Character printing defect detection method

Abstract

The invention discloses a character printing defect detection method, and the method comprises the steps: obtaining a local image of a character printing defect or a real-time character printing defect image, importing the character printing defect image into a system, and previewing the effect of the character printing defect image; performing defect detection on the character printing defect image, displaying a character printing defect detection result, performing defect region marking on image defects, and calculating a defect area; and recording a character printing defect detection result, generating a character printing defect detection template, and checking, switching or editing the character printing defect detection template. Compared with the prior art, the method is higher in detection accuracy, supports accurate recognition of character edge defects, quantifiable defects, supports area measurement of character defects, is convenient for customers to judge whether the character defects are qualified or not according to the defect degree, is high in deployment efficiency, does not need to collect a large batch of training samples, and is high in defect detection accuracy. A deformation tolerance mechanism avoids misjudgment caused by printing difference.

Description

Technical Field

[0001] The present disclosure generally relates to the field of image recognition technology, and particularly to a method for detecting character printing defects. Background Art

[0002] The detection of character printing defects is widely used in products such as watch engraving, electronic component identification, label coding, etc., and its quality is directly related to the product appearance and brand reputation. Current detection methods mostly adopt traditional image processing means such as deep learning recognition, OCR technology or template matching.

[0003] Among them, the template matching method is known for its stability and fast calculation speed, and is suitable for scenarios where the character arrangement pattern is clear. In the prior art, however, when the characters are arranged in a circular pattern, such as watch scales or back printing, or there are slight deformations in actual printing, the accuracy of traditional methods will decrease significantly. And deep learning algorithms require a large number of training samples, with high deployment costs, long training cycles, being difficult to interpret, and unable to accurately quantify the defect area, which are mainly manifested as follows.

[0004] 1. Poor adaptability: Traditional template matching cannot adapt to slight rotation, scaling or aspect ratio changes of characters, and often misjudges qualified characters as unqualified;

[0005] 2. Difficulty in dealing with circular characters: Conventional algorithms are difficult to handle characters arranged in a circular pattern in the Cartesian coordinate system, such as circular engraving on watches, and the complexity of extraction and comparison is high;

[0006] 3. Slow new product launch: Traditional algorithms usually require engineers to redesign or develop templates when launching new products, which is time-consuming and laborious;

[0007] 4. Deep learning depends on training data: It requires a large number of defect samples for training, and the results lack interpretability and cannot accurately quantify the defects.

[0008] Therefore, the prior art needs to be improved. Summary of the Invention

[0009] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a method for detecting character printing defects that can meet the specific requirements for character printing defect detection currently.

[0010] Based on one aspect of the embodiments of the present invention, embodiments of the present application provide a method for detecting character printing defects, including:

[0011] Obtaining a local image of a character printing defect, or a real-time character printing defect image collected by an image acquisition device, and importing the character printing defect image into the system to preview the effect of the character printing defect image;

[0012] Perform defect detection on the character printing defect image, display the character printing defect detection result, mark the defect area of the image defect, and calculate the defect area;

[0013] Record the character printing defect detection result, generate a character printing defect detection template, and view, switch, or edit the character printing defect detection template.

[0014] In another embodiment of the present application, the performing defect detection on the character printing defect image includes:

[0015] Perform polar coordinate transformation and image straightening on the character printing defect image;

[0016] The performing polar coordinate transformation and image straightening on the character printing defect image includes:

[0017] When the character printing defect image is arranged in a circular pattern, straighten the character printing defect image into a horizontal character sequence through polar coordinate transformation;

[0018] The straightening the character printing defect image into a horizontal character sequence through polar coordinate transformation includes:

[0019] Obtain the geometric center point of the character printing defect image, and use the geometric center point as the origin of the polar coordinate transformation;

[0020] Perform edge detection and boundary fitting on the geometric center point to improve the alignment degree of the character printing defect image;

[0021] Set the radius interval to be expanded for the polar coordinate transformation, and the radius interval to be expanded for the polar coordinate transformation is the character region range defined in the character printing defect image;

[0022] Take the geometric center point as the pole, sample each polar coordinate point within the radius interval to be expanded for the polar coordinate transformation, and map each polar coordinate point to the original image coordinates of the character printing defect image;

[0023] Perform bilinear interpolation sampling on the original image coordinates of the character printing defect image according to the set polar coordinate grid to obtain the straightened image matrix. The horizontal axis of the straightened image matrix represents the angle, the vertical axis represents the radius direction, and the character content within the radius interval to be expanded for the polar coordinate transformation changes from a curve to a horizontal arrangement.

