Image recognition type mobile phone screen defect detection device using deep learning

Abstract

The utility model discloses an image recognition type mobile phone screen defect detection device utilizing deep learning. A base is connected to a lifting platform so as to drive the lifting platform in the vertical direction; the one-dimensional driving table is arranged on the base and completes linear driving in a horizontal plane; the sample plate is mounted and driven on the one-dimensional driving table; the sample plate is driven by the one-dimensional driving table to enter and leave the frame body. According to the image recognition type mobile phone screen defect detection device based on deep learning provided by the utility model, light generated by the light source part is homogenized through the light homogenizing layer, so that bright spots which interfere with the detection accuracy are not easy to appear on a mobile phone; the sample plate is driven by the one-dimensional driving table to enter and be separated from the frame body, and the sample plate can be combined with and separated from the lower opening of the frame body through the work of the lifting platform; when the sample plate is combined with the lower opening, the detection work can be performed; when the sample plate is separated from the lower opening, the actions of sampling and lofting can be carried out.

Description

Technical Field

[0001] The utility model relates to the field of screen defect recognition, and in particular to an image recognition type mobile phone screen defect detection device using deep learning. Background Art

[0002] Smartphones have become an indispensable part of daily life. Whether it is work, study or entertainment, mobile phones play a vital role. The number of mobile phones has remained high and has shown an increasing trend year by year. In order to meet the continuous demand of the market, a large number of new mobile phones are put on the market every year. At the same time, with the acceleration of the replacement speed, a large number of old mobile phones are also eliminated. These old mobile phones need to undergo a series of tests and evaluations during the recycling and processing process in order to reuse or dispose of them reasonably.

[0003] As one of the core components of mobile phones, the quality and performance of mobile phone screens directly affect the user experience. In the production process of new mobile phones, defect detection of mobile phone screens is a crucial link. This can ensure that every mobile phone leaving the factory has a high-quality screen and provide users with a good visual experience. For the acceptance of old mobile phones, the inspection of mobile phone screens cannot be ignored. Since mobile phone screens are affected by various factors during use, such as falling and scratching, they are prone to damage or aging. Through the careful inspection of old mobile phone screens, their status and value can be accurately evaluated, providing a decision-making basis for subsequent reuse or disposal.

[0004] In terms of mobile phone screen defect detection, the use of advanced image processing technology and machine learning algorithms can realize automatic detection of mobile phone screens. This technology can not only greatly improve detection efficiency, but also reduce the error rate of manual operation, ensuring the accuracy and reliability of detection results. How to improve detection efficiency and detection accuracy remains the research focus of the above technologies. Utility Model Content

[0005] The main purpose of the utility model is to provide an image recognition type mobile phone screen defect detection device using deep learning, aiming to solve the problems of low detection efficiency and accuracy in screen defect image recognition.

[0006] In order to achieve the above-mentioned purpose, the utility model provides an image recognition type mobile phone screen defect detection device using deep learning, comprising:

[0007] A frame body, with an upper opening and a lower opening at the upper and lower ends respectively, and a light absorbing layer is arranged on the inner peripheral wall of the frame body;

[0008] A light environment portion, closed at the upper opening, comprising a light source portion and a light homogenizing layer arranged corresponding to the light source portion;

[0009] A camera unit is mounted on the peripheral wall of the frame and faces the lower opening;

[0010] A lifting platform is arranged below the frame;

[0011] A base, connected to the lifting platform to drive the lifting platform in a vertical direction;

[0012] A one-dimensional driving platform, arranged on the base to perform linear driving in a horizontal plane;

[0013] A sample plate, mounted and driven on the one-dimensional driving stage;

[0014] A processor controls the operation of the light environment unit, the camera unit, the lifting platform and the one-dimensional driving platform, and the processor receives data sent by the camera unit;

[0015] The sample plate is within the range where the one-dimensional driving stage drives the sample plate into and out of the frame.

