0008] The control module, whose analysis and recognition unit is used to process the acquired images and identify and detect defects.

[0009] A further technical improvement of the present invention is that the system also includes a mechanical conveying module, comprising a conveying device and an encoder, for conveying the plastic film product to the detection area at a predetermined speed. The image acquisition module is positioned above the mechanical conveying module.

[0010] A further technical improvement of the present invention is that the analysis and recognition unit of the control module preprocesses the acquired image, and the preprocessing steps include:

[0011] S1: Perform Gaussian filtering on the acquired image;

[0012] S2: The gray values ​​of the image in the horizontal and vertical directions are obtained by convolving with the template, and the gradient value and orientation angle at the corresponding position are obtained based on the gray values;

[0013] S3: Filter the gradient values ​​in step S2 and retain the image data corresponding to the above gradient values;

[0014] S4: Perform image enhancement operations on the retained image data.

[0015] A further technical improvement of the present invention is that the specific operation of filtering and retaining the image data corresponding to the gradient values ​​includes retaining the gradient value data in the neighborhood of the gradient value that is near the maximum value, and filtering the gradient values ​​outside this range.

[0016] A further technical improvement of the present invention is that the image enhancement operation is specifically contrast stretching, including histogram statistics and grayscale value mapping, which linearly stretches and maps the grayscale functions of the original image data in the horizontal and vertical directions according to a percentage.

[0017] A further technical improvement of the present invention is that: the analysis and identification unit performs defect identification through a defect judgment model, and the defect identification includes the identification of the color features and shape features of the defect;

[0018] The color second moment of the image data is calculated using a formula, and the color distribution range of the image is obtained by analyzing the magnitude of the color second moment.

[0019] By marking all gradient maxima and their neighborhood ranges in the image data, the distance between adjacent neighborhoods is calculated, and the neighborhood ranges with adjacent neighborhood distances less than a set value are formed into connected components. Then, the shape and structure of the connected components are analyzed, and the specific defect type is determined by combining different color distributions and shape structures.

[0020] A further technical improvement of the present invention is that the analysis and identification unit also classifies and statistically analyzes the identified defects, records the number of occurrences and coordinate locations of different defects, forms a scatter plot and performs correlation analysis, and provides guidance for production based on the analysis results.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] By preprocessing the acquired image data, data that does not need to be detected is filtered out, reducing the pressure on subsequent data analysis and recognition, and indirectly improving the accuracy and speed of subsequent recognition. Then, the defect judgment model is used to analyze the color distribution and connected component shape structure together, solving the problem that the existing technology cannot distinguish between different types of defects and has a high pressure on processing large amounts of data. Attached Figure Description

[0023] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0024] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0025] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0026] Please see Figure 1 As shown, a plastic film defect detection system includes a light source supplementation module, a mechanical transmission module, an image acquisition module, and a control module.

[0027] The light source supplement module is used to uniformly illuminate the surface of the plastic film to improve the image quality of the imaging. Specifically, the light source supplement module is set as an array of LED light groups.

[0028] The mechanical conveying module consists of a conveying device and an encoder, and is used to convey plastic film products to the inspection area at a predetermined speed.

[0029] The image acquisition module consists of a CCD line scan camera, a lens, and an image acquisition card. It is used for image data acquisition and transmission. The CCD camera acquires the image information of the plastic film product to be inspected completely according to a certain acquisition strategy and transmits the image data to the control module. The image acquisition card is used to acquire the image signal into the computer and save it on the hard drive in the form of a data file. The image acquisition card is the interface between the image acquisition part and the image processing part.

[0030] The control module includes an analysis and identification unit, a display unit, an alarm unit, and a communication unit. The analysis and identification unit is used to process the acquired images and identify and detect defects. The display unit is a screen that can display the current working status, making it convenient for staff to observe and adjust the parameters. The alarm unit vibrates and sounds an alarm when the system detects related defects in the film.

[0031] When the plastic film passes under the image acquisition module, its image is captured by the CCD linear array. The analysis and recognition unit in the control module first preprocesses the acquired image, which includes the following steps:

[0032] S1: First, the acquired image is processed by Gaussian filtering, which has the functions of data smoothing and noise reduction;

[0033] S2: Obtain the gray values ​​of the image in the horizontal and vertical directions by convolving with the template, and label them as G respectively. X G Y Substitute the grayscale values ​​in the horizontal and vertical directions into the following formula:

[0034]

[0035]

[0036] This allows us to obtain the gradient value and direction angle at the corresponding coordinates (a, b);

[0037] S3: Filter the gradient values ​​obtained above, retain the gradient values ​​in the neighborhood δ near the maximum gradient value, and filter the gradient values ​​outside this range.