[0024] In another embodiment of the present application, the performing defect detection on the character printing defect image further includes:

[0025] Perform character detection on the character printing defect image by combining template matching and area analysis;

[0026] The character detection of the character printing defect image by combining the template matching and area analysis includes:

[0027] Preprocess the character printing defect image to obtain a first processed image. The preprocessing includes performing gray-scale processing, filtering processing, edge enhancement processing, and binarization processing on the character printing defect image;

[0028] Extract the character region in the first processed image and compare it with the character printing defect detection template. Evaluate whether there are defects in the character region of the first processed image according to the overlap degree, edge consistency, and area difference;

[0029] If the area deviation of a certain region in the character region of the first processed image exceeds the set threshold, it is determined as a defect region and highlighted.

[0030] In another embodiment of the present application, the defect detection of the character printing defect image further includes:

[0031] Perform fine-tuning matching on the character printing defect image and the character printing defect detection template;

[0032] The fine-tuning matching of the character printing defect image and the character printing defect detection template includes:

[0033] Adjust the image samples in the character printing defect detection template to perform a set proportion adjustment of the set angle and / or scale and / or aspect ratio so that it matches the character printing defect image to be detected.

[0034] In another embodiment of the present application, the generation of the character printing defect detection template includes:

[0035] Automatic template generation. By importing the local image of the character printing defect or the real-time character printing defect image collected by the image acquisition device, automatically analyze the character region and generate a basic template region;

[0036] Manual drawing and editing. Freely draw regions on the interface through the automatically generated template, mark the contours of the characters of interest, and support zooming, rotating, and adjusting the shape.

[0037] In another embodiment of the present application, the gray-scale processing of the character printing defect image includes:

[0038] Obtain the gray-scale image of the character printing defect image and the pixel value of each pixel point in the gray-scale image. The pixel value of each pixel point is the brightness intensity of the pixel point.

[0039] In another embodiment of the present application, the filtering processing of the character printing defect image includes:

[0040] Perform Gaussian filtering on the grayscale processed character printing defect image to obtain a Gaussian processed image. The Gaussian filtering realizes the smoothing of the character printing defect image and retains the edge information of the character printing defect image;

[0041] Perform median filtering on the Gaussian processed image to obtain a median filtered image. The median filtering suppresses salt-and-pepper noise in the Gaussian processed image;

[0042] Perform edge enhancement on the median filtered image to obtain an edge enhanced image. The edge enhancement is used to enhance the character edge contrast in the median filtered image for easy binarization;

[0043] Perform binarization on the edge enhanced image to obtain a first processed image. The binarization is used to convert the edge enhanced image into a black-and-white binary image, enabling high-contrast separation between the character region and the background for easy contour extraction and area analysis.

[0044] In another embodiment of the present application, extracting the character region in the first processed image and comparing it with a character printing defect detection template, and evaluating whether there are defects in the character region of the first processed image according to the overlap degree, edge consistency, and area difference, includes:

[0045] Extract the character region in the first processed image to obtain the character contour of the first processed image;

[0046] Align the obtained character contour of the first processed image with the character detection region of the character printing defect detection template;

[0047] Calculate the intersection area and union area between the character contour of the first processed image and the character detection region of the character printing defect detection template;

[0048] Based on the intersection area and union area, calculate the overlap degree index between the character contour of the first processed image and the character detection region of the character printing defect detection template;

[0049] Based on the relationship between the overlap degree index and a set threshold, determine whether there are defects in the character region of the first processed image.

