A plastic product delivery quality detection device and detection method

Abstract

The application belongs to the technical field of image detection, and particularly relates to a quality detection device and method for plastic products. A three-dimensional model of a plastic product is established, and a database is established by extracting the features of the three-dimensional model. When the surface of the plastic product is detected, a detection image of the plastic product is obtained, HOG features of the detection image are extracted, a simulation image of the three-dimensional model with the same perspective is obtained according to the three-dimensional model, and HOG features of the simulation image are extracted. Through comparison between the detection image and the simulation image, various defects existing on the surface of the plastic product can be accurately detected, the detection accuracy is improved, and the limitations existing in the detection are reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of image detection, and particularly relates to a factory quality detection device and method for plastic products. BACKGROUND

[0002] Plastic products are a general term for life, industrial and other products processed by using plastic as the main raw material. Plastic is a kind of synthetic high polymer material with plasticity, and has the advantages of light weight, low cost and strong plasticity.

[0003] It forms the three indispensable synthetic materials in daily life together with synthetic rubber and synthetic fiber. Specifically, plastic is a material that can be molded into a certain shape under certain temperature and pressure conditions by using natural or synthetic resin as the main component and adding various additives, and keeps the shape unchanged at room temperature.

[0004] Due to the excellent plasticity of plastic, products with complex surfaces can be easily manufactured by injection molding. During injection molding, due to the shrinkage of plastic, surface defects such as cracking, shrinkage, overflow, and jet lines may occur on the surface of the plastic product. In order to improve product quality, the surface of the plastic product needs to be detected before it is put into the market. There are mainly two detection methods: 1. manual detection of the surface of the plastic product, which has low detection efficiency, large error and many uncontrollable subjective factors; 2. visual detection technology for detecting the surface of the plastic product, which has high detection efficiency and high detection accuracy, but has certain limitations for the appearance size and hole size of the product, and has certain limitations for the products with jet line defects or complex shapes. SUMMARY

[0005] The purpose of the present application is to provide ABCDE, which can accurately detect various defects on the surface of the plastic product by collecting images of the appearance of the product and extracting HOG features of the images, generating simulated detection images by a three-dimensional model and extracting HOG features of the images, and comparing the detection images with the simulated images, thereby improving the detection accuracy and reducing the limitations in detection.

[0006] The technical scheme adopted by the present application is as follows:

[0007] A factory quality detection method for plastic products, comprising the following steps:

[0008] Step A: inputting three-dimensional model data of a product and generating a three-dimensional model, and establishing a database of model features according to the model features on the three-dimensional model;

[0009] Step B: acquiring detection images of the product at multiple different angles by a detection device, and extracting HOG features of the detection images;

[0010] Step C: select one of the detection images, identify the HOG features and related information on the detection image, compare and match the HOG features extracted from the detection image with the model features, adjust the spatial orientation of the three-dimensional model to be consistent with the actual spatial orientation of the product through the HOG feature matching;

[0011] Step D: according to the image acquisition orientation and image acquisition angle of the detection device, perform image simulation acquisition on the three-dimensional model to generate a simulation image, and extract the HOG features of the simulation image;

[0012] Step E: compare the simulation image with the corresponding detection image to obtain the surface defects of the product.

[0013] As a preferred scheme of the plastic product factory quality detection method, in step A, the specific steps of generating model features from the three-dimensional model include: generating point features, line features, hole and groove features and other related features from the three-dimensional model, classifying the above model features, and establishing a database of the model features; wherein the database of the point features at least includes the minimum distance, horizontal distance and vertical distance between any two point features; the database of the line features includes point feature data at both ends of the line, length of the line, curvature of the line and other related information; the database of the hole and groove features includes spatial coordinates of the center point, diameter of the hole, size of the groove and other related information.

[0014] As a preferred scheme of the plastic product factory quality detection method, in step B, the specific steps of extracting the HOG features of the detection image are as follows:

[0015] Step B1: image gray processing;

[0016] Step B2: Gamma correction lookup table is used for Gamma correction of the image;

[0017] Step B3: image gradient calculation, convolution is performed with a small area template to calculate the gradient;

[0018] Step B4: statistic cell gradient direction histogram, divide the image into a plurality of cell unit cells, the cell unit cells do not overlap, in each cell, divide all gradient directions into n bins as the horizontal axis of the histogram, and the gradient value cumulative value corresponding to the angle range as the vertical axis of the histogram, and then perform statistics to obtain the gradient direction histogram;

[0019] Step B5: normalizing, combining m cell cells into a block, and normalizing the block;

[0020] Step B6: combining the feature vectors of each block to obtain the HOG feature of the image.