[0038] S4: Perform image enhancement operations on the retained image data. In this embodiment, contrast stretching is used, which can effectively enhance the thin film defect image and improve detection accuracy.

[0039] Contrast stretching consists of two parts:

[0040] (1) Histogram statistics

[0041] The two endpoints a1 and a2 are determined by setting the number of pixels at both ends of the histogram and the percentage of the total number of pixels in the image. Gray values ​​less than a1 are directly set to 0, and gray values ​​greater than a2 are directly set to 255. In this embodiment, the percentage is set to 0.005.

[0042] (2) Gray value mapping

[0043] In the histogram, the pixel grayscale values ​​before the two endpoints are mapped to pixel grayscale values ​​according to the formula:

[0044] Where G(X) and G(Y) are the grayscale functions in the horizontal and vertical directions of the original image data, G'(X) and G'(Y) are the enhanced grayscale functions, and N is the transformation rule. The transformation rule is to map the pixel grayscale values ​​between a1 and a2 to the range of 0 to 255 in a linear relationship.

[0045] It should be noted that after image preprocessing, a large amount of normal data is filtered out, which greatly reduces the pressure of subsequent data analysis and recognition and improves the accuracy and speed of subsequent recognition.

[0046] Ideally, a defect-free plastic film image would have uniform grayscale values. However, due to random factors such as the inhomogeneity of the actual film material, vibrations caused by changes in film tension, uneven illumination, and internal noise of the imaging device, even if the film surface is free of defects, its image grayscale values ​​cannot be completely uniform. Therefore, in the image preprocessing process, only the gradient maxima are retained. The image data within the neighborhood of these gradient maxima includes both defective and interfering image data.

[0047] A defect judgment model is constructed within the analysis and recognition unit, and defects are identified in the aforementioned retained image data using the defect judgment model.

[0048] In constructing the defect judgment model, identification and judgment are performed from two aspects: defect color features and shape features.

[0049] First, we extract the grayscale features from the grayscale image to analyze the color distribution. Since the grayscale values ​​of pixels corresponding to different colors are different, we use the second moment of color to characterize the color distribution width.

[0050] When calculating the second color moment of an image, the formula is used. The second color moment σ of the image is obtained;

[0051] Where, p i The value of the i-th pixel in the grayscale image of the surface of the plastic film is represented by N, and N represents the total number of pixels in the grayscale image of the surface.

[0052] The second color moment can effectively represent the color distribution in an image. The larger the σ is, the wider the color distribution range of the image, indicating that there may be more color values ​​in the surface grayscale image; conversely, the smaller the σ is, the narrower the color distribution range of the image, indicating that there may be fewer color values ​​in the surface grayscale image, that is, there may be only one color value.

[0053] In actual production, if bubbles or thin skin appear, the areas where bubbles or thin skin occur will appear bright white, resulting in a total of two color distributions. When dirt appears, due to the different thicknesses of the dirt or the different sources of contamination, more than two color distributions will appear. When cracks or wrinkles appear, the cracks or overlapping parts will appear dark gray, while the crack edges or the recessed parts of the wrinkles will appear bright white.

[0054] With the color distribution confirmed, a Cartesian coordinate system is established with the width of the plastic film as the vertical axis and the conveying direction of the plastic film as the horizontal axis. Then, each image captured by the CCD line scan camera can be matched with the corresponding coordinates, and each pixel can be assigned a coordinate value.

[0055] During image preprocessing, all gradient maxima and their neighborhood ranges are marked. When the distance between the neighborhoods of adjacent maxima is less than a set value, it is considered that the neighborhood ranges of the two maxima can form a connected region, and the entire connected region range is used as the object of defect detection.

[0056] When the determined detection range is a dotted circular structure, the defect is determined to be either a bubble or a black spot. Bubbles and black spots show opposite color distributions: bubble defects are bright white in the middle and have no color difference in the surrounding area, while black spot defects are dark black in the middle and have no color difference in the surrounding area. This allows us to determine the specific defect type.

[0057] When the determined detection range is a strip or linear structure, the defect is determined to be a crack or a wrinkle. At this time, the connected region also presents a strip structure. Furthermore, based on the width of the color distribution lines, crack defects and wrinkle defects can be distinguished. Obviously, the width of the color distribution lines corresponding to crack defects is smaller than that of wrinkle defects, so the specific defect type can be determined.

