CN115471684A - Injection molding workpiece template matching method, electronic equipment and storage medium - Google Patents

Abstract

The invention discloses an injection molding workpiece template matching method, electronic equipment and a storage medium, wherein J (psi) _ij ) The arbitration function implements a template matching mechanism. Comprises four main steps: preprocessing image data of an injection molding workpiece; extracting contour features of an image data set and establishing a standard template; obtaining matching precision and determining a matching threshold value and a statistical scaling value thereof for an injection molding workpiece by adopting a Hu moment-based contour matching algorithm; finally using a base based on J (psi) _ij ) And matching the workpiece to be detected with the standard template by a plastic template matching mechanism of the arbitration function. The method replaces the detection based on the gray value in the traditional detection method, so that the template matching of the injection molding workpiece has stability on the influence factors such as translation, rotation and scaling. The invention realizes the automatic detection of the defects of the injection molding workpiece, and has the advantages of high detection speed, and obviously improved accuracy and robustness compared with the traditional method.

Description

Injection molding workpiece template matching method, electronic equipment and storage medium

technical field

The invention relates to the field of defect detection of injection molded workpieces, in particular to a template matching method for injection molded workpieces, electronic equipment and a storage medium with translation, rotation and scaling stability of the workpiece image to be tested.

Background technique

In areas with a high degree of automation, injection molding production lines have basically achieved full automation. However, the quality inspection of injection molded workpieces still mainly uses manual visual inspection to detect surface defects of plastic parts. This open-loop working method is highly subjective and has weak real-time monitoring capabilities. Due to the different standards of manual inspection, the inspection standards are unstable. At the same time, from the perspective of the working environment, front-line workers also face dangers during the work process, and the labor intensity is relatively high.

At present, image recognition is also used to detect the surface of injection molded workpieces. However, under the influence of translation, rotation, zooming and other distortions in the actual shooting process of the plastic part image, the proportion of false alarms is large and the accuracy rate is low, and the actual application effect is not ideal. , can only be used in a few stations that are prone to consistent surface defects, and the applicability is very narrow.

Contents of the invention

The technical problem to be solved by the present invention is to provide a template matching method for injection molded workpieces with multivariable stability, electronic equipment and storage media. It can judge whether it is qualified or not, so as to solve the problems of strong subjectivity, low accuracy, poor robustness, narrow application range and high labor cost of the existing technical testing standards, and realize the automation of the quality testing process of plastic parts.

In order to solve the above technical problems, the present invention adopts the following technical solutions:

A method for matching templates of injection molded workpieces with multivariate stability, characterized in that it comprises the following steps:

S1: Collect the image of the injection molding workpiece and preprocess the collected image data of the injection molding workpiece;

S2: Contour feature extraction of image dataset and establishment of standard template data;

S3: The contour matching algorithm based on the Hu moment is used to obtain the matching accuracy for the injection molded workpiece;

S4: Use the J(ψ _ij ) arbitration function-based plastic part template matching mechanism to match the workpiece to be tested with the standard template.

Further, the preprocessing of the image data of the injection molded workpiece in step S1 includes the following steps: performing grayscale transformation processing on the original image, performing median filtering on the grayscale image, and performing sharpening processing on the smoothed image.

Further, the contour feature extraction of the image data set in step S2 uses the Canny edge detection algorithm. First, the Gaussian first-order derivative function filter is used to preprocess the image, and then the direction and magnitude of the gradient are calculated. Based on the non-maximization suppression theory Determine the more accurate edge information of the image, and set the corresponding threshold according to the actual image to obtain the best contour features, thereby establishing an image database of standard templates for various defect images.

Further, in step S2, the high threshold gray value is set to 240, and the low threshold gray value is set to 200.

Further, in step S3, the contour matching algorithm based on the Hu moment is used for the injection molded workpiece to obtain the matching accuracy, which includes the following steps: using the Hu moment function to establish the invariant moment data of the template image and the defect image, and combining the invariant moment data of the defect image with the The invariant moment data of the template image is matched to obtain the matching result, and the value of the matching result is used as a measure of the similarity of the geometric contour.

Further, the Hu moment function in step S3 is seven invariant moments composed of the second-order and third-order central moments of the image.

Further, in step S3, the following formula is used for matching between the invariant moment data of the defect image and the invariant moment data of the template image:

表示图片编号为i的标准模板的第k个Hu矩数据，

表示图片编号为i的待测图像的第k个Hu矩数据。Among them, _ψij represents the matching result, which is recorded as the matching accuracy, and its numerical value and similarity show an anti-correlation trend; i is used to record the number of the workpiece to be tested, j represents the number of matching times of the same workpiece image input, and k represents the processed image The order of the Hu moments from one to seven, A represents the standard template, and B represents the previous images to be tested;

Indicates the kth Hu moment data of the standard template with picture number i,

Indicates the kth Hu moment data of the image to be tested with the image number i.

