An apparatus and method for detecting hand mold defects in real time, storage medium

Abstract

The application discloses a device and method for real-time detection of hand mold defects, and a storage medium, and the method comprises the following steps: acquiring a hand mold surface image to be detected; pre-processing the hand mold surface image; extracting an ROI (Region of Interest) of the pre-processed hand mold surface image; backing up the ROI, and obtaining an ROI backup image; and based on the ROI backup image, realizing real-time detection of the hand mold defects. The application fills the blank in the field of hand mold defect detection. The application improves the reliability of defect detection of irregular surfaces and shapes, and realizes the differentiation of hand mold defects.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, specifically relating to a device and method for real-time detection of defects in hand molds, and a storage medium. Background Technology

[0002] The nitrile rubber latex market is projected to reach $2.4 billion by 2025. Whether it's latex gloves or nitrile gloves, the production process relies heavily on glove molds. The general manufacturing process for medical rubber gloves includes: ingredient preparation, mold cleaning, mold impregnation with a coagulant, drying, impregnation with rubber, edge rolling, leaching, impregnation with a release agent / chlorine treatment, demolding, vulcanization, drying and molding, initial packaging, and sterilization (if applicable). Most of these processes involve direct contact with the mold, making mold quality testing crucial for ensuring the stability and reliability of the entire product manufacturing process.

[0003] Currently, quality inspection of hand molds largely relies on manual labor. Manual inspection is inefficient, and its effectiveness depends on individual subjective observation, easily leading to missed or false positives. With the increasing demands for automated and intelligent production, manual inspection cannot meet the high standards and real-time requirements of hand mold inspection. Utilizing machine vision solutions can save significant labor resources, effectively reduce enterprise production costs, and bring considerable benefits to modern industrial production.

[0004] Obtaining intuitive, clear, and accurate information on hand mold defects is a crucial step in quality control and improvement. It is of great significance for preventing product defects and improving processes, while also providing qualitative understanding of product quality and quantitative data support for subsequent quality grading based on production status. Surface quality inspection during hand mold production not only eliminates defective products, reduces raw material waste, and improves product quality, but also significantly improves worker conditions and increases production efficiency. Furthermore, defect detection for hand molds is currently lacking in China. Summary of the Invention

[0005] This invention aims to address the shortcomings of existing technologies by proposing a device, method, and storage medium for real-time detection of defects in hand molds. This improves the reliability of defect detection for irregular surfaces and shapes and enables the differentiation of hand mold defects.

[0006] To achieve the above objectives, the present invention provides the following solution:

[0007] A device for real-time detection of defects in hand molds includes: a visible light sensor module, an image acquisition module, a parameter setting module, and an image processing module;

[0008] The visible light sensor module uses a reflective visible light sensor to detect the hand mold passing by in real time and trigger the image acquisition module to acquire an image of the hand mold surface.

[0009] The image acquisition module is used to acquire images of the surface of the hand mold to be detected;

[0010] The parameter setting module is used to provide brightness parameters when the image acquisition module acquires images;

[0011] The image processing module is used to realize real-time detection of defects in the hand mold based on the surface image of the hand mold.

[0012] The present invention also provides a method for real-time detection of defects in hand molds, which is based on a method for real-time detection of defects in hand molds and includes the following steps:

[0013] Acquire an image of the surface of the hand mold to be inspected;

[0014] The image of the hand mold surface is preprocessed;

[0015] Extract the region of interest (ROI) from the preprocessed surface image of the hand mold;

[0016] The region of interest (ROI) is backed up to obtain a backup map of the ROI.

[0017] Based on the backup map of the Region of Interest (ROI), real-time detection of defects in hand molds is achieved.

[0018] Preferably, Gaussian filtering is used to denoise the hand mold surface image, removing redundant pixels to achieve preprocessing.

