Coenzyme Q10 soft capsule appearance defect detection method based on image recognition

Abstract

The application discloses a coenzyme Q10 soft capsule appearance defect detection method based on image recognition and relates to the technical field of capsule detection. The application solves the problem that different types of defects such as "size deviation" and "breakage wrinkle" cannot be effectively recognized, and the reliability and pertinence of the detection result are insufficient. The application confirms standard images and a set of images to be analyzed from a large number of images, and then selects features to be analyzed, and focuses on the most significant difference images for elliptical perimeter comparison. This way can efficiently reduce the analysis range, accurately locate the area where wrinkles, breakage or size abnormalities may exist, and improve the pertinence and efficiency of defect analysis.

Description

Technical Field

[0001] This invention relates to the field of capsule detection technology, specifically to a method for detecting appearance defects in coenzyme Q10 soft capsules based on image recognition. Background Technology

[0002] Coenzyme Q10 soft capsules are a common health food and pharmaceutical dosage form, and their appearance quality is directly related to product stability, content safety, and consumer trust. In industrial production, soft capsules are prone to appearance defects such as deformation, wrinkles, breakage, and uneven size due to factors such as formulation (e.g., fluctuations in gelatin viscosity), process (e.g., deviations in the precision of pelleting machine molds, and uncontrolled drying temperature and humidity), and transportation collisions. These defects not only affect product compliance but may also lead to leakage of contents, oxidative deterioration, and even medication safety risks. Therefore, appearance defect detection is a core aspect of quality control in the production of coenzyme Q10 soft capsules.

[0003] The mainstream method for inspecting the appearance of soft capsules in the industry still relies on "manual sampling + visual observation": inspectors compare standard samples with the naked eye, using simple tools such as magnifying glasses and calipers to judge the shape, color, and surface condition of the soft capsules. However, this method has significant limitations: on the one hand, manual inspection relies on subjective experience, resulting in low accuracy in identifying minute defects (such as fine marks less than 0.1mm or slight local deformations), and is prone to missed detections and misjudgments due to visual fatigue; on the other hand, manual inspection is inefficient (single batch sampling inspection can take several hours), making it difficult to meet the full inspection requirements of continuous industrial production, especially in terms of comprehensive defect screening of the soft capsule's annular surface (such as hidden deformations on the sides and back of the capsule).

[0004] Some companies have attempted to introduce basic machine vision inspection solutions, but existing technologies mostly focus on single-angle image analysis and have not designed inspection logic adapted to the "ring-shaped three-dimensional structure" of soft capsules. For example, extracting contour features only from the frontal image cannot fully reflect the circumferential shape differences of the capsule. At the same time, existing solutions lack quantitative analysis of "contour feature dispersion," making it difficult to accurately identify defects that are "locally deformed but appear normal from a single angle" (such as one side of the capsule bulging and the other side flattened). Furthermore, no correlation verification mechanism between "contour features and elliptical circumference" has been established, making it impossible to effectively distinguish between different types of defects such as "size deviation" and "damage and wrinkles," resulting in insufficient reliability and specificity of the inspection results.

[0005] In summary, current coenzyme Q10 soft capsule appearance inspection technology suffers from pain points such as "low accuracy, poor efficiency, incomplete coverage, and weak defect differentiation ability." There is an urgent need for a detection method that can achieve comprehensive, high-precision, and automated defect identification and can quantitatively verify defect types to meet the strict quality control requirements of industrial production. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an image recognition-based method for detecting appearance defects in coenzyme Q10 soft capsules. This method solves the problem of insufficient reliability and specificity of detection results due to the inability to effectively distinguish between different types of defects such as "size deviation" and "damage and wrinkles".

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting appearance defects in coenzyme Q10 soft capsules based on image recognition, comprising the following steps:

[0008] Step 1: Place the soft capsule at the detection location and use a high-definition camera to acquire an overall image of the soft capsule. Perform contour detection on the overall image at the designated location to confirm the contour features associated with the overall image. Based on the confirmation results, identify whether the detection process is abnormal. The specific method is as follows:

[0009] Extract the overall image at the set location and mark it as the image to be processed. Confirm the RGB values ​​associated with different pixels in the image to be processed, and confirm the gray values ​​associated with corresponding pixels in the image to be processed according to the preset weights. Then, convert the image to be processed into a grayscale image based on the different gray values ​​associated with different pixels.

