Vision-based plastic mold injection molding defect analysis method, device and system

Abstract

The invention belongs to the technical field of mold manufacturing defect detection, and particularly relates to a vision-based plastic mold injection molding defect analysis method, device and system. The method comprises the following steps: configuring a set lighting point; activating an illumination source and focusing and illuminating an illumination point to generate light reflection; capturing light reflection by using camera equipment, and collecting optical image data; converting the optical image data into digital signals and storing the digital signals; integrating the converted digital signals into image matrix data; extracting feature data of optical feature line continuity in the image from the image matrix data; generating a defect graph by comparing the change of the optical characteristic line, and identifying the structure defect on the surface of the injection molding part; converting the image feature data to a frequency domain, and removing noise by using a filtering technology; performing inverse frequency domain transformation on the filtered data, and reconstructing a defect graph; and determining and recording the defect position of the injection molded part. The method improves the accuracy and efficiency of the whole defect detection process.

Description

Technical Field

[0001] The present invention belongs to the technical field of mold manufacturing defect detection, and particularly relates to a vision-based injection defect analysis method, device and system for plastic molds. Background Art

[0002] In modern manufacturing, plastic injection molding is a widely used production technology for quickly and cost-effectively producing large quantities of plastic parts. Quality control of plastic mold injection products is particularly important because the presence of defects directly affects the performance and aesthetics of the products. Common defect types include warping, bubbling, burning, shrinkage, etc., which are usually caused by improper mold design, wrong material selection or inappropriate processing parameters.

[0003] Traditional defect detection methods mainly rely on manual visual inspection, which is not only time-consuming and laborious, but also has poor accuracy and repeatability of detection results. Although some automated detection systems have been developed, such as those based on simple optics or infrared imaging, they can often only detect surface or larger defects and are difficult to identify minor or internal structural problems.

[0004] To solve the above problems, in recent years, automated defect detection systems based on advanced image processing technology have become a research hotspot. These systems use high-resolution cameras to capture images of injection-molded parts and analyze the image data through complex algorithms to identify various types of defects. However, existing technologies still have some limitations, such as slow processing speed, high dependence on ambient light, and difficulty in processing injection-molded parts with complex or irregular shapes. Summary of the Invention

[0005] (1) Technical Problems to be Solved

[0006] The present invention mainly aims at the above problems and proposes a vision-based injection defect analysis method, device and system for plastic molds, aiming to solve the problems of low efficiency, low accuracy and high dependence in traditional defect detection methods and achieve fast and accurate defect identification and analysis.

[0007] (2) Technical Solutions

[0008] To achieve the above object, the first aspect of the present invention provides a vision-based injection defect analysis method for plastic molds, including the following steps:

[0009] Configure a specific area on the surface of the injection-molded part to set lighting points;

[0010] Activate the light source and focus the light on the lighting points to generate light reflection;

[0011] Use a camera device to capture the light reflection from multiple different angles and collect the optical image data of the injection-molded part;

[0012] Convert the optical image data into a digital signal and store it in a data storage device;

[0013] Integrate the converted digital signals into a set of image matrix data;

[0014] Extract the characteristic data of the continuity of the optical feature lines in the image from the image matrix data;

[0015] Generate a defect map by comparing the changes in the optical feature lines to identify the structural defects on the surface of the injection molded part;

[0016] Convert the image feature data to the frequency domain and use filtering techniques to remove noise;

[0017] Perform an inverse frequency domain transform on the filtered data to reconstruct the defect map;

[0018] Determine and record the defect locations of the injection molded part.

[0019] Further, the steps of setting the lighting points include:

[0020] Preset multiple potential lighting areas on the surface of the injection molded part;

[0021] Use a photosensitive sensor to monitor the reflectivity of each potential lighting area to evaluate the lighting effect of each area;

[0022] Automatically select the area with the highest reflectivity as the lighting point according to the reflectivity data.

[0023] Further, the steps of capturing the light reflection from multiple different angles using a camera device include: irradiating the same position of the injection molded part with different lighting intensities and / or different lighting angles each time, so that the injection molded part generates light reflection and is received by the camera device.