Further, in step S4, use the J(ψ _ij ) arbitration function-based plastic part template matching mechanism to match the workpiece to be tested with the standard template according to the following arbitration function formula:

In the arbitration function J(ψ _ij ), i represents the number of the workpiece to be tested, and j represents the number of matching times of the same workpiece image input is also the number of times the same workpiece is input into the arbitration function, usually once or twice; hu _A [k] represents the standard The Hu invariant moment data of the template image, k represents the order of the Hu moment of the processed image from one to seven; Ω is the lower limit threshold of the accuracy of the qualified workpiece;

If the value of the arbitration function J(ψ _ij ) is 1, that is, the matching accuracy is lower than the threshold Ω, the workpiece represented by the image is a qualified product;

If the arbitration function value is 0, that is, the matching accuracy is higher than the threshold Ω, but lower than its statistical scaling value, the image will be preprocessed twice, and the feature information will be further extracted before matching. If the second matching accuracy is still If it is higher than the threshold, it is classified into the defect image data set;

If the function value is -1, that is, the statistical scaling value of the matching accuracy Ω is higher than the threshold, the artifact is directly classified as a defect image dataset.

An electronic device, comprising: a memory, a processor, and a computer program on the memory that can run on the processor, wherein the processor implements the steps of the above method when executing the computer program.

A transient or non-transitory computer-readable storage medium, on which a computer program is stored, is characterized in that, when the computer program is executed, the steps of the above method are realized.

Thus, the present invention discloses a new template matching method for injection molded workpieces with multi-variable stability, electronic equipment and storage media,

Compared with the prior art, the beneficial effects brought by the technical solution of the present invention are:

It replaces the template matching method based on the gray value, making the matching process robust to factors such as translation, rotation, and scaling.

Through the J(ψ _ij ) arbitration function in the present invention, secondary matching is performed on some images that may have misclassifications, and the new matching method that increases the error tolerance rate is conducive to enhancing the accuracy of defect classification of injection molding workpieces.

The system and electronic equipment of the present invention can realize the automatic detection of injection molded workpieces, improve the detection quality and ensure the detection consistency.

Description of drawings

The present invention will be further described below in conjunction with accompanying drawing and embodiment, in the accompanying drawing:

Fig. 1 is a flow chart of the template matching method for injection molded workpieces with multivariable stability of the present invention.

Fig. 2 is a comparison diagram of the original image (a) of the injection molded workpiece of the present invention and the image (b) after pretreatment.

Fig. 3 is the contrast diagram of high threshold (a) and low threshold (b) of Canny edge detection algorithm in the embodiment, wherein the two thresholds of (a) figure are 240,200 respectively; The two thresholds of (b) figure are 40 respectively , 20.

Fig. 4 is the feature extraction result of the present invention.

Fig. 5 is a specific flow chart of the template matching method proposed by the present invention.

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

The template matching method for injection molded workpieces with multivariable stability implemented according to the present invention is shown in FIGS. 1 and 5 . The input end is the collected image of the plastic part, and the output end is the judgment result of whether the image is qualified or not. According to the following embodiment, the judgment is realized based on the template matching result:

(1) Preprocessing of the image data of the injection molding workpiece: the original image is processed by grayscale transformation, the grayscale image is processed by median filtering, and the smoothed image is processed by image sharpening. The function of the gray scale conversion processing in this step is to improve the recognition degree of the injection molded workpiece, and the function of the smoothing processing is to eliminate image noise. The sharpening process is used to eliminate the blur caused by the smoothing process. The preprocessing results of the image data of the injection molding workpiece are shown in Figure 2.

(2) Contour feature extraction of image data set and establishment of standard template: the contour feature extraction of image data set uses Canny edge detection algorithm, and the corresponding threshold is set according to the actual image to obtain the best contour feature. The Canny edge detection algorithm in this embodiment adopts a high threshold as shown in Fig. 3 .

In this embodiment, contour feature extraction is performed on 8 types of defects that are most likely to occur in plastic parts in industrial production. 8 types of defects: warping of plastic parts, black spots on the surface, unstable dimensions, shrinkage depressions, silver streaks, and flashing , poor filling, cracking. The feature extraction results are shown in Figure 4.

(3) The contour matching algorithm based on the Hu moment is used for the injection molding workpiece to obtain the matching accuracy: the invariant moment data of the template image and the defect image are established by using the Hu moment function (equation 3 below), and the invariant moment data of the defect image are compared with the template The invariant moment data of the image is matched to obtain the matching result, and the value of the matching result is used as a measure of the similarity of the geometric contour. The Hu moment function is the seven invariant moments composed of the second-order and third-order central moments of the image.

The HU moment data of the template image set up described in this embodiment are as follows:

Hu1[1]=2.02425; Hu1[2]=4.46429; Hu1[3]=8.01779; Hu1[4]=8.98081; Hu1[5]=18.0195; Hu1[6]=-11.5543; Hu1[7]=17.499.

In this embodiment, the matching accuracy of the function to the standard workpiece is between 0.0871 and 0.1517. The lower limit introduced into this matching precision interval is denoted as Ω, which is used to establish the J(ψ _ij ) arbitration function, and Ω=0.15 in this embodiment.

Match the designed standard templates with the 8 types of defect images, and output the matching results as follows.