[0010] The Sobel algorithm is used to identify the horizontal and vertical gradients associated with individual pixels in a grayscale image. Based on the identified horizontal and vertical gradients, the overall gradient associated with the corresponding pixel is determined. The confirmed comprehensive gradient is compared with the preset value Y1. If the comprehensive gradient is greater than Y1, the corresponding pixel is recorded as the gradient pixel. Otherwise, no marking is performed.

[0011] The gradient pixels associated with the grayscale image are confirmed sequentially, and adjacent gradient pixels are connected to confirm the grayscale contour associated with the grayscale image. Then, the grayscale image is placed in a set of two-dimensional coordinate systems to confirm the two-dimensional coordinates associated with different points within the grayscale contour. The confirmed sets of two-dimensional coordinates are averaged to confirm the mean coordinates. The point where the mean coordinates are located is recorded as the center point of the current grayscale image.

[0012] Then, identify the two sets of contour points closest to and farthest from the center point on the grayscale contour. The distance between the closest contour point and the center point is denoted as L1, and the distance between the farthest contour point and the center point is denoted as L2. The contour feature Bz of the corresponding grayscale image is confirmed by using L2÷L1=Bz. Then, the contour features associated with different images to be processed are confirmed. The variance of the confirmed contour features Bz is processed to confirm the calibration variance. If the calibration variance > Y2, it means that there is an abnormality in the detection process, and a detection abnormality signal is directly generated.

[0013] If the calibration variance is less than or equal to Y2, it means that the detection process is normal and no processing is required. Y2 is a preset value.

[0014] Step 2: When an anomaly occurs during the detection process, directly confirm the standard image from the overall image at the set location. Then, based on the adjacent features of the standard image, confirm the image set to be analyzed from several sets of acquired overall images. Finally, select the features to be analyzed based on the gradual change features of the edge contours of adjacent images within the image set to be analyzed. The specific method is as follows:

[0015] Extract the overall image at the specified location, and extract the contour feature Bz associated with the corresponding overall image. k Where k represents different set positions, and then the confirmed contour feature Bz k Compare with the preset standard feature BJ: If |Bz k If -BJ|≤0.02, the corresponding overall image is marked as a standard image; otherwise, no marking is performed.

[0016] The specific method for verifying the image set to be analyzed is as follows:

[0017] Confirm the set position associated with the standard image, identify whether there are standard images adjacent to the set position, and if so, take the first set of images of the adjacent standard images as the initial image and the second set of images as the final image according to the direction of image acquisition, and confirm a set of "initial image-final image" image columns.

[0018] The several sets of overall images acquired during the detection process are sorted to confirm an image set. Based on the confirmed image series, the initial image, the final image, and other images associated with the initial image and the final image are removed from the image set to obtain the image set to be analyzed.

[0019] The specific method for selecting features to be analyzed from the image set to be analyzed is as follows:

[0020] Based on the confirmed image set to be analyzed, the same confirmation method for the contour features of the images to be processed is adopted to confirm the contour features of different single images in the image set to be analyzed. From the confirmed different contour features belonging to different images, the maximum and minimum values ​​are confirmed. The single images associated with the maximum value and the single images associated with the minimum value are recorded as features to be analyzed.

[0021] Step 3: Based on the different contour features associated with different features to be analyzed, confirm whether the perimeters of the ellipses associated with the two features to be analyzed are consistent. Based on the confirmation results, generate the associated signals for display. The specific method is as follows:

[0022] Based on the different contour features associated with different features to be analyzed, the nearest distance L1 and the farthest distance L2 associated with the corresponding contour features are identified, and the following method is used: Confirm the eccentricity e associated with the feature to be analyzed, and 0 < e < 1;

[0023] use: , where t is the integral variable, and dt is the derivative of the integral variable t;

[0024] The perimeters of the ellipses associated with the two features to be analyzed are confirmed. It is then determined whether the two confirmed sets of ellipse perimeters are consistent. If they are consistent, a size difference signal is directly generated and displayed; if they are inconsistent, a capsule anomaly signal is directly generated.