[0024] Further, the method of converting the optical image data into a digital signal includes:

[0025] Receive the optical image data captured by the camera device;

[0026] Input the received optical image data into an image processing unit;

[0027] In the image processing unit, perform grayscale processing to convert the color image data into grayscale image data;

[0028] Apply an image coding algorithm to encode the grayscale image data into a digital signal.

[0029] Further, the method of extracting the characteristic data of the continuity of the optical feature lines in the image from the image matrix data includes the following steps:

[0030] Perform gradient calculation on the image matrix data;

[0031] Record the gradient direction and gradient magnitude of each pixel point;

[0032] Convert the gradient direction and gradient magnitude into polar coordinate representation;

[0033] Perform curve fitting on each point in the polar coordinate system to obtain a preliminary feature line;

[0034] Calculate the continuity of the fitted curve and mark the discontinuous points;

[0035] Delete or correct the data corresponding to the discontinuous points;

[0036] Update the fitted curve to obtain the final optical feature line.

[0037] Furthermore, the formula for performing gradient calculation on the image matrix data is as follows:

[0038]

[0039] where A is the image matrix; G x and G y are the gradient matrices in the horizontal and vertical directions respectively.

[0040] Furthermore, the formula for the gradient magnitude is:

[0041]

[0042] The formula for the gradient direction is:

[0043]

[0044] Furthermore, the method for generating a defect map by comparing the changes in the optical feature line and identifying the structural defects on the surface of the injection molded part includes the following steps:

[0045] Perform segmentation processing on the extracted final optical feature line and calculate the curvature and direction change of each segment of the feature line;

[0046] Register the feature lines at different time points or different angles, compare the feature line shapes at corresponding positions, and identify the variant points;

[0047] Construct a change matrix from the variant points to the whole feature line, and match the change matrix with a pre-set defect template;

[0048] Mark the matching area and generate a preliminary defect map;

[0049] Apply an edge detection algorithm to the marked area in the preliminary defect map and extract the connected components in the edge detection result;

[0050] Calculate the area, perimeter, and other geometric properties of each connected component, screen the effective defect regions based on geometric property thresholds, confirm the screened defect regions, and record their positions;

[0051] Generate the final defect map and save it to the data storage device.

[0052] To achieve the above object, a second aspect of the present invention provides a vision-based injection molding defect analysis device for plastic molds, including:

[0053] An illumination component for configuring a specific area on the surface of the injection molded part to set illumination points and activating an illumination source to focus on illuminating the illumination points to generate light reflection;

[0054] A camera component for capturing light reflections from multiple different angles and collecting optical image data of the injection molded part;

[0055] An image processing unit for converting the optical image data into digital signals and integrating them into a set of image matrix data;

[0056] A feature extraction unit for extracting feature data on the continuity of optical feature lines in the image from the image matrix data;

[0057] A defect recognition unit for generating a defect map by comparing the changes in optical feature lines and identifying the structural defects on the surface of the injection molded part;

[0058] A frequency domain conversion unit for converting the image feature data into the frequency domain, using filtering techniques to remove noise, and performing an inverse frequency domain transformation on the filtered data to reconstruct the defect map;

[0059] A storage unit for determining and recording the defect positions of the injection molded part.

[0060] To achieve the above object, a third aspect of the present invention provides a vision-based injection molding defect analysis system for plastic molds, including:

[0061] A vision-based injection molding defect analysis device for plastic molds;

[0062] A control module for coordinating and controlling the work processes of each component;

[0063] A display module for displaying the defect map and related analysis results;

[0064] A database module for storing historical defect data and defect templates;

[0065] A communication module for performing data exchange with an external system;

[0066] Wherein: the control module includes:

[0067] The main controller is used to execute the automated operation of the defect analysis process;

[0068] The parameter setting unit is used to set and adjust various parameters of the analysis system;

[0069] The database module includes:

[0070] The defect record library is used to store the defect location information and image data of each analysis;

[0071] The defect template library is used to store pre-set defect templates and identify defect types by matching defect templates.