[0016] Furthermore, the mobile phone screen defect detection device also includes a robot arm that performs two-dimensional movement in a plane, and the robot arm completes a pick-and-place action in a vertical direction, and the robot arm is arranged corresponding to the sample plate.

[0017] Furthermore, the image-based mobile phone screen defect detection device also includes a purge function unit controlled by the processor, and the purge function unit includes a purge unit and a negative pressure unit both introduced from the peripheral wall of the frame.

[0018] Further, the sample plate comprises a lower sample plate and an upper sample plate detachably disposed on the lower sample plate, wherein the lower sample plate is connected to the one-dimensional driving platform.

[0019] Furthermore, the light source unit and the light-homogenizing layer cooperate in a side-entry manner, and there are multiple light source units, which are light sources of different colors.

[0020] Furthermore, a roughening layer is provided on the outer surface of the light absorbing layer.

[0021] Furthermore, the number of the camera parts is four, and they are evenly arranged around the peripheral wall of the frame.

[0022] Furthermore, the one-dimensional driving platform is a screw-driven type or a private server-driven type.

[0023] Furthermore, a sample groove is provided on the upper surface of the sample plate.

[0024] Furthermore, a chamfered structure is provided on the inner periphery of the upper portion of the sample tank.

[0025] The utility model provides an image recognition type mobile phone screen defect detection device using deep learning, which evens out the light generated by the light source part through a uniform light layer, so that bright spots that interfere with the detection accuracy are less likely to appear on the mobile phone; the sample plate is driven by a one-dimensional drive table to enter and leave the range of the frame, and the lifting platform can combine and separate the sample plate from the lower opening of the frame; when the sample plate is combined with the lower opening, the optical environment in the frame has been stabilized and the detection work can be carried out; when the sample plate is separated from the lower opening, sampling and placement actions can be carried out; the mobile phones can be transported to the range of the frame in batches through the one-dimensional drive table, thereby improving the detection speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of a frame in an image recognition-type mobile phone screen defect detection device using deep learning according to an embodiment of the present utility model;

[0027] Figure 2 This is a schematic diagram of the light environment part of the image recognition type mobile phone screen defect detection device using deep learning in one embodiment of the utility model (the top part is hidden);

[0028] Figure 3 This is a working schematic diagram (sampling process) of an image recognition-based mobile phone screen defect detection device using deep learning in one embodiment of the utility model;

[0029] Figure 4 This is a working schematic diagram (testing process) of an image recognition-based mobile phone screen defect detection device using deep learning in one embodiment of the utility model.

[0030] The realization of the purpose, functional features and advantages of the utility model will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0031] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0032] Those skilled in the art will appreciate that, unless expressly stated, the singular forms "one", "said", "above", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present utility model refers to the presence of the features, integers, steps, operations, elements, units, modules, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, units, modules, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0033] Those skilled in the art will understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as the general understanding of ordinary technicians in the field to which the utility model belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0034] Reference Figures 1 to 4 In one embodiment of the utility model, a mobile phone screen defect detection device using image recognition using deep learning includes:

[0035] The frame 100 has an upper opening and a lower opening at its upper and lower ends respectively, and a light absorbing layer is disposed on the inner peripheral wall of the frame 100;

[0036] A light environment portion 200 is enclosed in the upper opening, and the light environment portion 200 includes a light source portion 210 and a light uniforming layer 220 arranged corresponding to the light source portion 210;

[0037] The camera unit 300 is mounted on the peripheral wall of the frame 100 and faces the lower opening;

[0038] A lifting platform 400 is disposed below the frame 100;

[0039] A base 500 connected to the lifting platform 400 to drive the lifting platform 400 in a vertical direction;

[0040] A one-dimensional driving platform 600 is disposed on the base 500 to perform linear driving in a horizontal plane;

[0041] A sample plate 700, mounted and driven on the one-dimensional driving stage 600;

[0042] A processor controls the operation of the light environment unit 200, the camera unit 300, the lifting platform 400 and the one-dimensional driving platform 600, and receives data sent by the camera unit 300;

[0043] The sample plate 700 is driven by the one-dimensional driving stage 600 to enter and leave the frame 100 .