[0025] This invention provides a method for detecting appearance defects in coenzyme Q10 soft capsules based on image recognition. Compared with existing technologies, it has the following advantages:

[0026] This invention precisely extracts contour features by processing multi-angle images of soft capsules with grayscale and gradient calculation, and combines variance analysis and other methods to keenly identify whether the detection process is abnormal. It effectively avoids the omission of defects caused by single-angle detection or simple visual observation, greatly improves the accuracy of defect detection, and ensures that no potential appearance problems are overlooked.

[0027] When the detection process encounters anomalies, the standard image and the set of images to be analyzed can be quickly identified from a large number of images. Then, the features to be analyzed are selected, and the images with the most significant differences are compared for ellipse perimeter. This method can efficiently narrow down the analysis scope, accurately locate areas that may have wrinkles, damage, or dimensional abnormalities, and improve the targeting and efficiency of defect analysis.

[0028] This method strictly controls the appearance quality of coenzyme Q10 soft capsules, can promptly detect unqualified products, and prevent soft capsules with appearance defects (such as damage leading to content contamination, abnormal size affecting efficacy, etc.) from entering the market, effectively ensuring the safety of consumers' medication, and also helping to maintain the company's product quality reputation and brand image.

[0029] The entire testing process relies on technologies such as image recognition and algorithm calculation, which reduces the subjective errors and labor costs of manual testing, and realizes the automation and intelligence of testing. This meets the needs of modern industrial production for efficient and accurate quality testing, and helps to improve the production testing efficiency and intelligence level of enterprises. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figure 1 This application provides a method for detecting appearance defects in coenzyme Q10 soft capsules based on image recognition, comprising the following steps:

[0033] Step 1: Place the soft capsule at the detection location and use a high-definition camera to acquire an overall image of the soft capsule (this is a circular acquisition process, which can effectively confirm the overall image of the soft capsule). Perform contour detection on the overall image at the set location to confirm the contour features associated with the overall image. Based on the confirmation results, identify whether the detection process is abnormal. If it is abnormal, the subsequent analysis process needs to be executed. If it is normal, no processing is required. This detection process is a sampling detection process.

[0034] The specific method for identifying whether the current detection process is abnormal is as follows:

[0035] The overall image at the set location is extracted and marked as the image to be processed. Specifically, the set location is predetermined by the operator, generally corresponding to the front, left, right, and rear of the capsule. When confirming the location, there is corresponding angle data. The data can be confirmed and marked according to the corresponding angle features. The RGB values ​​associated with different pixels in the image to be processed are confirmed, and the gray values ​​associated with corresponding pixels in the image to be processed are confirmed according to the preset weights. Based on the different gray values ​​associated with different pixels, the image to be processed is converted into a grayscale image. Specifically, the RGB values ​​associated with a single pixel are R, G, and B values, where different values ​​are associated with different weights. The sum of multiple weights is 1, and the grayscale value is 0.299×R+0.587×G+0.114×B. The method of image grayscale processing is common in existing technologies, so it will not be elaborated on here.

[0036] The Sobel algorithm is used to confirm the horizontal and vertical gradients associated with a single pixel in a grayscale image. (When analyzing and confirming a pixel, the Sobel algorithm determines the vertical and vertical gradients associated with the current pixel based on the different grayscale values ​​associated with its eight surrounding pixels. The weighting factor is between -2 and 2, and convolution summation is used to confirm the gradient belonging to the current pixel. Since this method of gradient data confirmation is common in existing technologies, it will not be elaborated upon here.) Based on the confirmed horizontal and vertical gradients, the comprehensive gradient associated with the corresponding pixel is determined. The confirmed comprehensive gradient is compared with the preset value Y1. If the comprehensive gradient is greater than Y1, the corresponding pixel is recorded as the gradient pixel. Otherwise, no marking is performed.

