Contour recognition method for all-in-one AI intelligent image recognition sensor

Abstract

The present application relates to the field of image processing technology, and discloses a contour recognition method for an integrated AI intelligent image recognition sensor, the method comprising: the CPU automatically adjusts the exposure brightness and the binarization threshold a to generate several effect graphs for the target; the user selects an effect graph as the target image, and the CPU records the photographing brightness value and the binarization threshold a corresponding to the target image; the user identifies the template image, and the CPU automatically recognizes the edge contour information of the target image based on the template image, and classifies the contour corresponding to the target image as simple or complex; when the contour corresponding to the target image is classified as simple, the CPU uses a contour similarity algorithm to judge the edge contour of the target image; otherwise, a skeleton detection algorithm is used to judge the edge contour of the target image; and the edge contour recognition result of the target image is output. By adopting the present application, it is possible to automatically recognize the contours of different target images, improve computing efficiency, and ensure the accuracy of the recognition results.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and specifically to a contour recognition method for an integrated AI intelligent image recognition sensor. Background Art

[0002] Traditional machine vision equipment is generally composed of a PC host, a camera, an independent light source, and a corresponding algorithm system. Looking at some existing machine vision equipment, it can be seen that they generally have the following shortcomings:

[0003] 1. The device is large in size, and PC installation requires a relatively large space;

[0004] 2. The wiring is long and prone to interference; the host must be placed in a fixed position, and the camera generally needs to be placed above or to the side of the object to be detected, at a certain distance from the host, and must be connected with a dedicated data cable (including power cable);

[0005] 3. It has high technical requirements for users. Manual focusing is required during installation to find the most suitable focusing distance. Due to the difference in the naked eye's recognition of zoom clarity, it will affect the accuracy of contour edge detection.

[0006] Therefore, a new visual inspection technology is urgently needed. Summary of the invention

[0007] The purpose of this application is to provide a contour recognition method for an integrated AI intelligent image recognition sensor to solve the technical problems raised in the above background technology.

[0008] To achieve the above objectives, the present application discloses the following technical solution: a contour recognition method for an integrated AI intelligent image recognition sensor, the method comprising the following steps:

[0009] According to the target selected by the user, the CPU automatically adjusts the exposure brightness and the binarization threshold a to take multiple photos of the target and generate several effect images;

[0010] The user selects one of the generated renderings as a target image, and the CPU records the camera brightness value and the binarization threshold a corresponding to the target image;

[0011] The user marks the template image with a rectangular frame, and the CPU automatically identifies the edge contour information of the target image based on the template image, and classifies the contour corresponding to the target image as simple or complex after analyzing the edge contour information;

[0012] When the contour corresponding to the target image is classified as simple, the CPU uses a contour similarity algorithm to determine the edge contour of the target image; when the contour corresponding to the target image is classified as complex, the CPU uses a skeleton detection algorithm to determine the edge contour of the target image;

[0013] Based on the judgment result, the edge contour recognition result of the target image is output.

[0014] Preferably, the contour similarity algorithm specifically includes:

[0015] Obtaining the grayscale value of the target image;

[0016] Performing median filtering on the target image;

[0017] Performing binarization processing on the target image;

[0018] Extracting a maximum contour based on sub-pixels in the target image;

[0019] Calculating the similarity between the target image and the maximum contour;

[0020] Compare the calculated similarity with the preset similarity;

[0021] When the similarity between the target image and the maximum surrounding contour is greater than the preset similarity, the maximum surrounding contour is output as the edge contour of the target image; otherwise, a prompt indicating recognition failure is output.

[0022] Preferably, the skeleton detection algorithm specifically includes:

[0023] Obtaining the grayscale value of the target image;

[0024] Performing median filtering on the target image;

[0025] Performing binarization processing on the target image;

[0026] Extracting a peripheral contour based on sub-pixels of the target image;

[0027] Performing concave contour extraction on the outer contour;

[0028] Calculating the similarity between the target image and the concave contour;

[0029] Compare the calculated similarity with the preset similarity;

[0030] When the similarity between the target image and the concave contour is greater than the preset similarity, the concave contour is output as the edge contour of the target image; otherwise, a prompt indicating recognition failure is output.