[0050] In another embodiment of the present application, extracting the character region in the first processed image and comparing it with a character printing defect detection template, and evaluating whether there are defects in the character region of the first processed image according to the overlap degree, edge consistency, and area difference, further includes:

[0051] Extract the character region in the first processed image to obtain the character contour of the first processed image;

[0052] Calculate the filled area of the character contour of the first processed image, and obtain the standard area of the character detection region of the character printing defect detection template;

[0053] Calculate the area difference between the filled area and the standard area;

[0054] Based on the relationship between the area difference and the set threshold, determine whether there are defects in the character region of the first processed image.

[0055] In the embodiments of the present application, by obtaining a local image of a character printing defect, or a real-time character printing defect image collected by an image acquisition device, and importing the character printing defect image into the system to preview the effect of the character printing defect image; performing defect detection on the character printing defect image, and displaying the character printing defect detection result, marking the defect area of the image defect, and calculating the defect area; recording the character printing defect detection result, generating a character printing defect detection template, and viewing, switching, or editing the character printing defect detection template. Compared with the prior art, the detection accuracy of the present application is higher, it supports accurate identification of character edge defects, and the area determination error is less than 5%; the defects can be quantified, it supports area measurement of character defects, and it is convenient for customers to determine whether it is qualified according to the degree of defects; the deployment efficiency is fast, there is no need to collect a large number of training samples, and there is no need for secondary development. It only takes about 10 minutes from importing the image to going online with the detection template; the defect detection accuracy is high, and the deformation tolerance mechanism avoids misjudgment caused by printing differences; the computing resource occupancy is low: no GPU is required, and it can run stably only with an ordinary industrial PC, which is suitable for embedded deployment. Description of the Drawings

[0056] By reading the detailed description of the non-restrictive embodiments with reference to the following drawings, other features, purposes, and advantages of the present application will become more obvious:

[0057] Figure 1 It is a flowchart of a character printing defect detection method provided by an embodiment of the present application. Detailed Embodiments

[0058] The following further details the present application with reference to the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the related invention, rather than limiting the invention. Additionally, it should be noted that for the sake of description, only parts related to the invention are shown in the drawings.

[0059] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The following will detail the present application with reference to the drawings and embodiments.

[0060] In one embodiment, as Figure 1 shown, a method for detecting character printing defects is provided.

[0061] Please refer to Figure 1 , which shows a flowchart of the method for detecting character printing defects to which the embodiments of the present application can be applied.

[0062] As Figure 1 shown, the method for detecting character printing defects includes:

[0063] Step 101, obtain a local image of the character printing defect, or a real-time character printing defect image collected by an image acquisition device, and import the character printing defect image into the system to preview the effect of the character printing defect image;

[0064] Step 102, perform defect detection on the character printing defect image, display the character printing defect detection result, mark the defect area of the image defect, and calculate the defect area;

[0065] Step 103, record the character printing defect detection result, generate a character printing defect detection template, and view, switch, or edit the character printing defect detection template.

[0066] The performing defect detection on the character printing defect image includes:

[0067] Perform polar coordinate transformation and image straightening on the character printing defect image;

[0068] The performing polar coordinate transformation and image straightening on the character printing defect image includes:

[0069] When the character printing defect image is arranged in a circular pattern, straighten the character printing defect image into a horizontal character sequence through polar coordinate transformation;

[0070] The straightening the character printing defect image into a horizontal character sequence through polar coordinate transformation includes:

[0071] Obtain the geometric center point of the character printing defect image, and use the geometric center point as the origin of the polar coordinate transformation;

[0072] Perform edge detection and boundary fitting on the geometric center point to improve the alignment degree of the character printing defect image;

[0073] Set the radius interval to be expanded for the polar coordinate transformation, and the radius interval to be expanded for the polar coordinate transformation is the character region range defined in the character printing defect image;

[0074] Taking the geometric center point as the pole, sample each polar coordinate point within the radius interval to be expanded for the polar coordinate transformation, and map each polar coordinate point to the original image coordinates of the character printing defect image;

[0075] Perform bilinear interpolation sampling on the original image coordinates of the character printing defect image according to the set polar coordinate grid to obtain the straightened image matrix. The horizontal axis of the straightened image matrix represents the angle, and the vertical axis represents the radius direction. The character content within the radius interval to be expanded for the polar coordinate transformation changes from a curve to a horizontal arrangement.