[0072] (III) Beneficial effects

[0073] Compared with the prior art, a vision-based injection defect analysis method, device and system for plastic molds provided by the present invention adopt multi-angle and multi-intensity lighting technologies in cooperation with advanced image processing algorithms, significantly improving the accuracy and efficiency of defect detection. By setting specific lighting points and capturing light reflections from multiple angles, this system can generate more detailed optical image data. In addition, by using the gradient calculation and frequency domain analysis of image matrix data, the system can effectively extract defect features from complex backgrounds, and through automated feature line continuity detection and defect template matching technologies, quickly and accurately locate and classify structural defects. This method not only reduces the dependence on operator skills but also greatly shortens the detection cycle.

[0074] Specifically, in the present invention, the system first uses the gradient calculation of image matrix data to analyze the change degree of each pixel point. By calculating the gradient amplitude and direction of each point in the image, the object edges and other important features can be effectively detected. Then, through frequency domain analysis, the system converts the image from the time domain to the frequency domain, which enables more effective removal of noise through filtering techniques and highlights useful feature information. In addition, the system also adopts feature line continuity detection to determine the location of defects by calculating the continuity of image feature lines. The defect template matching technology matches the detected features with pre-set defect templates to quickly and accurately identify and classify specific structural defects. Brief description of the drawings

[0075] Figure 1 It is a schematic flowchart of a vision-based injection defect analysis method for plastic molds disclosed in this application.

[0076] Figure 2 It is a schematic working diagram of an illumination component disclosed in this application.

[0077] Figure 3 It is an arrangement diagram of a multi-angle camera device disclosed in this application.

[0078] Figure 4 A comparison chart of images before and after filtering disclosed in this application.

[0079] Figure 5 A schematic diagram of defect identification disclosed in this application.

[0080] Figure 6 A schematic diagram of optical feature line extraction disclosed in this application.

[0081] Figure 7 An overall system architecture diagram disclosed in this application.

[0082] Figure 8 An example diagram of plastic gear crack detection disclosed in this application.

[0083] Figure 9 A database structure diagram disclosed in this application.

[0084] Figure 10 A function diagram of the control module disclosed in this application. Detailed implementation manners

[0085] The present invention will be described in detail below with reference to the accompanying drawings. The technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0086] This embodiment provides a vision-based injection molding defect analysis method for plastic molds, as Figure 1 shown. The method includes the following steps:

[0087] Step S100: Configure specific areas on the surface of the injection molded part to set illumination points;

[0088] First, select one or more specific areas on the surface of the injection molded part as illumination points. These areas can be selected in typical or defect-prone parts of the injection molded part. Use a high-precision position adjustment device to accurately set the positions of the illumination points to ensure that the light source can evenly and accurately illuminate these areas.

[0089] Step S200: Activate the illumination source and focus the illumination on the illumination points to generate light reflection;

[0090] Use a light source control system to activate the illumination source, such as an LED or a laser, and ensure that the light is accurately focused on the previously set illumination points through an optical focusing system (such as a lens or a mirror).

[0091] Step S300: Use a camera device to capture light reflections from multiple different angles and collect optical image data of the injection molded part;

[0092] Deploy high-resolution camera devices at multiple angles to capture light reflected from different angles. These camera devices can operate synchronously to collect optical image data in real time.

[0093] Step S400: Convert the optical image data into digital signals and store them in a data storage device

[0094] Use an image processing unit to convert the captured optical image data into digital signals. This conversion includes image correction and enhancement processing to improve image quality. The converted digital signals need to be stored in a high-speed and secure data storage device, such as an SSD hard drive.

[0095] Step S500: Integrate the converted digital signals into a set of image matrix data;

[0096] Use image processing software to integrate digital signals from multiple angles to generate a complete set of image matrix data. In this step, algorithms are applied to register and stitch the image data to ensure the continuity and integrity of the images.

[0097] Step S600: Extract feature data on the continuity of optical feature lines in the image from the image matrix data;

[0098] Through advanced image analysis software, extract key optical feature lines from the image matrix, such as the edges and reflection intensities of light rays. Analyze the continuity of these feature lines to identify potential defects.