[0031] Preferably, when the target image is binarized, a binarization threshold is adopted.

[0032] Preferably, the integrated AI intelligent image recognition sensor includes a DC / DC power supply, an MCU, a button module, a display module, a light source module, an input / output interface module, a focusable lens, a network connection module, a storage module and the CPU; the DC / DC power supply, the MCU, the button module, the display module, the light source module, the input / output interface module, the focusable lens, the network connection module and the storage module are all integrated on an MCU baseboard, the CPU is arranged on a CPU core board, and the MCU baseboard is communicatively connected to the CPU core board through a board-to-board connection terminal; this integrated AI intelligent image recognition sensor integrates an embedded 3D GPU and a computing power NPU;

[0033] The button module, the display module, the light source module, and the input / output interface module are respectively connected to the MCU control, the DC / DC power supply is connected to a DC20-28V power supply, and the MCU is also connected to a status indicator light for displaying the working status.

[0034] Preferably, the storage module includes a storage hard disk and a TF card storage unit.

[0035] Preferably, the DC / DC power supply includes a first power supply unit for outputting +4V power to power the CPU core board, a second power supply unit for outputting +5V power to power the focusable lens, a third power supply unit for outputting +3.3V power to power the MCU, and a fourth power supply unit for outputting +1.8V power to the power-on timing of the CPU core board.

[0036] Preferably, the DC / DC power supply further includes an EMC protection circuit.

[0037] Preferably, the input-output interface module includes three external input interfaces and one external output interface, wherein the external output interface switches between NPN output and PNP output through the control of the MCU.

[0038] Beneficial effects: The contour recognition method for an integrated AI intelligent image recognition sensor of the present application, through the configuration of a dual algorithm, combined with the simple / complex pre-judgment of the target image contour, can automatically recognize the contours of different target images, improve computing efficiency and ensure the accuracy of the recognition results. Furthermore, the sensor used in this method adopts an integrated structure with a small device size and less wiring. Compared with traditional machine vision equipment, it has obvious advantages in cost, size, installation, maintenance and stability. In addition, the accuracy of the contour recognition results is improved with the application of this method. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 Flow chart of the contour recognition method for the integrated AI intelligent image recognition sensor provided in the embodiment of the present application

[0041] Figure 2 A system block diagram of an integrated AI intelligent image recognition sensor provided in an embodiment of the present application;

[0042] Figure 3 is a circuit diagram of a first power supply unit;

[0043] Figure 4 is a circuit diagram of a second power supply unit;

[0044] Figure 5 is a circuit diagram of a third power supply unit;

[0045] Figure 6 is a circuit diagram of a fourth power supply unit;

[0046] Figure 7 This is a schematic diagram of the power-on timing control pins of the CPU core board;

[0047] Figure 8 This is a schematic diagram of the external clock circuit of the MCU core board;

[0048] Fig. 9 It is a schematic diagram of the communication circuit between the CPU core board and the external I / O;

[0049] Fig.10 This is a schematic diagram of the EMC protection circuit;

[0050] Fig.11This is a schematic diagram of the IO signal communication / power input interface circuit;

[0051] Fig.12 This is a schematic diagram of the HDMI connector;

[0052] Fig.13 This is a schematic diagram of the pin connections corresponding to the three external input interfaces;

[0053] Fig.14 A schematic diagram of a circuit for an external output interface;

[0054] Fig.15 This is a pin connection diagram of the board-to-board socket J1;

[0055] Fig.16 This is a pin connection diagram of the board-to-board socket J2;

[0056] Fig.17 This is a pin connection diagram of the board-to-board socket J3;