[0044] In the prior art, how to improve the detection efficiency and accuracy of screen defect image recognition remains a research focus.

[0045] In the present invention, the frame 100 is used as the detection space. The upper and lower ends of the frame 100 are respectively an upper opening and a lower opening, and a light absorbing layer is provided on the inner peripheral wall of the frame 100. The light absorbing layer optimizes the optical environment in the frame 100, and specifically prevents the occurrence of reflection phenomenon to a large extent.

[0046] The light environment part 200 is closed at the upper opening, and the light environment part 200 includes a light source part 210 and a light homogenizing layer 220 arranged corresponding to the light source part 210. The light generated by the light source part 210 is homogenized by the light homogenizing layer 220, so that bright spots that interfere with the detection accuracy are not likely to appear in the mobile phone. The specific working mode of the light source part 210 can be direct-down or side-injection. The light emitted from the light homogenizing layer 220 can be pure light of various colors or mixed color light. The light homogenizing layer 220 can also be combined with various types of optical films to further enhance the optical environment in the frame 100.

[0047] The camera unit 300 is installed on the peripheral wall of the frame 100 and opens downward. The function of the camera unit 300 is to obtain an optical image inside the frame 100 from a specific angle, providing a basis for the detection of the mobile phone screen.

[0048] The lifting platform 400 is arranged below the frame 100. The base 500 is connected to the lifting platform 400. The lifting platform 400 is used to drive the entire base 500 to reciprocate in the vertical direction. A one-dimensional driving platform 600 is arranged on the base 500, and a sample plate 700 is arranged on the one-dimensional driving platform 600. The sample plate 700 is driven by the one-dimensional driving platform 600 to enter and leave the range of the frame 100. The lifting platform 400 can combine and separate the sample plate 700 with the lower opening of the frame 100. When the sample plate 700 is combined with the lower opening, the optical environment in the frame 100 has been stabilized. Detection work can be carried out; when the sample plate 700 is separated from the lower opening, sampling and placement actions can be carried out (which can be automatic or manual).

[0049] The processor coordinates and controls the light environment unit 200, the camera unit 300, the lifting platform 400 and the one-dimensional driving platform 600 to complete the test action. The specific control method of the processor can be completed by a preset program in the single-chip microcomputer, or by real-time detection and control through corresponding sensors, or a combination of the two control methods. The processor receives the data sent by the camera unit 300 and compares it with the database formed after deep learning, thereby completing the detection process.

[0050] In summary, the light generated by the light source unit 210 is homogenized by the light-uniforming layer 220, so that bright spots that interfere with detection accuracy are not likely to appear in the mobile phone; the sample plate 700 is driven by the one-dimensional driving platform 600 to enter and leave the range of the frame 100, and the lifting platform 400 can combine and separate the sample plate 700 with the lower opening of the frame 100; when the sample plate 700 is combined with the lower opening, the optical environment in the frame 100 has been stable, and detection work can be carried out; when the sample plate 700 is separated from the lower opening, sampling and placement actions can be performed; the one-dimensional driving platform 600 can be used to transport mobile phones in batches to the range of the frame 100, thereby improving the detection speed.

[0051] In one embodiment, the mobile phone screen defect detection device also includes a robot arm that performs two-dimensional motion in a plane, and the robot arm completes a pick-and-place action in a vertical direction. The robot arm is arranged corresponding to the sample plate 700.

[0052] In this embodiment, considering the need for automation, a manipulator is introduced to complete the process of picking up and placing the mobile phone, thereby improving work efficiency. The specific structure of the manipulator is not the focus. It has a two-dimensional drive platform that can achieve two-dimensional movement to complete the transfer action; it has a vertical drive structure so that the manipulator can move in the vertical direction to achieve sampling and placement actions. The sampling of the manipulator can be either negative pressure type or grasping type. The manipulator removes the mobile phone from the conveyor line and transfers it to the sample plate 700.