[0019] Preferably, the method for extracting the region of interest (ROI) is as follows:

[0020] The preprocessed hand mold surface image is converted to grayscale to obtain a grayscale image;

[0021] The grayscale image is subjected to dilation processing;

[0022] The grayscale image after the dilation process is binarized to obtain the edges;

[0023] The point set of the edge is selected by drawing a box to obtain the smallest positive bounding rectangle;

[0024] The region of interest (ROI) is obtained by extending the smallest positive bounding rectangle by pixels.

[0025] Preferably, the ROI region of interest backup image includes: a first ROI grayscale image, a second ROI grayscale image, and a three-channel ROI image.

[0026] Preferably, real-time detection of hand mold defects includes: determining whether the hand mold has material stuck to it, determining whether the hand mold has missing fingers, and determining whether the hand mold has cracks.

[0027] Preferably, the method for determining whether a hand mold has material residue is as follows:

[0028] The ROI three-channel image is converted from BGR to HSV color space.

[0029] A preset threshold is used to detect material residue on the hand mold based on the HSV color gamut space, and a Boolean variable is output to determine whether there is material residue on the hand mold.

[0030] The preferred method for determining whether a hand mold has missing fingers is as follows:

[0031] The first ROI grayscale image is divided into blocks for selection, resulting in three selection areas;

[0032] For the three selected areas, perform connected component detection to distinguish between forward and reverse hands and detect missing fingers in the hand mold;

[0033] If the number of connected components in the three selected areas are 5, 3, and 2 respectively, then the hand mold is a normal hand and there are no missing fingers;

[0034] If the number of connected components in the three selected areas are 5, 2, and 3 respectively, then the hand mold is a reverse hand and there are no missing fingers;

[0035] If the number of connected components in the three selected areas is less than a preset value, then the hand mold has missing fingers.

[0036] The preferred method for determining whether a hand mold has cracks is as follows:

[0037] Morphological gradient processing is performed on the grayscale image of the second ROI to obtain edge information.

[0038] The edge information is processed using a combined algorithm of dilation and closing operations, and a combined algorithm of erosion and opening operations, respectively.

[0039] The difference between the processing results of the combined dilation and closure operation algorithm and the combined erosion and opening operation algorithm is obtained.

[0040] By changing the shape and size of the structural elements, an opening operation is performed on the difference of the composite operation results to determine whether there are cracks in the hand mold.

[0041] The present invention also provides a storage medium storing machine-executable instructions, which, when invoked and executed by a processor, cause the processor to implement a method for real-time detection of defects in a hand mold.

[0042] Compared with the prior art, the beneficial effects of the present invention are as follows: the device can measure the dynamic hand mold in the production process in real time without contacting the workpiece; the device has a high degree of integration, the device body is simple and clear, the manufacturing cost is low, and the maintenance is convenient; as a modular detection system, the device can be moved according to the actual situation and is easy to use; the measurement process is quick and intuitive, without the need for other extra cumbersome operations, while maintaining relatively high measurement accuracy and reducing measurement errors. Attached Figure Description

[0043] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a schematic diagram of the overall structure of the device for real-time detection of defects in hand molds according to Embodiment 1 of the present invention;

[0045] Figure 2 This is a front view of the device in Embodiment 1;

[0046] Figure 3 This is a left view of Embodiment 1;

[0047] Figure 4 This is a flowchart of a method for real-time detection of hand mold defects according to Embodiment 2 of the present invention;

[0048] Figure 5 To obtain the hand mold image to be detected;

[0049] Figure 6 Regions A, B, and C are representative regions within the selected region of interest.

[0050] Figure 7 The graphs show the number of connected regions, where Figure (a) shows the number of connected regions for the three selected regions being 5, 3, and 2; Figure (b) shows the number of connected regions for the three selected regions being 3, 3, and 2; and Figure (c) shows the number of connected regions for the three selected regions being 5, 2, and 2.