[0076] Specifically, straightening the character printing defect image into a horizontal character sequence through polar coordinate transformation can improve the visibility and consistency of template drawing, reduce the positional interference between characters, improve the recognition speed, and facilitate subsequent defect detection and area comparison. The straightened image has a consistent character arrangement direction, making the template easier to define; operations such as region template matching and area analysis on characters are more concise and accurate; it is especially suitable for scenarios with small character spacing and large circular curvatures;

[0077] When the detection objects are arranged in a circular or semi-circular shape along a certain center point, traditional straight-line template matching methods are difficult to apply directly. This application adopts polar coordinate transformation technology to expand the circular characters into a linear arrangement, facilitating subsequent template production and defect detection.

[0078] The advantages of the above method are: adaptability, capable of handling arbitrarily centrosymmetric character arrangements; intuitive operation, the expanded image is a linear image, which is easy for users to label and debug; improving the detection accuracy, reducing misjudgment and missed judgment caused by character bending; enhancing the template generalization ability, the same template can be adapted to multiple circular character images with different diameters but similar structures.

[0079] The defect detection of the character printing defect image further includes:

[0080] Adopt a combination of template matching and area analysis to perform character detection on the character printing defect image;

[0081] The combination of template matching and area analysis for character detection on the character printing defect image includes:

[0082] Preprocess the character printing defect image to obtain a first processed image. The preprocessing includes performing gray-scale processing, filtering processing, edge enhancement processing, and binarization processing on the character printing defect image;

[0083] Extract the character region in the first processed image and compare it with the character printing defect detection template, and evaluate whether there are defects in the character region of the first processed image according to the overlap degree, edge consistency, and area difference;

[0084] If the area deviation of a certain area in the character region of the first processed image exceeds a set threshold, it is determined as a defective area and highlighted.

[0085] The defect detection of the character printing defect image further includes:

[0086] Performing fine-tuning matching between the character printing defect image and the character printing defect detection template;

[0087] The performing fine-tuning matching between the character printing defect image and the character printing defect detection template includes:

[0088] Adjusting the image samples in the character printing defect detection template to perform a set proportion adjustment of the set angle and / or scale and / or aspect ratio so that it matches the character printing defect image to be detected.

[0089] Specifically, by performing fine-tuning matching between the character printing defect image and the character printing defect detection template, misjudgment caused by slight character deformation can be reduced.

[0090] The generating of the character printing defect detection template includes:

[0091] Automatic template generation, by importing local images of character printing defects or real-time character printing defect images collected by an image acquisition device, automatically analyzing the character region to generate a basic template region;

[0092] Manual drawing and editing, freely drawing regions on the interface through the automatically generated template, marking the outlines of characters of interest, and supporting zooming, rotation, and shape adjustment.

[0093] The performing gray-scale processing on the character printing defect image includes:

[0094] Obtaining the gray-scale image of the character printing defect image and the pixel value of each pixel point in the gray-scale image, and the pixel value of each pixel point is the brightness intensity of the pixel point.

[0095] In another embodiment of the present application, the performing filtering processing on the character printing defect image includes:

[0096] Performing Gaussian filtering processing on the gray-scale processed character printing defect image to obtain a Gaussian processed image, and the Gaussian filtering processing realizes the smoothing processing of the character printing defect image and retains the edge information of the character printing defect image;

[0097] Performing median filtering on the Gaussian processed image to obtain a median filtered image, and the median filtering suppresses salt-and-pepper noise in the Gaussian processed image;

[0098] Perform edge enhancement processing on the median-filtered image to obtain an edge-enhanced processed image. The edge enhancement processing is used to improve the character edge contrast in the median-filtered image, facilitating binarization.

[0099] Perform binarization processing on the edge-enhanced processed image to obtain a first processed image. The binarization processing is used to convert the edge-enhanced processed image into a black-and-white binary image, enabling a high-contrast separation between the character region and the background, facilitating contour extraction and area analysis.