[0021] As a preferred scheme of the plastic product factory quality detection method, in step C, the specific steps of identifying the HOG feature and its related information on the detection image, comparing and matching the HOG feature extracted from the detection image with the model feature include:

[0022] Step C1: selecting three point features from the selected detection image, constructing a plane through the three points, and establishing a coordinate system with one of the point features as the origin ;

[0023] Step C2: searching for point features matched with the three point features in step C1 through the model feature database combined with the line feature, hole and groove feature and other model features, and establishing a coordinate system , and The origins of the two feature points are matched;

[0024] Step C3: taking as a reference, synchronously adjusting and the three-dimensional model so that the vector direction of is consistent with the vector direction of , and the spatial orientation of the three-dimensional model is completely consistent with the actual spatial orientation of the product.

[0025] As a preferred scheme of the plastic product factory quality detection method, in step C3, the specific steps of making the vector direction of consistent with the vector direction of include:

[0026] Step C3-1: establishing two straight edge lines through the three point features in step C1, wherein The origin of is the starting point of the straight edge line, and the other two point features are the endpoints of the two straight edge lines; two inclined edge lines are established through the three point features in step C2, wherein The origin of

[0027] Step C3-2: a virtual right triangle is established by the corresponding straight side line and the oblique side line, and the included angle of the straight side line and the oblique side line is obtained, according to the above-mentioned included angle, the vector direction of is adjusted, so that the vector direction of is consistent with the vector direction of

[0028] A plastic product factory quality detection device, at least comprising: a plurality of image acquisition elements for image acquisition of the plastic product from multiple dimensions; a feeding module for conveying the plastic product; a sorting module for sorting the products with surface defects.

[0029] A computer device comprising a storage element and a processing unit, the storage element internally storing a detection program for plastic product surface detection, and the processing unit can execute the steps of the plastic product factory quality detection method of any one of the above-mentioned methods when running the detection program.

[0030] A computer readable and writable storage medium, which stores a three-dimensional software and a detection program, the three-dimensional software is used to establish a three-dimensional model of the plastic product, and the detection program is executed by the processing unit to realize the steps of the plastic product factory quality detection method of any one of the above-mentioned methods.

[0031] The technical effects obtained by the present application are: the present application can accurately detect various defects (including spray lines, cracks, and deformation of modeling surface) on the surface of the plastic product by image acquisition of the appearance of the product and extraction of the HOG features of the image, comparison of the detection image and the simulation image, improvement of the detection accuracy, and reduction of the limitations in the detection process; the present application can perform multi-dimensional image acquisition of the product by multiple image acquisition elements, and can detect the product without placing it at a specific angle, which is convenient for disordered detection of the product. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 It is a system flowchart of the present application;

[0033] Figure 2 It is a schematic diagram of the present application when the detection equipment collects the image of the plastic product;

[0034] Figure 3 It is a schematic diagram of the present application when the detection image is detected;

[0035] Figure 4 It is a schematic diagram of the present application when the three-dimensional model is detected;

[0036] Figure 5 ​​​​​A schematic diagram of the right triangle of the present application A schematic diagram of the right triangle of the present application

[0037] Figure 6 A schematic diagram of the internal structure of the computer device of the present application. DETAILED DESCRIPTION

[0038] In order to make the purpose and advantages of the present application more clear and explicit, the present application will be specifically described below in combination with examples. It should be understood that the following text is only used to describe one or several specific embodiments of the present application, and does not strictly limit the protection scope of the present application. EMBODIMENT

[0039] As shown in Figure 1 and Figure 2 , the first embodiment of the present application provides a method for detecting the quality of a plastic product before leaving the factory, which comprises the following steps:

[0040] Step A: input the three-dimensional model data of the product and generate a three-dimensional model, identify the model features on the three-dimensional model, generate point features, line features, hole and groove features and other related features according to the three-dimensional model, classify the above model features, and establish a model feature database; wherein the database of the point features at least includes the minimum distance, horizontal distance and vertical distance between any two point features; the database of the line features includes the point feature data at both ends of the line, the length of the line, the curvature of the line and other related information; the database of the hole and groove features includes the spatial coordinates of the center point, the diameter of the hole, the size of the groove and other related information;