[0057] Fig.18 This is a diagram showing the pin connections of the board-to-board socket J4. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0059] In this article, the term "comprising" is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, the elements defined by the sentence "comprising..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0060] This embodiment discloses a Figure 1 The contour recognition method for the integrated AI intelligent image recognition sensor shown includes the following steps;

[0061] According to the target selected by the user, the CPU automatically adjusts the exposure brightness and the binarization threshold a to take multiple photos of the target and generate several effect images;

[0062] The user selects one of the generated renderings as a target image, and the CPU records the camera brightness value and the binarization threshold a corresponding to the target image;

[0063] The user marks the template image with a rectangular frame, and the CPU automatically identifies the edge contour information of the target image based on the template image, and classifies the contour corresponding to the target image as simple or complex after analyzing the edge contour information;

[0064] When the contour corresponding to the target image is classified as simple, the CPU uses a contour similarity algorithm to determine the edge contour of the target image; when the contour corresponding to the target image is classified as complex, the CPU uses a skeleton detection algorithm to determine the edge contour of the target image;

[0065] Based on the judgment result, the edge contour recognition result of the target image is output.

[0066] The contour similarity algorithm specifically includes:

[0067] Obtaining a grayscale value of the target image;

[0068] Performing median filtering on the target image;

[0069] Performing binarization processing on the target image, the binarization threshold is a;

[0070] Extracting a maximum contour based on sub-pixels in the target image;

[0071] Calculating the similarity between the target image and the maximum contour;

[0072] Compare the calculated similarity with the preset similarity;

[0073] When the similarity between the target image and the maximum surrounding contour is greater than the preset similarity, the maximum surrounding contour is output as the edge contour of the target image; otherwise, a prompt indicating recognition failure is output.

[0074] And, the skeleton detection algorithm specifically includes:

[0075] Obtaining a grayscale value of the target image;

[0076] Performing median filtering on the target image;

[0077] Performing binarization processing on the target image, the binarization threshold is a;

[0078] Extracting a peripheral contour based on sub-pixels of the target image;

[0079] Performing concave contour extraction on the outer contour;

[0080] Calculating the similarity between the target image and the concave contour;

[0081] Compare the calculated similarity with the preset similarity;

[0082] When the similarity between the target image and the concave contour is greater than the preset similarity, the concave contour is output as the edge contour of the target image; otherwise, a prompt indicating recognition failure is output.

[0083] Based on the above method, the contour recognition method for the integrated AI intelligent image recognition sensor of this embodiment is configured with two edge contour recognition algorithms, which can distinguish scenes with different edge complexities, improve recognition accuracy, and thus adapt to different usage scenarios. At the same time, combined with the concave surface detection algorithm, it can remove unnecessary burrs, better reflect the outer contour features, and improve contour recognition accuracy.

[0084] Corresponding to the above-mentioned contour recognition method for an integrated AI intelligent image recognition sensor, the present embodiment discloses an integrated AI intelligent image recognition sensor (hereinafter referred to as the sensor). It can be understood that the contour recognition method of the sensor is the aforementioned contour recognition method for an integrated AI intelligent image recognition sensor. Specifically, the hardware composition of the sensor includes a DC / DC power supply, an MCU, a key module, a display module, a light source module, an input / output interface module, a focusable lens, a network connection module, a storage module, and the CPU. Among them, the DC / DC power supply, the MCU, the key module, the display module, the light source module, the input / output interface module, the focusable lens, the network connection module, and the storage module are all integrated on an MCU baseboard, the CPU is arranged on the CPU core board, the CPU communicates with the focusable lens via a USB interface, and the MCU baseboard is connected to the CPU core board through a board-to-board connection terminal. In this way, the MCU baseboard transfers the interface function of the CPU, and also integrates all functional interfaces detected by the sensor, thereby reducing the volume of the sensor. At the same time, the input and output logic functions of external communication control are integrated into the baseboard, and the input trigger signal is synchronously connected to the CPU core board and the MCU baseboard, so that the CPU can quickly capture camera images and the MCU can turn on the lighting in time. The synchronous response of the dual processing platform improves the system reaction speed and the imaging quality of the original image. In terms of software, the sensor integrates an embedded 3D GPU and a computing power NPU. It is feasible that the embedded 3D GPU uses the embedded ARM Mali G6103D GPU, and the computing power NPU can quickly calculate the deep AI model.