[0100] Specifically, performing binarization processing on the edge-enhanced processed image is to convert the image into a black-and-white binary image, enabling a high-contrast separation between the character region and the background, facilitating contour extraction and area analysis.

[0101] Extract the character region in the first processed image and compare it with the character printing defect detection template. Evaluate whether there are defects in the character region of the first processed image according to the overlap degree, edge consistency, and area difference, including:

[0102] Extract the character region in the first processed image to obtain the character contour of the first processed image.

[0103] Align the obtained character contour of the first processed image with the character detection region of the character printing defect detection template.

[0104] Calculate the intersection area and union area between the character contour of the first processed image and the character detection region of the character printing defect detection template.

[0105] Based on the intersection area and union area, calculate the overlap degree index between the character contour of the first processed image and the character detection region of the character printing defect detection template.

[0106] Based on the relationship between the overlap degree index and the set threshold, determine whether there are defects in the character region of the first processed image.

[0107] The extraction of the character region in the first processed image and the comparison with the character printing defect detection template, and the evaluation of whether there are defects in the character region of the first processed image according to the overlap degree, edge consistency, and area difference, further include:

[0108] Extract the character region in the first processed image to obtain the character contour of the first processed image.

[0109] Calculate the filled area of the character contour of the first processed image and obtain the standard area of the character detection region of the character printing defect detection template.

[0110] Calculate the area difference between the filled area and the standard area.

[0111] According to the relationship between the area difference and the set threshold, determine whether there are defects in the character area of the first processed image.

[0112] Specifically, the system allows users to define personalized templates for each character position, so as to quickly adapt to various product character layouts without coding.

[0113] When finally outputting the detection result, the system will mark the defective characters and highlight them with a red frame on the graph; display the defective area, position and possible reasons, such as breakage, offset, etc.; record the detection log and support exporting the detection result image and report.

[0114] This application converts circular characters into linear arrangements through polar coordinate transformation, greatly simplifying the character comparison process;

[0115] By combining automatic template generation and manual editing, it supports the rapid launch of new products and adapts to different character arrangement structures;

[0116] Through the deformation tolerance mechanism: introducing an adaptive template matching strategy for angle, scale, and aspect ratio to improve the robustness of the system;

[0117] Implement high-speed and accurate comparison, and achieve fast and interpretable defect evaluation through efficient image preprocessing and area quantization.

[0118] The above description is only a preferred embodiment of this application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solution formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the technical solutions formed by mutually replacing the above features with the (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for detecting character printing defects, characterized in that: include: Acquire a local image of a character printing defect, or acquire a real-time character printing defect image through an image acquisition device, and import the character printing defect image into the system to preview the character printing defect image effect; Perform defect detection on character printing defect images and display the character printing defect detection results, mark the defect area of ​​the image defects, and calculate the defect area; Record character printing defect detection results, generate character printing defect detection templates, and view, switch or edit character printing defect detection templates.

2. The character printing defect detection method according to claim 1, characterized in that: The defect detection of the character printing defect image comprises: Carry out polar coordinate transformation and image straightening on character printing defect images; The polar coordinate transformation and image straightening of the character printing defect image comprises: When the character printing defect image is arranged in a circle, the character printing defect image is straightened into a horizontal character sequence by polar coordinate transformation; The method of straightening the character printing defect image into a horizontal character sequence by polar coordinate transformation includes: Obtaining the geometric center point of the character printing defect image, and using the geometric center point as the origin of polar coordinate transformation; Performing edge detection and boundary fitting on the geometric center point to improve the alignment of the character printing defect image; Setting a radius interval that needs to be expanded by polar coordinate transformation, wherein the radius interval that needs to be expanded by polar coordinate transformation is a character area range defined in the character printing defect image; Taking the geometric center point as the pole, sampling each polar coordinate point within the radius interval that needs to be expanded for the polar coordinate transformation, and mapping each polar coordinate point to the original image coordinates of the character printing defect image; Bilinear interpolation sampling is performed on the original image coordinates of the character printing defect image according to the set polar coordinate grid to obtain a straightened image matrix, wherein the horizontal axis of the straightened image matrix represents the angle, and the vertical axis represents the radius direction. The character content within the radius range that needs to be expanded by the polar coordinate transformation changes from a curve to a horizontal arrangement.