[0041] Step B: obtain multiple detection images of the product at different angles through a detection device, process the detection images, and extract the HOG features of the detection images;

[0042] Step C: select one of the detection images, identify the HOG features and their related information on the detection image, compare and match the extracted HOG features on the detection image with the model features, adjust the spatial orientation of the three-dimensional model to make it completely consistent with the actual spatial orientation of the product through HOG feature matching;

[0043] Step D: according to the image acquisition orientation and image acquisition angle of the detection device, perform image simulation acquisition on the three-dimensional model to generate simulation images, and extract the HOG features of the simulation images;

[0044] Step E: compare the simulation images with the corresponding detection images to obtain the surface defects of the product.

[0045] Further, in step B, the specific steps of extracting the HOG features of the detection images are as follows:

[0046] Step B1: image gray processing, the empirical formula for image gray processing is: ;

[0047] Step B2: Gamma correction: using Gamma correction lookup table to perform Gamma correction on the image;

[0048] Specifically, the Gamma correction lookup table is pre-established.

[0049] Step B3: image gradient calculation, using horizontal edge operator and vertical edge operator , the gradient is calculated by convolution with a small area template, wherein the pixel point module length calculation formula is , and the pixel point gradient direction calculation formula is ;

[0050] It should be noted that, is the value of the pixel point after Gamma correction, , ;

[0051] Step B4: statistic cell gradient direction histogram, divide the image into several cell unit cells, the cell unit cells do not overlap, in each cell, divide all gradient directions into n bins as the horizontal axis of the histogram, and the gradient value cumulative value corresponding to the angle range as the vertical axis of the histogram, and then statistic to obtain the gradient direction histogram;

[0052] Step B5: normalization, combine m cell unit cells into a block (preferably m=4), and normalize the block, wherein the normalization empirical formula is: ;

[0053] Here, is the block pixel value after data normalization, is the block pixel value before data normalization, is the maximum value of the image pixel, is the minimum value of the image pixel;

[0054] Further, in order to avoid image details or edge blur, m is preferably 4.

[0055] Step B6: combine the feature vector of each block to obtain the HOG feature of the image.

[0056] Here, the steps for establishing the Gamma correction lookup table include: performing normalization, pre-compensation, and inverse normalization calculations on a total of 256 elements with grayscale pixel values ​​from 0 to 255, and then storing the results into the Gamma correction lookup table. The formula for calculating the pre-compensation coefficient is as follows: The formula for inverse normalization is: ,in, This is the original pixel value of that pixel. The Gamma value, for example, if the original value of a pixel is 185 and the Gamma value is 2.1, is calculated using the formula: , Similarly, by calculating a total of 256 elements in the grayscale pixel values ​​from 0 to 255, a Gamma correction lookup table is established;

[0057] For example, when pixel A When the value is 240 and the Gamma value is 2.1, the Gamma correction lookup table can be used to quickly retrieve the result. .

[0058] Furthermore, in step C, the specific steps of identifying the HOG features and related information on the detected image, and comparing and matching the extracted HOG features with the model features, include:

[0059] Step C1: From the selected detection image, select three point features, construct a plane using the three points, and establish a coordinate system with one of the point features as the origin. ;

[0060] Step C2: Using the model feature database and combining the model features such as line features and hole / groove features, retrieve point features that match the three point features described in Step C1, and establish a coordinate system. ,and The origin and The origin is the two matching feature points;

[0061] Step C3: with For reference, adjust synchronously And 3D models, making vector direction and The vector directions are consistent, which makes the spatial orientation of the 3D model completely consistent with the actual spatial orientation of the product.

[0062] In one specific embodiment, such as Figures 2 to 4 As shown, three point features are selected from the chosen detection image. , and ,Establish and a line between , the length of which is a line between , the length of which is ;

[0063] Here, since the detection images are all two-dimensional images, and are projection values, wherein the projection plane is the plane where the image acquisition element that acquires the detection image is located, denoted as ;

[0064] A plane is established through , with as the reference point, as the reference plane, as the +X axis to establish a coordinate system ; ;

[0065] It should be noted that is parallel to ;

[0066] As shown in Figure 4 , through the model special database in step A, combined with the model characteristics such as line features, hole and groove features, point features matching the three point features in step C1 are searched , and , a line between and is established, the length of which is , a line between and is established, the length of which is ; ;

[0067] A plane is established through , with as the reference point, as the reference plane, as the +X axis to establish a coordinate system ; ;

[0068] With as the reference, adjust and the three-dimensional model synchronously, so that the vector direction of is the same as that of ​​​The vector directions are consistent, which makes the spatial orientation of the 3D model completely consistent with the actual spatial orientation of the product.