[0085] Looking back at the hardware part, the button module, the display module, the light source module, and the input / output interface module are respectively connected to the MCU control, the DC / DC power supply is connected to the DC20~28V power supply, and the MCU is also connected to a status indicator light for displaying the working status. In this embodiment, the CPU can be, but is not limited to, a domestic SOC based on the Cortex-A76+A55 architecture, an eight-core 64-bit CPU, with a main frequency of up to 2.4GHz, with super general computing performance, and can support the computing power requirements of contour recognition. The storage module includes a storage hard disk and a TF card storage unit. It is feasible that the storage hard disk can be, but is not limited to, a built-in 32 / 64G storage hard disk, supporting application data graphs for more than 3 months, and the TF card storage unit is a pluggable TF card slot and a corresponding TF card, so as to realize data backup at any time and quickly copy parameters to a new host for use. The light source module includes a built-in camera light source LED and an external light source interface, and the display module is a built-in OLED color display.

[0086] In the hardware design, the input voltage of the MCU baseboard is 24V, and the DC / DC power supply is configured with two power supply isolations, which is beneficial to improve the stability of the system and resist interference. Specifically, the DC / DC power supply includes: circuit structure and component composition such as Figure 3 The first power supply unit shown in the figure is used to output +4V power to power the CPU core board, and the circuit structure and components are as follows Figure 4 The second power supply unit shown in the figure is used to output +5V power to power the focusable lens, and the circuit structure and components are as follows Figure 5 The third power supply unit shown in FIG. 1 is used to output a +3.3V power supply to power the MCU, and the circuit structure and components are as shown in FIG. Figure 6 The fourth power supply unit shown is used to output +1.8V power to the power-on timing of the CPU core board. Among them, the power-on timing control pin of the CPU core board is as follows Figure 7 As shown, it is controlled by MCU. The external clock circuit of MCU core board is as follows Figure 8 As shown in the figure, the communication circuit between the CPU core board and the external I / O is as follows Fig. 9 As shown, secondly, the hardware design also arranges Fig.10 The EMC protection circuit shown in Figure 1 can realize power input protection. The MCU's IO signal communication / power input interface circuit is as follows: Fig.11 As shown, the MCU baseboard is configured as follows Fig.12 The HDMI connector shown in the figure can easily meet the needs of high-definition display.

[0087] In addition, in the hardware design, the input-output interface module includes three external input interfaces and one external output interface, wherein the pin connections corresponding to the three external input interfaces are shown as follows: Fig.13 As shown, the circuit of the external output interface is as follows Fig.14 As shown, the external output interface switches between NPN output and PNP output through the MCU control.

[0088] In this embodiment, the board-to-board connection terminal used by the sensor can be Figure 15-18 The pin connection diagram is shown below, where: Fig.15 The board-to-board socket J1 on the CPU mainboard is shown, which is used to connect the MCU core board. Fig.16 The board-to-board socket J2 on the CPU mainboard is shown, which is used to connect the MCU core board. Fig.17 The board-to-board socket J3 on the CPU mainboard is shown, which is used to connect the MCU core board. Fig.18 Shown is the board-to-board socket J4 on the CPU mainboard, which is used to connect the MCU core board.

[0089] In summary, the sensor disclosed in this embodiment adopts an integrated structure, with a small device size and less wiring. Compared with traditional machine vision equipment, it has obvious advantages in cost, size, installation, maintenance and stability. In addition, the application of the contour recognition method for the integrated AI intelligent image recognition sensor disclosed in this embodiment improves the accuracy of the contour recognition results.