[0099] It should be noted that: Optical feature lines can reveal microscopic changes and irregularities on the surface of an object. When light shines on an object, it will be evenly reflected on a smooth and defect-free surface, while at a defective place such as a crack, depression or protrusion, the reflection path of the light will change. This change will form interrupted or distorted optical feature lines in the optical image. By analyzing the patterns and continuity of these feature lines, the structural defects on the surface of the object can be accurately identified and located, because the differences in optical properties between a normal surface and an abnormal surface result in clearly visible changes in the feature lines in the image.

[0100] Step S700: Generate a defect map by comparing the changes in optical feature lines and identify the structural defects on the surface of the injection molded part;

[0101] Step S800: Convert the image feature data into the frequency domain and use filtering techniques to remove noise;

[0102] To further improve the image quality, the image feature data is transformed into the frequency domain, and various filtering techniques (such as low-pass or high-pass filters) are applied to remove image noise (see Figure 4 ).

[0103] Step S900: Perform an inverse frequency domain transform on the filtered data to reconstruct the defect map;

[0104] After the filtering process is completed, an inverse transform is performed on the data to return from the frequency domain to the spatial domain to reconstruct a clear defect map.

[0105] S1000: Determine and record the defect locations of the injection-molded parts

[0106] Finally, based on the reconstructed defect map, the specific defect locations on the injection-molded parts are determined and marked for recording. This information will be used for quality control and subsequent production improvement.

[0107] As Figure 2 shown, step S100 involves the process of setting illumination points, which specifically includes the following sub-steps: First, multiple potential illumination areas are preset on the surface of the injection-molded part, and these areas are selected according to the structural characteristics of the injection-molded part and the expected detection area. Then, a photosensitive sensor is used to monitor the reflectivity of these potential illumination areas in real time, and the illumination effect of each area is evaluated by analyzing the data collected by the sensor. Finally, the system will automatically select the area with the highest reflectivity as the final illumination point. This method ensures that the selection of the illumination point can maximize the improvement of the image quality and detection effect, thereby improving the accuracy and efficiency of the entire defect detection process.

[0108] As Figure 3 shown, step S300 involves using a camera device to capture light reflections from multiple different angles. The specific operation is to irradiate the same position of the injection-molded part multiple times, and each irradiation uses a different illumination intensity and / or a different illumination angle. By changing the illumination conditions, different light reflection effects are induced on the surface of the injection-molded part, and these effects are subsequently captured by the camera device. Through this method, the optical characteristics of this position under different illumination conditions can be obtained, providing richer data for subsequent image analysis.

[0109] In a specific embodiment of the present invention, step S400 involves the process of converting optical image data into digital signals. First, the captured optical image data is received by a camera device. These optical image data are then input into an image processing unit. In the image processing unit, grayscale processing is first performed, that is, converting color image data into grayscale image data. Grayscale processing helps to simplify subsequent image processing steps because grayscale images only contain luminance information, thereby reducing the amount of data while highlighting structural features. After that, an image coding algorithm is applied to encode the processed grayscale image data and convert it into a digital signal.

[0110] As Figure 6 shown, step S600 involves extracting characteristic data of the optical feature line continuity from the image matrix data, specifically including:

[0111] Step S601: Calculate the gradient of the image matrix data;

[0112] Step S602: Record the gradient direction and gradient magnitude of each pixel point;

[0113] Step S603: Convert the gradient direction and gradient magnitude into polar coordinate representation;

[0114] Step S604: Perform curve fitting on each point in the polar coordinate system to obtain a preliminary feature line;

[0115] Step S605: Calculate the continuity of the fitted curve and mark the discontinuous points;

[0116] Step S606: Delete or correct the data corresponding to the discontinuous points;

[0117] Step S607: Update the fitted curve to obtain the final optical feature line.

[0118] By calculating the image gradient and converting this data into polar coordinate representation, curve fitting can be performed more naturally and accurately to identify continuous and discontinuous feature lines in the image. Gradient calculation can reveal the rate of change of luminance for each pixel in the image, and can determine where there are potential boundaries or feature lines, where the gradient direction and magnitude determine the orientation and intensity of the edges. Conversion to polar coordinate representation simplifies the processing of angles and distances, making it more intuitive and effective to extract feature lines from image data. Using polar coordinate data, curve fitting can be performed on the feature lines to combine scattered points into continuous lines. Continuity analysis helps to identify possible breaks or anomalies in the fitted curve.