[0069] Furthermore, in step C3, the following is made: vector direction and The specific steps for aligning vector directions include:

[0070] Step C3-1: Establish two straight lines using the three point features described in step C1, wherein, The origin is the starting point of the straight edge, and the other two points mentioned above are the ending points of the two straight edges, respectively; two hypotenuses are established using the three point features mentioned in step C2, wherein, The origin is the starting point of the two hypotenuses, and the other two points mentioned above are the ending points of the two hypotenuses, and the two straight lines and the two hypotenuses correspond one-to-one.

[0071] Step C3-2: Construct a virtual right triangle using the corresponding straight side and hypotenuse, and calculate the angle between the straight side and hypotenuse. Adjust the triangle according to this angle. The vector direction makes vector direction and The vector directions are consistent;

[0072] In one specific embodiment, such as Figure 5 As shown, using and Construct a virtual right triangle ,in, Let it be the hypotenuse. and Substituting into the standardized function, we obtain ,by The Y-axis is the axis, and the point features are... Rotate around the center point Its rotation angle is The standardized function is: ;

[0073] use and Construct a virtual right triangle ,in, Let it be the hypotenuse. and Substituting into the standardized function, we obtain ,by The X-axis in the diagram is the axis, and the point features are... Rotate around the center point Its rotation angle is The standardized function is: and according to and adjusting the vector direction of such that the vector direction of is consistent with the vector direction of .

[0074] A plastic product factory quality detection device, at least comprising: a plurality of image acquisition elements, the image acquisition elements are used for image acquisition of the plastic product from multiple dimensions; a feeding module, the feeding module is used for conveying the plastic product; a sorting module, the sorting module is used for sorting the products with surface defects.

[0075] As shown in Figure 6 , the present application also provides a computer device, which can be a server or other terminal with data processing capability. The computer device comprises a processing unit, a storage element, a network interface, a database display screen and an input device connected through a system bus. Among them, the processing unit of the computer design is used to provide computing and control ability. The storage element of the computer device comprises a non-volatile storage medium and an internal storage element. The non-volatile storage medium stores an operating system, a detection program and a database. The internal storage provides an environment for the operation of the operating system and the detection program in the non-volatile storage medium. The database of the computer device is used to store all the data required in the quality detection process of the plastic product. The network interface of the computer device is used to communicate with the external terminal through the network connection. The detection program is executed by the processing unit to realize the steps of the plastic product factory quality detection method.

[0076] Those skilled in the art can understand that Figure 6 the structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied.

[0077] An embodiment of the present application also provides a computer readable and writable storage medium, which stores a three-dimensional software and a detection program, the three-dimensional software is used to establish a three-dimensional model of the plastic product, and the detection program is executed by the processing unit to realize the steps of the plastic product factory quality detection method.

[0078] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by detecting programs instructing relevant hardware, and the detecting programs can be stored in a non-volatile computer readable storage medium. When the detecting programs are executed, the processes of the above-mentioned embodiment methods can be included. Any reference to storage elements, storage, databases or other media in the present application and used in the embodiments can include non-volatile and / or volatile storage elements. Non-volatile storage elements can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile storage elements can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM).

[0079] It should be noted that, in this document, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, device, article or method that comprises a list of elements does not include only those elements recited, but can also include other elements not expressly listed or inherent to such process, device, article or method. Without more limitations, an element defined by the statement "comprising a" does not exclude the existence of additional identical elements in the process, device, article or method that includes the element.

[0080] The above description is only the preferred embodiment of the present application, and it should be noted that for those skilled in the art, without departing from the principles of the present application, some improvements and refinements can be made, which should also be considered as the protection scope of the present application. The structures, devices and operation methods not specifically described and explained in the present application are implemented according to the conventional means in the art, unless otherwise specified and limited.