[0058] The color distribution structure of dirt defects is irregular, and the defined detection range is also without a specific shape. At the same time, the color distribution may show more than two colors.

[0059] The analysis and identification unit also classifies and statistically analyzes the detected defects, records the frequency and coordinate location of different defects, marks points in the constructed virtual Cartesian coordinate system, and generates a scatter plot. Correlation analysis is performed on the scatter plot, assuming a banded distribution, and a linear regression equation and correlation coefficient are derived. A correlation coefficient greater than 0.8 indicates a banded distribution; a correlation coefficient less than 0.7 indicates a uniform distribution.

[0060] When the distribution appears in a banded pattern, it indicates a problem with the plastic film production equipment, and the alarm unit will issue a stop and inspection signal.

[0061] When the defect distribution is uniform, it indicates that there is no correlation between the defect locations and that the defect occurrence is sporadic. In this case, a virtual moving frame is set within a certain range, and the defect density within the moving frame is calculated in real time. When the defect density exceeds the set value, the alarm unit generates a defect density signal. At this time, the quality of the plastic film is poor, and it is necessary to extract the defect type and quantity within this range for specific problem investigation.

[0062] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A plastic film defect detection system, characterized in that, include: A light source supplement module is used to uniformly illuminate the surface of the plastic film to supplement the imaging brightness; The image acquisition module is used to acquire and transmit image data from the surface of the plastic film. The control module, whose analysis and recognition unit is used to process the acquired images and identify and detect defects; During image preprocessing, all gradient maxima and their neighborhood ranges in the image are marked. When the distance between the neighborhoods of adjacent maxima is less than a set value, it is considered that the neighborhood ranges of the two maxima can form a connected region, and the entire connected region range is used as the object of defect detection. The analysis and identification unit identifies defects through a defect judgment model, and the defect identification includes the identification of the color and shape features of the defects. Using formula The color second moment of the preprocessed image data was calculated. The color distribution range of the image is obtained by analyzing the magnitude of the second-order color moments; among which, The grayscale value of the i-th pixel in the grayscale image of the surface of the plastic film is represented by N, and N represents the total number of pixels in the grayscale image of the surface. The coordinates of all gradient maxima in the image grayscale gradient and their neighborhood ranges are marked. The distance between adjacent neighborhoods is calculated, and the neighborhood ranges with adjacent neighborhood distances less than a set value are formed into connected components. Then, the shape and structure of the connected components are analyzed. The specific defect type is determined by combining different color distributions and shape structures. If the shape and structure are circular or dot-like, the defects are bubbles or black spots, which are opposite in color distribution; if the shape and structure are strip-like or linear, the defects are cracks or wrinkles, which can be further distinguished by the difference in line width; if the shape and color distribution are irregular, the defects are dirt.

2. The plastic film defect detection system according to claim 1, characterized in that, It also includes a mechanical conveying module, comprising a conveying device and an encoder, for conveying the plastic film product to the detection area at a predetermined speed, wherein the image acquisition module is positioned above the mechanical conveying module.

3. The plastic film defect detection system according to claim 1, characterized in that, The analysis and recognition unit of the control module preprocesses the acquired image. The preprocessing steps include: S1: Perform Gaussian filtering on the acquired image; S2: The gray values ​​of the image in the horizontal and vertical directions are obtained by convolving with the template, and the gradient value and orientation angle at the corresponding position are obtained based on the gray values; S3: Filter the gradient values ​​in step S2 and retain the image data corresponding to the above gradient values; S4: Perform image enhancement operations on the retained image data.

4. The plastic film defect detection system according to claim 3, characterized in that, The specific operation of filtering and retaining the image data corresponding to the gradient values ​​includes retaining the gradient value data in the neighborhood of the gradient value that is near the maximum value, and filtering the gradient values ​​outside this range.

5. A plastic film defect detection system according to claim 3, characterized in that, The image enhancement operation is specifically contrast stretching, which includes histogram statistics and grayscale mapping, linearly stretching and mapping the grayscale functions of the original image data in the horizontal and vertical directions according to a percentage.

6. The plastic film defect detection system according to claim 1, characterized in that, The analysis and identification unit also classifies and statistically analyzes the identified defects, records the frequency and coordinate location of different defects, forms a scatter plot, performs correlation analysis, and provides guidance for production based on the analysis results.

Images