[0119] Gradient calculation is a basic method for edge detection in image processing. By calculating the gradient of the image matrix, the places with the most significant luminance changes in the image can be determined, and the formula is as follows:

[0120]

[0121] Among them, A is the image matrix, and G x and G y are the gradient matrices in the horizontal and vertical directions respectively.

[0122] After obtaining the gradient matrices, by calculating the gradient magnitude and direction of each pixel point, it is used to identify the intensity and direction of the edge. The formula for the gradient magnitude is as follows:

[0123]

[0124] The gradient magnitude G is the Euclidean distance between G x and G y , representing the edge intensity at this point.

[0125] The calculation formula for the gradient direction is:

[0126]

[0127] The gradient direction Θ is the edge direction at this point, obtained from the arctangent of the ratio of G y and G x , representing the direction of the edge.

[0128] Convert the Cartesian coordinates (G x , G y ) of the gradient to polar coordinates (G, Θ). Using the gradient information (magnitude and direction) extracted from the image, the identified edges can be curve-fitted to form a more coherent and smooth edge representation. By checking the continuity of the fitted curve, the interrupted or discontinuous parts can be identified and marked. These discontinuities are caused by image noise or defects; in the continuity check method, the discontinuous points are found by analyzing the derivative of the curve or performing differential calculations; data smoothing or interpolation processing is performed on these discontinuous points to restore the continuity of the feature line. Through this series of steps, important features such as edges and lines in the image are effectively extracted.

[0129] See Figure 5, in step S700, the method of generating a defect map by comparing the changes in optical feature lines at different times or different angles and identifying structural defects on the surface of an injection-molded part includes the following steps: First, process the extracted optical feature lines in segments, and calculate the curvature and direction changes of each segment; then, register the feature lines, compare the morphologies at corresponding positions to identify variant points; next, construct a matrix describing the changes from the variant points to the entire feature line, and match it with a pre-set defect template; mark and generate a preliminary defect map within the matching area; then, apply an edge detection algorithm (such as the Canny algorithm) to the preliminary defect map, extract connected components; screen out effective defect areas by calculating geometric properties (such as the minimum of area or perimeter) and record their positions; finally, generate a defect map and save it to a data storage device.

[0130] Suppose we want to detect whether there are tiny cracks on the surface of a plastic gear:

[0131] As Figure 8 shown, first, use an optical scanner to scan the plastic gear to obtain the optical feature lines on its surface. These feature lines reflect the changes in the gear surface morphology. Divide these feature lines into several segments every 10 millimeters; for each segment, calculate the curvature and direction changes. For example, on a certain segment, if the feature line suddenly has a sharp bend, it indicates that there is an anomaly at that position, that is, a potential crack.

[0132] Next, scan the gear again from different time points or different angles, and register these feature lines to make them aligned; after registration, compare the feature lines at the same positions in each scan. If the morphology of the feature line at a certain position changes significantly (for example, from smooth to irregular), identify that position as a variant point.

[0133] Starting from the identified variant points, construct a matrix describing the changes from these points to the entire feature line; the matrix records the coordinates, curvature change values, and direction change values of each variant point; match this matrix with a pre-set crack defect template. For example, if the change matrix in a certain area highly matches the crack template, mark that area.

[0134] According to the matching result, mark the areas where cracks may exist on the two-dimensional image of the gear to generate a preliminary defect map.

[0135] Perform edge detection on the preliminary defect map, and use the Canny edge detection algorithm to identify the crack edges. Extract connected components, that is, in the edge detection result, find all sets of connected pixels.

[0136] Calculate the geometric properties of each connected component: 1. Area: the number of pixels contained in the crack area; 2. Perimeter: the edge length of the crack area; for example, assume that we define an area with an area greater than 50 square pixels and a perimeter greater than 30 pixels as a valid crack. According to these criteria, filter out the true crack areas and record their specific locations.

[0137] As Figure 5 shown, based on the filtered valid crack areas, generate the final defect map and save the final defect map to a data storage device, such as a hard disk or cloud storage, for subsequent analysis and quality control.