[0053] In one embodiment, the incident angle of the camera unit 300 is 40 to 50 degrees.

[0054] In this embodiment, the incident angle of the camera unit 300 is limited so that the mobile phone will not have a mirror image of the light environment unit 200, thereby improving the accuracy of detection.

[0055] In one embodiment, the image-based mobile phone screen defect detection device further includes a purge function unit controlled by the processor, and the purge function unit includes a purge unit and a negative pressure unit both introduced from the peripheral wall of the frame body 100 .

[0056] In this embodiment, considering that the working environments in different inspection workshops are different and dust-free environments cannot be met in some cases, a purge function unit is introduced, and the purge unit of the purge function unit is used to process the environment inside the frame 100, and the air introduced by the purge unit is further extracted out of the frame 100 through the negative pressure unit. In implementation, the purge unit and the negative pressure unit can be connected, that is, negative pressure and positive pressure can be achieved simultaneously through one fan, thereby simplifying the complexity of the entire structure.

[0057] In one embodiment, the sample plate 700 includes a lower sample plate 700 and an upper sample plate 700 detachably disposed on the lower sample plate 700 , wherein the lower sample plate 700 is connected to the one-dimensional driving stage 600 .

[0058] In this embodiment, the sample plate 700 is set to be split, the lower sample plate 700 serves as a base, and the upper sample plate 700 serves as a sample support. When the mobile phone model changes or the detection placement position is adjusted, adaptation can be achieved by replacing the upper sample plate 700.

[0059] In one embodiment, the light source unit 210 and the light homogenizing layer 220 are matched in an edge-entry manner, and there are multiple light source units 210 of different colors.

[0060] In this embodiment, considering that different screen defects have different degrees of difficulty in distinguishing under different color light environments, the number of light source units 210 is set to multiple, and in order to facilitate the convenient working switch of the light source unit 210, the uniform light effect adopts the side-incident type. Specifically, during the working process, the appropriate light source unit 210 can be selected for the mobile phone, or turned on in turn for independent detection.

[0061] In one embodiment, a roughened layer is disposed on the outer surface of the light absorbing layer.

[0062] In this embodiment, considering that the smooth light-absorbing layer may still produce reflections, roughening (matte treatment) is performed on its outer surface to further reduce the reflections. For example, micro-protrusions or micro-holes are added. Through the above matte treatment, the reflection effect is further reduced.

[0063] Reference Figures 1 to 4 In one embodiment, the number of the camera units 300 is four, and they are evenly arranged around the peripheral wall of the frame body 100 .

[0064] In this embodiment, considering that different defects have different degrees of difficulty in identifying at different angles, multiple camera units 300 are distributed in multiple directions around the frame 100, so that more comprehensive and accurate detection results can be obtained by performing corresponding comparisons respectively. Specifically, during the detection process, the images obtained by each camera unit 300 are compared with the corresponding deep learning result library. In some cases, the images obtained by different camera units 300 can also be combined for analysis.

[0065] In one embodiment, the one-dimensional driving platform 600 is a screw-driven type or a private-server-driven type.

[0066] In this embodiment, two models of one-dimensional driving stage 600 are provided for selection, which have the advantages of accurate driving position, mature technology and simple control.

[0067] Reference Figures 3 to 4 In one embodiment, a sample groove (not shown) is disposed on the upper surface of the sample plate 700 .

[0068] In this embodiment, a sample slot is introduced, and the position of the mobile phone on the sample plate 700 is fixed, and the optical environment is also fixed. Therefore, it is less difficult for the processor to find defects through image recognition, and the possibility of errors in defect features found by deep learning is also reduced.

[0069] In one embodiment, a chamfered structure is provided on the inner periphery of the upper portion of the sample slot.

[0070] In this embodiment, the chamfered structure provided on the sample slot plays a guiding role in the process of placing the mobile phone, and the difficulty of automatic placement or manual placement is reduced.