[0051] Figure descriptions: 1-Strip light source; 2-Camera bracket; 3-Camera lens; 4-Optical bracket; 5-Upper base; 6-Connector; 7-Base; 9-Mounting plate; 10-Industrial control cabinet; 11-Display screen. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] Example 1

[0055] like Figure 1-3 As shown, a device for real-time detection of defects in hand molds includes: a visible light sensor module, an image acquisition module, a parameter setting module, and an image processing module;

[0056] The visible light sensor module uses a reflective visible light sensor to detect the hand mold passing by in real time and trigger the image acquisition module to acquire an image of the hand mold surface;

[0057] The image acquisition module is used to acquire images of the surface of the hand mold to be inspected; specifically, it includes an industrial camera and a lens.

[0058] The parameter setting module provides brightness parameters for the image acquisition module when acquiring images. Specifically, it includes a light source and a gain module. The light source is part of the image acquisition module and does not require brightness parameter settings; it uses fixed parameters from the manufacturer's factory specifications. It is an illumination device. It can enhance the contrast of defects, making them easier to detect.

[0059] The image processing module is used to realize real-time detection of defects in the hand mold based on the surface image of the hand mold, and specifically includes: a computer.

[0060] The device also includes: a power supply for powering the entire device; and an industrial control cabinet for housing the computer and the reflective visible light sensor.

[0061] Specifically, this device also includes: 1-strip light source; 2-camera bracket; 3-camera lens; 4-photoelectric bracket; 5-upper seat; 6-connector; 7-base; 9-mounting plate; 10-industrial control cabinet; 11-display screen.

[0062] Device connection method: The reflective visible light sensor module is placed inside the industrial control cabinet. The visible light emitted by the sensor is perpendicular to the axis of the hand mold on the glove production line. The sensor trigger is connected to the camera via wiring. The camera trigger is connected to the gain module. The camera needs to be connected to the computer's acquisition card via wiring. The light source is placed on the left and right sides of the lens, facing the hand mold being photographed. The hand mold being detected is parallel to the black background and forms a certain angle with the ground, so that the hand mold's own weight can be used. The cameras above and below the hand mold on one side of the production line do not interfere with each other.

[0063] Image Acquisition Process: By aligning the hand mold with a black background cloth parallel to the ground at a certain angle, the hand mold surface is positioned for optimal photography. Two cameras and lenses are positioned offset on either side of the hand mold to acquire the first and second images. During production, the reflective visible light sensor module is triggered by the hand mold to acquire the image of the hand mold to be inspected. This image is then processed to obtain information such as the size, color, number of fingers, and surface defects and damage, comprehensively determining whether the hand mold has material residue, missing fingers, or cracks. A modular structure ensures that it can be adjusted according to on-site needs. Specifically, after the gloves are removed from the hand mold at the end of production, the hand mold triggers the camera on the camera bracket 2 via a visible light sensor fixed on the photoelectric bracket 4. Simultaneously, the strip light source 1 is triggered to acquire a real-time image of the properly exposed hand mold to be inspected. These images are captured from the hand mold production line, with the shooting angle directly in front of the hand mold's palm. Using C++ and OpenCV libraries, the acquired hand mold images are saved for further processing.

[0064] Example 2

[0065] like Figure 4 As shown, the present invention also provides a method for real-time detection of defects in hand molds, implemented based on a device for real-time detection of defects in hand molds, comprising the following steps:

[0066] Obtain an image of the surface of the hand mold to be detected, such as Figure 5 As shown;

[0067] Preprocess the image of the hand mold surface;

[0068] Extract the region of interest (ROI) from the preprocessed hand mold surface image;

[0069] Backup the regions of interest (ROI) to obtain a backup map of the ROI.

[0070] Based on the ROI (Region of Interest) backup map, real-time detection of hand mold defects is achieved.

[0071] Gaussian filtering is used to denoise the hand mold surface image, removing redundant pixels and achieving preprocessing.

[0072] The method for extracting the Region of Interest (ROI) is as follows:

[0073] The preprocessed hand mold surface image is converted to grayscale to obtain a grayscale image;

[0074] Dilation is applied to the grayscale image;

[0075] Binarize the dilated grayscale image to obtain the edges;

[0076] Select the point set at the edge to obtain the smallest positive bounding rectangle;

[0077] Extend the smallest positive bounding rectangle by pixels to obtain the region of interest (ROI).