Claims

1. A method for detecting the outgoing quality of a plastic product, characterized in that: It comprises the following steps: Step A: inputting three-dimensional model data of a product and generating a three-dimensional model, and establishing a model feature database according to model features on the three-dimensional model; Step B: acquiring detection images of the product at multiple different angles through a detection device, and extracting HOG features of the detection images; Step C: selecting one of the detection images, identifying HOG features and related information on the detection image, comparing and matching the HOG features extracted from the detection image with model features, and adjusting a spatial orientation of the three-dimensional model to make it completely consistent with an actual spatial orientation of the product through the HOG feature matching; Step D: performing image simulation acquisition on the three-dimensional model according to an image acquisition orientation and an image acquisition angle of the detection device, generating a simulation image, and extracting HOG features of the simulation image; Step E: comparing the simulation image with a corresponding detection image to obtain surface defects of the product.

2. The method for detecting the quality of a plastic product before delivery according to claim 1, wherein: In the step A, the specific steps of generating model features from the three-dimensional model include: generating point features, line features, hole and groove features and other related features from the three-dimensional model, classifying the above model features, and establishing a database of the model features; wherein the database of the point features at least includes minimum distances, horizontal distances and vertical distances between any two point features; the database of the line features at least includes point feature data at both ends of the line, length of the line and curvature of the line; and the database of the hole and groove features at least includes spatial coordinates of a center point, diameter of a hole and size of a groove.

3. The method for detecting the quality of a plastic product before it leaves the factory according to claim 1, characterized in that: In the step B, the specific steps of extracting HOG features of the detection image are as follows: Step B1: image gray processing; Step B2: Gamma correction lookup table is used to perform Gamma correction on the image; Step B3: image gradient calculation, convolution is performed with a small area template to calculate the gradient; Step B4: statistical cell gradient direction histogram, the image is divided into a plurality of cell unit cells, the cell unit cells do not overlap, in each cell, all gradient directions are divided into n bins as the horizontal axis of the histogram, and the gradient value cumulative value corresponding to the angle range is taken as the vertical axis of the histogram, and statistics are performed to obtain the gradient direction histogram; Step B5: normalization, combining m cell unit cells into a block, and normalizing the block; Step B6: combining the feature vectors of each block to obtain the HOG features of the image.

4. The method for detecting the quality of a plastic product before it leaves the factory according to claim 2, characterized in that: In the step C, the specific steps of identifying HOG features and related information on the detection image, comparing and matching the HOG features extracted from the detection image with model features include: Step C1: Select three point features from the selected detection image, construct a plane with the three points, and establish a coordinate system with one of the point features as the origin ; Step C2: through the model feature database, combined with the line feature, hole feature, retrieve the point feature matched with the three point features in step C1, and establish a coordinate system , and The origin of The origin of the two feature points is matched; Step C3: adjusting the vector direction of the three-dimensional model and the vector direction of the product actual space orientation in step C2 synchronously, so that the spatial orientation of the three-dimensional model and the product actual space orientation are completely consistent. the spatial orientation of the three-dimensional model and the product actual space orientation are completely consistent.​​​ 5. The method of claim 4, wherein: In the step C3, the vector direction of and the vector direction of are made consistent by the following specific steps. Step C3-1: Establish two straight edge lines by the three point features in Step C1, wherein, the origin is the starting point of the straight edge line, and the other two point features are the end points of the two straight edge lines respectively; Step C3-2: Establish two oblique edge lines by the three point features in Step C2, wherein, the origin is the starting point of the oblique edge line, and the other two point features are the end points of the two oblique edge lines respectively, and the two straight edge lines and the two oblique edge lines correspond one by one; Step C3-2: A virtual right triangle is established by the corresponding straight side line and the oblique side line, and the included angle of the straight side line and the oblique side line is obtained, according to the above-mentioned included angle, the vector direction of is adjusted so that the vector direction of is consistent with the vector direction of .​​ 6. A device for detecting the quality of a plastic product at the time of shipment, which is suitable for use in a method for detecting the quality of a plastic product at the time of shipment according to any one of claims 1 to 5, characterized by At least including: A plurality of image acquisition elements for image acquisition of the plastic product from multiple dimensions; A feeding module for conveying the plastic product; A sorting module for sorting the products with surface defects.

7. A computer device, comprising: The device comprises a storage element and a processing unit, the storage element internally stores a detection program for surface detection of plastic products, and the processing unit can execute the steps of the quality detection method of the plastic products according to any one of claims 1-5 when the detection program is executed.

8. A computer readable and writable storage medium having stored thereon a three-dimensional software and a detection program, characterized by, The three-dimensional software is used to establish a three-dimensional model of the plastic product, and the detection program is executed by the processing unit to realize the quality detection method of the plastic product according to any one of claims 1-5.

Images