[0037] The gradient pixels associated with the grayscale image are confirmed sequentially, and adjacent gradient pixels are connected to confirm the grayscale contour associated with the grayscale image. Then, the grayscale image is placed in a set of two-dimensional coordinate systems to confirm the two-dimensional coordinates associated with different points within the grayscale contour. The confirmed sets of two-dimensional coordinates are averaged to confirm the mean coordinates. The point where the mean coordinates are located is recorded as the center point of the current grayscale image.

[0038] The two sets of contour points closest to and farthest from the center point are identified on the grayscale contour. The distance between the closest contour point and the center point is denoted as L1, and the distance between the farthest contour point and the center point is denoted as L2. The contour feature Bz of the corresponding grayscale image is confirmed by L2 ÷ L1 = Bz. Then, the contour features associated with different images to be processed are confirmed. The variance of the confirmed contour features Bz is processed to confirm the calibration variance. If the calibration variance > Y2, it means that there is an abnormality in the detection process and an abnormality signal is directly generated. Otherwise, it means that the detection process is normal and no processing is required. Y2 is a preset value, and its specific value is determined by the operator based on experience.

[0039] Specifically, a capsule is placed at a detection location, and images of eight designated locations around it are acquired to confirm the image to be processed. The confirmed image is then converted to grayscale to obtain a corresponding grayscale image. Based on the grayscale values ​​associated with different pixels within the grayscale image, gradient points around the grayscale image are identified and used as the edge contour. The center point of the corresponding grayscale image is then identified based on the associated edge contour. Two sets of points, the closest and farthest from the center point, are identified on the edge contour and denoted as the nearest and farthest points, respectively. The contour features of the capsule are confirmed based on the distance characteristics associated with the nearest and farthest points. The degree of dispersion between the contour feature values ​​is analyzed to determine the specific differences between capsule contours. Large differences indicate significant variations in the shape of the capsule's four sides, indicating a substandard state. Normally, the contour features of the images associated with the four sides of the soft capsule should be relatively consistent and similar. Therefore, by comparing the differences in contour features of images from different directions, it can be confirmed whether there are any abnormalities in the detection process.

[0040] Step 2: When there is an anomaly in the current detection process, the standard image is directly confirmed from the overall image at the set location. Then, based on the adjacent features of the standard image, the set of images to be analyzed is confirmed from several sets of overall images. Then, based on the gradient features of the edge contours of adjacent images in the set of images to be analyzed, the features to be analyzed are selected. Specifically, different set locations are associated with different overall images, and different overall images have different contour features. Based on the preset standard features, the standard image is confirmed from multiple different overall images.

[0041] The specific method for verifying standard images is as follows:

[0042] Extract the overall image at the specified location, and extract the contour feature Bz associated with the corresponding overall image. k Where k represents different set positions, and then the confirmed contour feature Bz k Compare with the preset standard feature BJ: If |Bz k If -BJ|≤0.02, the corresponding overall image will be marked as a standard image; otherwise, no marking will be performed. Specifically, the preset standard feature BJ is the proportional feature associated with the standard size, which is generally 3, that is, the ratio between the length and width inside the capsule is 3.

[0043] The specific method for verifying the image set to be analyzed is as follows:

[0044] Confirm the set position associated with the standard image, identify whether there is a standard image adjacent to the set position, and if so, take the first set of images of the adjacent standard images as the initial image and the second set of images as the final image (that is, a part of the surface of an elliptical facade, which is in a normal state) according to the direction of image acquisition, and confirm a set of "initial image-final image" image series.

[0045] The several sets of overall images acquired during the detection process are sorted to confirm an image set. Based on the confirmed image series, the initial image, the final image, and other images associated with the initial image and the final image are removed from the image set to obtain the image set to be analyzed.