[0078] Specifically, after completing the image preprocessing, the ROI (Region of Interest) of the hand mold is processed. Since the trigger position for acquiring the hand mold is fixed and its position is basically centered, the preprocessed image is converted to grayscale, dilated, and then binarized. All edges are selected, and the point sets of the edges are bounded to determine the smallest positive bounding rectangle. To ensure the complete ROI of the hand mold is selected, the selected smallest positive bounding rectangle is extended outwards by 10 pixels to form the ROI of the hand mold.

[0079] Specifically, the ROI (Region of Interest) backup images include: the first ROI grayscale image, the second ROI grayscale image, and the ROI three-channel image.

[0080] Real-time detection of hand mold defects includes: determining whether the hand mold has material residue, whether the hand mold has missing fingers, and whether the hand mold has cracks.

[0081] The method to determine whether there is material residue on the hand mold is as follows:

[0082] The ROI three-channel image is converted from BGR to HSV color space.

[0083] A preset threshold is used to detect material residue on the hand mold based on the HSV color gamut space, and a boolean variable is output to determine whether there is material residue on the hand mold.

[0084] The preferred method for determining whether a hand mold has missing fingers is as follows:

[0085] The first ROI grayscale image is divided into blocks for selection, resulting in three selection areas (representative regions A, B, and C within the ROI), as shown below. Figure 6 As shown;

[0086] For the three selected areas, perform connected component detection to distinguish between forehand and backhand hands and detect missing fingers in the hand mold;

[0087] If the number of connected components in the three selected areas are 5, 3, and 2 respectively, then the hand model is a normal hand and there are no missing fingers;

[0088] If the number of connected components in the three selected areas are 5, 2, and 3 respectively, then the hand model is a reverse hand and there are no missing fingers;

[0089] If the number of connected components in the three selected areas is less than the preset value, then the hand mold has missing fingers.

[0090] The number of connected components is determined based on Figure 6 The three choices A, B, and C are related, with Figure 7 For example, the number of connected components in the three selection areas of hand mold (a) is 5, 3, and 2; the number of connected components in the three selection areas of hand mold (b) is 3, 3, and 2; and the number of connected components in the three selection areas of hand mold (c) is 5, 2, and 2.

[0091] If it is a reverse hand model, the positions of areas B and C are swapped, and the output result is 5, 2, 3.

[0092] The method for determining whether a hand mold has cracks is as follows:

[0093] Morphological gradient processing is performed on the grayscale image of the second ROI to obtain edge information (a continuous set of pixels);

[0094] The edge information (i.e., processing the pixel set and highlighting the edge information features) is processed by using a combined algorithm of dilation and closing operation and a combined algorithm of erosion and opening operation respectively.

[0095] The difference between the results of the combined algorithm of dilation and closing and the combined algorithm of erosion and opening is obtained.

[0096] By transforming the shape and size of the structuring elements and performing an opening operation on the difference of the composite operation results, the system can determine whether there are cracks in the hand mold. Here, the structuring elements refer to the key parameters of the operator that performs morphological algorithm processing on the grayscale image of the second ROI, specifically B1, B2, and B3.

[0097] In the combined algorithm of dilation and closing operations, the kernel matrix value is 1 / 4 of the kernel matrix value of the combined algorithm of erosion and opening operations. The difference between the two combined operation results obtained above is much better than the results of processing by a single morphological algorithm.

[0098] To achieve the desired goal, following the above algorithm flow, the improved morphological operator constructed using the basic operators is as follows:

[0099]

[0100] Where B1, B2, and B3 are structuring elements; G is the image after morphological gradient calculation, i.e.

[0101]

[0102] • For closing operations, For opening operation, Θ represents the expansion operation, and Θ represents the erosion operation.