[0138] To further demonstrate the superiority of a vision-based injection molding defect analysis method for plastic molds provided by the present invention, the present invention also provides a vision-based injection molding defect analysis device for plastic molds, as Figure 7 shown, the device includes:

[0139] An illumination component for configuring a specific area on the surface of the injection molded part to set illumination points and activating an illumination source to focus on illuminating the illumination points to generate light reflection;

[0140] A camera component for capturing light reflections from multiple different angles and collecting optical image data of the injection molded part;

[0141] An image processing unit for converting the optical image data into digital signals and integrating them into a set of image matrix data;

[0142] A feature extraction unit for extracting feature data on the continuity of optical feature lines in the image from the image matrix data;

[0143] A defect recognition unit for generating a defect map by comparing the changes in optical feature lines and identifying structural defects on the surface of the injection molded part;

[0144] A frequency domain conversion unit for converting the image feature data into the frequency domain, using filtering techniques to remove noise, and performing an inverse frequency domain transformation on the filtered data to reconstruct the defect map;

[0145] A storage unit for determining and recording the defect locations of the injection molded part.

[0146] As Figure 9 、 Figure 10 shown, an embodiment of the present invention also provides a vision-based injection molding defect analysis system for plastic molds, and the system includes:

[0147] A vision-based injection molding defect analysis device for plastic molds;

[0148] A control module for coordinating and controlling the work processes of each component;

[0149] A display module for displaying defect diagrams and related analysis results;

[0150] A database module for storing historical defect data and defect templates;

[0151] A communication module for data exchange with external systems;

[0152] Among them: the control module includes:

[0153] A main controller for performing automated operations of the defect analysis process;

[0154] A parameter setting unit for setting and adjusting various parameters of the analysis system;

[0155] The database module includes:

[0156] A defect record library for storing defect location information and image data of each analysis;

[0157] A defect template library for storing pre-set defect templates to identify defect types by matching defect templates.

[0158] In the description of the present invention, a large number of specific details are set forth. However, it is understood that the embodiments of the present invention may be practiced without these specific details. In some instances, well-known methods, structures, and technologies have not been shown in detail so as not to obscure the understanding of this specification. Similarly, it should be understood that, in order to streamline the disclosure of the present invention and assist in understanding one or more of the various inventive aspects, in the above description of the exemplary embodiments of the present invention, the various features of the present invention are sometimes grouped together in a single embodiment, figure, or description thereof. However, the disclosed method should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected by the claims, the inventive aspects lie in less than all the features of the preceding disclosed single embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into the detailed description, where each claim stands on its own as a separate embodiment of the present invention. It should be noted that, without conflict, the embodiments and features in the embodiments of the present application may be combined with each other. The present invention is not limited to any single aspect, nor to any single embodiment, nor to any arbitrary combination and / or permutation of these aspects and / or embodiments. Moreover, each aspect and / or embodiment of the present invention may be used alone or in combination with one or more other aspects and / or their embodiments.

[0159] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A vision-based plastic mold injection defect analysis method, characterized in that: The steps include: Configure specific areas of the molded part surface to set lighting points; activating the illumination source and focusing illumination on the illumination point to generate light reflection; Use camera equipment to capture light reflections from multiple angles and collect optical image data of injection molded parts; Converting optical image data into digital signals and storing them in a data storage device; Integrate the converted digital signals into a set of image matrix data; Extracting feature data of continuity of optical feature lines in the image from the image matrix data; By comparing the changes in optical characteristic lines, a defect map is generated to identify structural defects on the surface of injection molded parts; Convert image feature data to frequency domain and use filtering technology to remove noise; Perform inverse frequency domain transformation on the filtered data to reconstruct the defect map; Determination and documentation of defect locations in injection molded parts.

2. A visual-based plastic mold injection defect analysis method according to claim 1, characterized in that: The steps for setting the lighting point include: Preset multiple potential lighting areas on the surface of the injection molded part; Use photosensors to monitor the reflectivity of each potential lighting area to evaluate the lighting effect of each area; Based on the reflectivity data, the area with the highest reflectivity is automatically selected as the lighting point.