[0046] Specifically, the set positions are in eight directions or four directions, etc. There is a corresponding overall image in the corresponding direction. If there is a standard image, but the set positions associated with the standard image are not adjacent, then the image series cannot be confirmed. If they are adjacent, for example, the overall images in the two directions of "due east position" and "due south position" are both standard images, then the several other sets of images associated between the two directions are also standard images, and no further confirmation is required.

[0047] During the image acquisition process, it is not just about acquiring images at a single set location, but also about acquiring images in a circular pattern. Step one only analyzes the images at the set location, but does not perform an overall analysis because the number of images would be too large during the overall analysis. However, this does not mean that images at different locations are not acquired during the image acquisition process, but rather that images are acquired from all locations. This can be understood as a continuous acquisition process.

[0048] The specific method for selecting the features to be analyzed from the image set to be analyzed is as follows:

[0049] Based on the confirmed image set to be analyzed, the same confirmation method for the contour features of the images to be processed is adopted to confirm the contour features of different single images in the image set to be analyzed. From the confirmed different contour features belonging to different images, the maximum and minimum values ​​are confirmed. The single images associated with the maximum value and the single images associated with the minimum value are recorded as features to be analyzed.

[0050] Specifically, each image has corresponding contour features. Within the corresponding image set, there are minimum contour features and maximum contour features. The minimum contour features and maximum contour features are associated with different images, and the corresponding images are the features to be analyzed. These features to be analyzed will be further analyzed and verified.

[0051] Step 3: Based on the different contour features associated with different features to be analyzed, confirm whether the perimeters of the ellipses associated with the two features to be analyzed are consistent. Based on the confirmation results, generate the associated signals for display.

[0052] The specific method for confirming whether the perimeters of the two feature ellipses to be analyzed are the same is as follows:

[0053] Based on the different contour features associated with different features to be analyzed, the nearest distance L1 and the farthest distance L2 associated with the corresponding contour features are identified, and the following method is used: Confirm the eccentricity e associated with the feature to be analyzed, and 0 < e < 1;

[0054] use: , where t is the integral variable (eccentric angle), which is predetermined by the operator and has a value range of [0, 0.5π], and dt is the derivative of the integral variable t;

[0055] The perimeter of the ellipse associated with the two features to be analyzed is confirmed. It is then identified whether the two confirmed ellipse perimeters are consistent. If they are consistent, a size difference signal is directly generated and displayed. If they are inconsistent, a capsule anomaly signal is directly generated (generally, wrinkles, damage, or other conditions will cause the perimeter of the corresponding different faces to be inconsistent, so the corresponding signal can be directly generated and displayed).

[0056] Specifically, in the actual verification process, no matter how the two surfaces change, as long as there is no damage, their perimeters will be consistent. Therefore, by selecting the two feature surfaces with the most obvious differences from the corresponding image set and comparing and verifying the two feature surfaces, the differences between the corresponding surfaces can be effectively confirmed based on the characteristic changes in the perimeters of the two surfaces. This allows for the assessment of whether the corresponding capsule is damaged or whether there are any abnormalities in size during the manufacturing process. Through actual comparison and verification, the corresponding abnormalities can be effectively and quickly identified.