[0090] In the embodiments provided in the present application, it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic units designed to implement the functions described herein or their combination. For software implementation, part or all of the flow of the embodiment can be completed by instructing the relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable storage medium or transmitted as one or more instructions or codes on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein the communication medium includes any medium that is convenient for transmitting a computer program from one place to another. The storage medium can be any available medium that a computer can access. The computer-readable storage medium can include but is not limited to RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of an instruction or data structure and can be accessed by a computer.

[0091] Finally, it should be noted that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A contour recognition method for an integrated AI intelligent image recognition sensor, characterized in that: The method comprises the following steps: According to the target selected by the user, the CPU automatically adjusts the exposure brightness and the binarization threshold a to take multiple photos of the target and generate several effect images; The user selects one of the generated renderings as a target image, and the CPU records the camera brightness value and the binarization threshold a corresponding to the target image; The user marks the template image with a rectangular frame, and the CPU automatically identifies the edge contour information of the target image based on the template image, and classifies the contour corresponding to the target image as simple or complex after analyzing the edge contour information; When the contour corresponding to the target image is classified as simple, the CPU uses a contour similarity algorithm to determine the edge contour of the target image; when the contour corresponding to the target image is classified as complex, the CPU uses a skeleton detection algorithm to determine the edge contour of the target image; Based on the judgment result, the edge contour recognition result of the target image is output.

2. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 1, characterized in that: The contour similarity algorithm specifically includes: Obtaining a grayscale value of the target image; Performing median filtering on the target image; Performing binarization processing on the target image; Extracting a maximum contour based on sub-pixels in the target image; Calculating the similarity between the target image and the maximum contour; Compare the calculated similarity with the preset similarity; When the similarity between the target image and the maximum surrounding contour is greater than a preset similarity, the maximum surrounding contour is output as the edge contour of the target image; otherwise, a prompt indicating recognition failure is output.

3. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 1, characterized in that: The skeleton detection algorithm specifically includes: Obtaining a grayscale value of the target image; Performing median filtering on the target image; Performing binarization processing on the target image; Extracting a peripheral contour based on sub-pixels of the target image; Performing concave contour extraction on the outer contour; Calculating the similarity between the target image and the concave contour; Compare the calculated similarity with the preset similarity; When the similarity between the target image and the concave contour is greater than a preset similarity, the concave contour is output as the edge contour of the target image; otherwise, a prompt indicating recognition failure is output.

4. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 2 or 3, characterized in that: When the target image is binarized, the binarization threshold used is a.

5. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 1, characterized in that: The integrated AI intelligent image recognition sensor includes a DC / DC power supply, an MCU, a key module, a display module, a light source module, an input / output interface module, a focusable lens, a network connection module, a storage module and the CPU; the DC / DC power supply, the MCU, the key module, the display module, the light source module, the input / output interface module, the focusable lens, the network connection module and the storage module are all integrated on an MCU baseboard, the CPU is arranged on a CPU core board, and the MCU baseboard is communicatively connected to the CPU core board through a board-to-board connection terminal; this integrated AI intelligent image recognition sensor integrates an embedded 3D GPU and a computing power NPU; The button module, the display module, the light source module, and the input / output interface module are respectively connected to the MCU control, the DC / DC power supply is connected to a DC20-28V power supply, and the MCU is also connected to a status indicator light for displaying the working status.

6. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 5, characterized in that: The storage module includes a storage hard disk and a TF card storage unit.

7. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 5, characterized in that: The DC / DC power supply includes a first power supply unit for outputting +4V power to power the CPU core board, a second power supply unit for outputting +5V power to power the focusable lens, a third power supply unit for outputting +3.3V power to power the MCU, and a fourth power supply unit for outputting +1.8V power to the CPU core board power-on timing.

8. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 7, characterized in that: The DC / DC power supply also includes an EMC protection circuit.

9. The contour recognition method for an integrated AI intelligent image recognition sensor according to claim 5, characterized in that: The input-output interface module includes three external input interfaces and one external output interface, wherein the external output interface switches between NPN output and PNP output through the control of the MCU.

Images