3. The method for analyzing plastic mold injection defects based on vision according to claim 1, characterized in that: The step of capturing light reflection from multiple different angles using a camera device includes: irradiating the same position of the injection molded part with different lighting intensities and / or different lighting angles each time, so that the injection molded part generates light reflection and is received by the camera device.

4. The method for analyzing plastic mold injection defects based on vision according to claim 1, characterized in that: Methods for converting optical image data into digital signals include: receiving optical image data captured by a camera device; inputting the received optical image data into an image processing unit; In the image processing unit, grayscale processing is performed to convert the color image data into grayscale image data; Apply image coding algorithms to encode grayscale image data into digital signals.

5. The method for analyzing plastic mold injection defects based on vision according to claim 1, characterized in that: The method for extracting characteristic data of continuity of optical characteristic lines in an image from image matrix data comprises the following steps: Calculate the gradient of the image matrix data; Record the gradient direction and gradient amplitude of each pixel; Convert the gradient direction and gradient magnitude into polar coordinates; Perform curve fitting on each point in the polar coordinate system to obtain preliminary characteristic lines; Calculate the continuity of the fitted curve and mark the discontinuity points; Delete or correct the data corresponding to the discontinuity points; Update the fitting curve to obtain the final optical characteristic line.

6. A visual-based plastic mold injection defect analysis method according to claim 5, characterized in that: The formula for calculating the gradient of image matrix data is as follows: Where A is the image matrix; G x and G y are the gradient matrices in the horizontal and vertical directions, respectively.

7. A visual-based plastic mold injection defect analysis method according to claim 6, characterized in that: The calculation formula of gradient amplitude is: The calculation formula for the gradient direction is:

8. The method for analyzing plastic mold injection defects based on vision according to claim 5, characterized in that: The method of generating a defect map and identifying structural defects on the surface of an injection molded part by comparing changes in optical characteristic lines includes the following steps: The extracted final optical characteristic lines are processed in segments, and the curvature and direction change of each segment of the characteristic line are calculated; Register the feature lines at different time points or angles, compare the feature line shapes at corresponding positions, and identify the variation points; Construct a change matrix from the mutation point to the entire feature line, and match the change matrix with the pre-set defect template; Mark matching areas and generate preliminary defect maps; Applying edge detection algorithm to the marked area in the preliminary defect map and extracting connected components in the edge detection results; Calculate the area, perimeter and other geometric characteristics of each connected component, screen the effective defect area based on the geometric characteristic threshold, confirm the screened defect area and record its location; The final defect map is generated and saved to a data storage device.

9. A vision-based plastic mold injection defect analysis device, characterized in that: include: A lighting assembly, used to configure a specific area on the surface of the injection molded part to set a lighting point, and activate a lighting source to focus on illuminating the lighting point to generate light reflection; A camera assembly, used to capture light reflections from multiple different angles and collect optical image data of the injection molded parts; An image processing unit, used for converting optical image data into digital signals and integrating them into a set of image matrix data; A feature extraction unit, used for extracting feature data of continuity of optical feature lines in the image from the image matrix data; The defect recognition unit generates a defect map by comparing the changes in the optical characteristic lines and identifies the structural defects on the surface of the injection molded parts; The frequency domain conversion unit is used to convert the image feature data into the frequency domain, use filtering technology to remove noise, and perform inverse frequency domain transformation on the filtered data to reconstruct the defect map; Storage unit for determining and documenting the location of defects in injection molded parts.

10. A vision-based plastic mold injection defect analysis system, characterized in that: include: The vision-based plastic mold injection defect analysis device as claimed in claim 9; Control module, used to coordinate and control the workflow of each component; Display module, used to display defect diagrams and related analysis results; Database module, used to store historical defect data and defect templates; Communication module, used for data exchange with external systems; Wherein: the control module comprises: A main controller for performing automated operations of the defect analysis process; Parameter setting unit, used to set and adjust various parameters of the analysis system; The database module includes: Defect record library, used to store defect location information and image data for each analysis; The defect template library is used to store pre-set defect templates and identify defect types by matching defect templates.

Images