[0071] In summary, the image recognition type mobile phone screen defect detection device using deep learning provided by the utility model uniformizes the light generated by the light source part 210 through the light homogenizing layer 220, so that bright spots that interfere with the detection accuracy are not likely to appear in the mobile phone; the sample plate 700 is driven by the one-dimensional driving platform 600 to enter and leave the range of the frame 100, and the lifting platform 400 can combine and separate the sample plate 700 with the lower opening of the frame 100; when the sample plate 700 is combined with the lower opening, the optical environment in the frame 100 has been stabilized and detection work can be carried out; when the sample plate 700 is separated from the lower opening, sampling and placement actions can be performed; the one-dimensional driving platform 600 can be used to transport mobile phones in batches to the range of the frame 100, thereby improving the detection speed.

[0072] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An image recognition-based mobile phone screen defect detection device using deep learning, characterized in that: include: A frame body (100) having an upper opening and a lower opening at the upper and lower ends respectively, and a light absorbing layer is provided on the inner peripheral wall of the frame body (100); A light environment portion (200) is sealed at the upper opening, the light environment portion (200) comprising a light source portion (210) and a light homogenizing layer (220) arranged corresponding to the light source portion (210); A camera unit (300) is mounted on a peripheral wall of the frame (100) and faces the lower opening; A lifting platform (400) is arranged below the frame (100); A base (500) connected to the lifting platform (400) to drive the lifting platform (400) in a vertical direction; A one-dimensional driving platform (600) is arranged on the base (500) to perform linear driving in a horizontal plane; A sample plate (700), mounted on and driven by the one-dimensional driving stage (600); a processor, controlling the operation of the light environment unit (200), the camera unit (300), the lifting platform (400) and the one-dimensional driving platform (600), the processor receiving data sent by the camera unit (300); The sample plate (700) is driven by the one-dimensional driving platform (600) to enter and leave the frame (100).

2. The image recognition type mobile phone screen defect detection device using deep learning according to claim 1 is characterized in that: The mobile phone screen defect detection device also includes a robot arm that performs two-dimensional movement in a plane, the robot arm completes a pick-and-place action in a vertical direction, and the robot arm is arranged corresponding to the sample plate (700).

3. The image recognition type mobile phone screen defect detection device using deep learning according to claim 1 is characterized in that: The image-based mobile phone screen defect detection device also includes a purge function unit controlled by the processor, and the purge function unit includes a purge unit and a negative pressure unit both introduced from the peripheral wall of the frame (100).

4. The image recognition type mobile phone screen defect detection device using deep learning according to claim 1, characterized in that: The sample plate (700) comprises a lower sample plate (700) and an upper sample plate (700) detachably arranged on the lower sample plate (700), wherein the lower sample plate (700) is connected to the one-dimensional driving platform (600).

5. The image recognition type mobile phone screen defect detection device using deep learning according to claim 1, characterized in that: The light source unit (210) and the light homogenizing layer (220) cooperate in a side-entry manner, and there are a plurality of light source units (210), which are light sources of different colors.

6. The image recognition type mobile phone screen defect detection device using deep learning according to any one of claims 1 to 5, characterized in that: The outer surface of the light absorbing layer is provided with a roughening layer.

7. The image recognition type mobile phone screen defect detection device using deep learning according to any one of claims 1 to 5, characterized in that: The number of the camera units (300) is four, and they are evenly arranged around the peripheral wall of the frame (100).

8. The image recognition type mobile phone screen defect detection device using deep learning according to any one of claims 1 to 5, characterized in that: The one-dimensional drive platform (600) is a screw drive type or a private server drive type.

9. The image recognition type mobile phone screen defect detection device using deep learning according to any one of claims 1 to 5, characterized in that: The upper surface of the sample plate (700) is provided with a sample groove.

10. The image recognition type mobile phone screen defect detection device using deep learning according to claim 9, characterized in that: The inner periphery of the upper part of the sample tank is provided with a chamfered structure.

Images