[0057] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0058] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An image recognition-based coenzyme Q10 soft capsule appearance defect detection method, characterized by, The method comprises the following steps: Step one, place the soft capsule at the detection position, and use a high-definition camera to obtain an overall image of the soft capsule, and perform contour detection on the overall image at the set position, confirm the contour features associated with the overall image, and according to the confirmation result, identify whether the current detection process is abnormal; Step two, when the current detection process is abnormal, directly confirm the standard image from the overall image at the set position, then according to the adjacent features of the standard image, confirm the set of images to be analyzed from the obtained several sets of overall images, and then select the features to be analyzed according to the gradual change features of the edge contours of the adjacent images in the set of images to be analyzed; In the step two, the specific way of confirming the standard image is: extracting the overall image at the set position, and extracting the contour feature Bz belonging to the corresponding overall image k where k represents different set positions, and then comparing the identified contour feature Bz k with the preset standard feature BJ: if |Bz k -BJ|≤0.02, the corresponding overall image is marked as a standard image, otherwise, no marking is performed. In the step two, the specific way of confirming the set of images to be analyzed is: Confirm the set position associated with the standard image, identify whether there is a standard image adjacent to the set position, if there is, then according to the direction of image acquisition, the previous group of images of the adjacent standard image is regarded as the initial image, and the next group of images is regarded as the final image, and an image column of "initial image-final image" is confirmed; Sort the several sets of overall images obtained in the detection process, confirm a set of images, and according to the confirmed image column, remove the initial image, the final image and other images associated between the initial image and the final image from the set of images to obtain the set of images to be analyzed; In the step two, the specific way of selecting the features to be analyzed from the set of images to be analyzed is: According to the confirmed set of images to be analyzed, the same confirmation method of confirming the contour features of the images to be processed is used to confirm the contour features of different single images in the set of images to be analyzed, and from the confirmed different contour features belonging to different images, the maximum value and the minimum value are confirmed, and the single image associated with the maximum value and the single image associated with the minimum value are both recorded as the features to be analyzed; Step three, based on the different contour features associated with different features to be analyzed, confirm whether the circumferences of the two features to be analyzed are consistent, and according to the confirmation result, generate and display the associated signal.

2. The image recognition-based coenzyme Q10 soft capsule appearance defect detection method according to claim 1, characterized in that, In the step one, the specific way of identifying whether the current detection process is abnormal is: Extract the overall image at the set position, and mark it as an image to be processed, confirm the RGB values associated with different pixel points in the image to be processed, and according to the preset weight, confirm the gray values associated with the corresponding pixel points in the image to be processed, and according to the different gray values associated with different pixel points, convert the image to be processed into a gray image; The Sobel algorithm is used to confirm the horizontal gradient and vertical gradient associated with a single pixel point in a gray-scale image, and according to the confirmed horizontal gradient and vertical gradient, the comprehensive gradient associated with the corresponding pixel point is confirmed The confirmed comprehensive gradient is compared with a preset value Y1, if the comprehensive gradient > Y1, the corresponding pixel point is marked as a gradient pixel point, otherwise, no mark is made. Confirm the gradient pixel points associated with the gray image in sequence, connect the adjacent gradient pixel points, confirm the gray contour associated with the gray image, and then place the gray image in a two-dimensional coordinate system to confirm the two-dimensional coordinates associated with different points in the gray contour, and perform mean value processing on the confirmed several sets of two-dimensional coordinates to confirm the mean value coordinates, and mark the point of the mean value coordinates as the center point of the current gray image; And confirm the two groups of contour points closest to and farthest from the center point on the gray scale contour, record the distance between the closest contour point and the center point as L1, record the distance between the farthest contour point and the center point as L2, and confirm the contour feature Bz of the corresponding gray scale image by adopting: L2 ÷ L1 = Bz. 3.The image recognition-based coenzyme Q10 soft capsule appearance defect detection method according to claim 2, characterized in that, Then confirm the contour features associated with different images to be processed, and perform variance processing on the confirmed several groups of contour features Bz to confirm the calibration variance. 4.The image recognition-based coenzyme Q10 soft capsule appearance defect detection method according to claim 1, characterized in that, If the calibration variance > Y2, it represents that there is an abnormality in the current detection process, and a detection abnormality signal is directly generated. According to different profile features associated with different to-be-analyzed features, the nearest distance L1 and the farthest distance L2 associated with the corresponding profile feature are confirmed, and the following is adopted: The eccentricity e associated with the corresponding to-be-analyzed feature is confirmed, and 0 Adopted: where t is an integral variable, where dt is a differential of the integral variable t; If the calibration variance ≤ Y2, it represents that the current detection process is normal, and no processing is required.

5. The image recognition-based coenzyme Q10 soft capsule appearance defect detection method according to claim 4, characterized in that, In step three, the specific way to confirm whether the circumferences of the two features to be analyzed are consistent is: Confirm the circumferences of the two features to be analyzed, identify whether the two confirmed circumferences are consistent, and if so, directly generate a size difference signal for display. If not, a capsule abnormality signal is directly generated.

Images