(4) Use the J(ψ _ij ) arbitration function-based plastic part template matching mechanism to match the workpiece to be tested with the standard template: combined with the data of the standard template, the statistical scaling value of the threshold Ω in this example is 0.2542. Except for defects such as unstable dimensions of plastic parts that will enter the secondary pretreatment process, the other seven types of defects were successfully detected. The images with unstable size are also accurately classified after the second preprocessing.

After the new method proposed by the present invention is visualized, it is shown in FIG. 5 .

The electronic equipment in the embodiment of the present invention includes: a memory, a processor, and a computer program on the memory that can run on the processor, wherein the processor implements the steps of the method described above when executing the computer program . The electronic equipment can be arranged on the corresponding workstation, and realized in the form of a camera and a memory processor.

Of course, the present invention can also directly be a transitory or non-transitory computer-readable storage medium on which a computer program is stored, and it is characterized in that, when the computer program is executed, the steps of the above method are realized.

It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should belong to the protection scope of the appended claims of the present invention.

Claims (10)

1. A method for matching injection molding workpiece templates with multivariable stability is characterized by comprising the following steps:

s1: collecting an injection molding workpiece image and preprocessing the collected injection molding workpiece image data;

s2: extracting contour features of an image data set and establishing standard template data;

s3: obtaining the matching precision of the injection molding workpiece by adopting a Hu moment-based contour matching algorithm;

s4: using a base based on J (psi) _ij ) And matching the workpiece to be detected with the standard template by a plastic template matching mechanism of the arbitration function.

2. The method for matching templates of injection-molded workpieces with multivariate stability as claimed in claim 1, wherein the pre-processing of image data of injection-molded workpieces in step S1 comprises the steps of: and carrying out gray level transformation processing on the original image, carrying out median filtering on the gray level image, and carrying out sharpening processing on the smoothed image.

3. The method for matching the templates of injection-molded workpieces with multivariate stability as claimed in claim 1, wherein the profile features of various standard images of the image dataset are extracted in step S2, a classical Canny edge detection algorithm is used, the images are preprocessed by a first derivative function filter of gaussian first, then the direction and amplitude of the gradient are calculated, more accurate edge information of the images is determined based on the non-maximization inhibition theory, and the optimal profile features are obtained by setting corresponding thresholds according to the actual images, thereby establishing an image database of the standard templates of various defective images.

4. The method as claimed in claim 1, wherein the high threshold gray-level value is set to 240 and the low threshold gray-level value is set to 200 in step S2.

5. The injection molding workpiece template matching method with multivariable stability as claimed in claim 1, wherein the step S3 of obtaining the matching precision by applying the profile matching algorithm based on the Hu moment to the injection molding workpiece comprises the following steps: and establishing invariant moment data of the template image and the defect image by using the Hu moment function, matching the invariant moment data of the defect image with the invariant moment data of the template image to obtain a matching result, and taking the value of the matching result as the measurement of the similarity of the geometric profile.

6. The method for matching injection molding workpiece templates with multivariate stability as claimed in claim 1, wherein the Hu moment function in step S3 is seven invariant moments formed by combining the second and third central moments of the image.

7. The method for matching templates of injection-molded workpieces with multivariate stability as claimed in claim 1, wherein in step S3, the invariant moment data of the defect image and the invariant moment data of the template image are matched using the following formula:

wherein psi _ij Expressing a matching result, recording the matching result as matching precision, wherein the numerical value and the similarity degree of the matching result show an inverse correlation trend; i is used for recording the number of a workpiece to be detected, j represents the matching times of the same workpiece image, k represents the sequence of Hu moments of the processed image from one to seven, A represents a standard template, and B represents the image to be detected of previous times;

the kth Hu moment data representing the standard template of picture number i,

and k-th Hu moment data of the image to be measured with the picture number i.

8. The method of claim 1, wherein J (ψ) is used in step S4 _ij ) The arbitration function model matching mechanism matches the workpiece to be tested with the standard model according to the following arbitration function J (psi) _ij ) Matching by using a formula:

wherein i represents the number of the workpiece to be detected, j represents the number of times that the same workpiece image is input for matching, and is also the number of times that the same workpiece is input into an arbitration function, and generally the number of times is one or two; hu (pharmaceutical) _A [k]Hu invariant moment data representing a standard template image, k representing the sequence of Hu moments of a processed image from one to seven; omega is the lower limit threshold of the precision of the qualified workpiece;

if the arbitration function J (psi) _ij ) If the value is 1, namely the matching precision is lower than the threshold value omega, the workpiece represented by the image is a qualified product;

if the arbitration function value is 0, namely the matching precision is higher than the threshold omega but lower than the statistical scaling value, performing secondary preprocessing on the image, further extracting characteristic information and then matching, and classifying the image into a defect image data set if the matching precision of the second time is still higher than the threshold;

and if the function value is-1, namely the statistical scaling value of the matching precision omega is higher than the threshold value, directly classifying the workpiece into a defect image data set.

9. An electronic device, comprising: memory, processor and computer program on the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any of the claims 1 to 8 when executing the computer program.

10. A transitory or non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program is configured to perform the steps of the method according to any one of claims 1 to 8 when executed.

Images