[0103] Compared with the original edge detection image, the edge image obtained by processing the edge image with the above-mentioned improved morphological algorithm is clearer, smoother and more accurate, and some broken edges are also connected, thus successfully obtaining better contour information.

[0104] By obtaining all the outlines of the hand mold, the results are classified, and hand molds with defects such as cracks and dirt can be selected.

[0105] Example 3

[0106] The present invention also provides a storage medium storing machine-executable instructions, which, when called and executed by a processor, cause the processor to implement a method for real-time detection of defects in hand molds.

[0107] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for detecting hand mold defects in real time, characterized in that, The method is realized by a device for real-time detection of hand mold defects, and the device comprises: a visible light sensor module, an image acquisition module, a parameter setting module, and an image processing module; the visible light sensor module adopts a reflective visible light sensor, is used for detecting a real-time passing hand mold, and triggers the image acquisition module to acquire a hand mold surface image; the image acquisition module is used for acquiring a hand mold surface image to be detected; the parameter setting module is used for providing a brightness parameter when the image acquisition module acquires an image; the image processing module is used for realizing real-time detection of hand mold defects based on the hand mold surface image; the method comprises the following steps: acquiring a hand mold surface image to be detected; preprocessing the hand mold surface image; extracting an ROI region of interest of the preprocessed hand mold surface image; backing up the ROI region of interest to obtain an ROI region of interest backup image; realizing real-time detection of hand mold defects based on the ROI region of interest backup image; the ROI region of interest backup image comprises a first ROI gray image, a second ROI gray image, and an ROI three-channel image; the real-time detection of hand mold defects comprises judging whether the hand mold has material sticking, judging whether the hand mold has missing fingers, and judging whether the hand mold has cracks; the method for judging whether the hand mold has material sticking is as follows: performing BGR to HSV color space conversion on the ROI three-channel image; presetting a threshold value, performing hand mold material sticking detection based on the HSV color space, outputting a Boolean variable, and realizing judgment of whether the hand mold has material sticking; the method for judging whether the hand mold has missing fingers is as follows: performing block selection processing on the first ROI gray image to obtain three selected areas; performing connected domain detection on the three selected areas to distinguish right and left hands and perform hand mold missing finger detection; if the number of connected domains of the three selected areas is 5, 3, and 2 respectively, the hand mold is a right hand and does not have missing fingers; if the number of connected domains of the three selected areas is 5, 2, and 3 respectively, the hand mold is a left hand and does not have missing fingers; if the number of connected domains of the three selected areas is less than a preset value, the hand mold has missing fingers; the method for judging whether the hand mold has cracks is as follows: performing morphological gradient processing on the second ROI gray image to obtain edge information respectively using an inflation and closing operation composite algorithm and an erosion and opening operation composite algorithm to process the edge information; subtracting the processing result of the inflation and closing operation composite algorithm from the processing result of the erosion and opening operation composite algorithm to obtain a composite operation result difference; performing opening operation on the composite operation result difference by changing the shape and size of the structure element to realize judgment of whether the hand mold has cracks.

2. The method for real-time detection of hand mold defects according to claim 1, wherein a Gaussian filter is used to perform noise reduction processing on the hand mold surface image, remove redundant pixel points of the hand mold surface image, and realize preprocessing.

3. The method for detecting hand mold defects in real time according to claim 1, wherein, the method for extracting the ROI region of interest is as follows: performing graying on the preprocessed hand mold surface image to obtain a gray image; performing inflation processing on the gray image; The gray-scale image subjected to the expansion processing is binarized to obtain an edge; A minimum circumscribed rectangle is obtained by frame selection on a point set of the edge; The minimum circumscribed rectangle is extended by pixel points to obtain the ROI region of interest.

4. A storage medium, characterized by The storage medium stores machine executable instructions, and when the machine executable instructions are called and executed by the processor, the machine executable instructions cause the processor to implement the method for detecting hand mold defects in real time according to any one of claims 1 to 3.

Images