3. The character printing defect detection method according to claim 2, characterized in that: The defect detection of the character printing defect image also includes: The character detection of character printing defect images is carried out by combining template matching with area analysis; The method of combining template matching with area analysis to detect characters in a character printing defect image includes: Preprocessing the character printing defect image to obtain a first processed image, wherein the preprocessing includes grayscale processing, filtering processing, edge enhancement processing, and binarization processing on the character printing defect image; Extracting a character region in the first processed image and comparing it with a character printing defect detection template, and evaluating whether the character region in the first processed image has defects based on overlap, edge consistency, and area difference; If the area deviation of a certain area of ​​the character area of ​​the first processed image exceeds a set threshold, it is determined to be a defective area and is highlighted.

4. The character printing defect detection method according to claim 3, characterized in that: The defect detection of the character printing defect image also includes: Fine-tuning and matching the character printing defect image with the character printing defect detection template; The fine-tuning and matching of the character printing defect image with the character printing defect detection template includes: The image sample in the character printing defect detection template is adjusted to set the angle and / or scale and / or the aspect ratio to match the character printing defect image to be detected.

5. The character printing defect detection method according to claim 1, characterized in that: The generating of the character printing defect detection template comprises: Automatic template generation: by importing local images of character printing defects or real-time character printing defect images acquired by image acquisition equipment, the character area is automatically analyzed to generate the basic template area; Manual drawing and editing: freely draw areas on the interface through automatically generated templates, mark the outlines of characters of interest, and support scaling, rotation, and shape adjustment.

6. The character printing defect detection method according to claim 3, characterized in that: The grayscale processing of the character printing defect image includes: A grayscale image of the character printing defect image and a pixel value of each pixel in the grayscale image are obtained, wherein the pixel value of each pixel is the brightness intensity of the pixel.

7. The character printing defect detection method according to claim 3, characterized in that: The filtering process of the character printing defect image comprises: Performing Gaussian filtering on the grayscale processed character printing defect image to obtain a Gaussian processed image, wherein the Gaussian filtering realizes smoothing of the character printing defect image and retains edge information of the character printing defect image; Performing median filtering on the Gaussian processed image to obtain a median filtered image, wherein the median filtering suppresses salt and pepper noise on the Gaussian processed image; Performing edge enhancement processing on the median filtered image to obtain an edge enhanced processed image, wherein the edge enhancement processing is used to enhance the edge contrast of characters in the median filtered image to facilitate binarization; The edge enhanced processed image is binarized to obtain a first processed image, wherein the binarization is used to convert the edge enhanced processed image into a black and white binary image so that the character area is separated from the background by a high contrast, which is convenient for contour extraction and area analysis.

8. The character printing defect detection method according to claim 3, characterized in that: The extracting of the character region in the first processed image and comparing it with the character printing defect detection template, and evaluating whether the character region in the first processed image has defects according to the overlap, edge consistency and area difference, includes: Extracting a character region in the first processed image to obtain a character outline of the first processed image; Aligning the character contour of the acquired first processed image with the character detection area of ​​the character printing defect detection template; Calculating the intersection area and the union area of ​​the character contour of the first processed image and the character detection area of ​​the character printing defect detection template; Calculating an overlap index between the character outline of the first processed image and the character detection area of ​​the character printing defect detection template according to the intersection area and the union area; According to the relationship between the overlap index and the set threshold, it is determined whether there is a defect in the character area of ​​the first processed image.

9. The character printing defect detection method according to claim 8, characterized in that: The extracting of the character region in the first processed image and comparing it with the character printing defect detection template, and evaluating whether the character region in the first processed image has defects according to the overlap, edge consistency and area difference, further includes: Extracting a character region in the first processed image to obtain a character outline of the first processed image; Calculating the filling area of ​​the character outline of the first processed image, and obtaining the standard area of ​​the character detection area of ​​the character printing defect detection template; Calculating the area difference between the filling area and the standard area; According to the relationship between the area difference and the set threshold, it is determined whether there is a defect in the character area of ​​the first processed image.

Images