Contour recognition method and device, processing equipment, electronic equipment and storage medium

Abstract

The application discloses a contour recognition method and device, a processing device, electronic equipment and a storage medium. The contour recognition method comprises the following steps: acquiring a scene image, wherein the scene image comprises a target object; performing downsampling on the scene image to obtain a down-sampled image; performing semantic segmentation on the down-sampled image to obtain a semantic segmentation image; and performing guided up-sampling filtering on the semantic segmentation image based on the scene image to obtain a contour image containing a contour of the target object. In this way, the recognition efficiency and accuracy of the contour can be considered, and the accuracy of the contour recognition is also high under the condition of ensuring the recognition efficiency.

Description

Technical Field

[0001] This application belongs to the field of object recognition technology, and in particular relates to a contour recognition method, contour recognition device, processing equipment, electronic equipment and computer-readable storage medium. Background Technology

[0002] Recognizing object contours has important applications in laser engraving, die-cutting, and 3D printing tasks. By recognizing object contours through camera images, objects can be located and segmented. This technology is used in tasks such as object recognition, target localization, engraving pattern replication, automatic plating, and foreign object detection.

[0003] However, current contour recognition methods are either slow or have low accuracy, making it difficult to balance both. Summary of the Invention

[0004] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a contour recognition method, contour recognition device, processing equipment, electronic equipment, and computer-readable storage medium, which can balance recognition efficiency and recognition accuracy.

[0005] In a first aspect, this application provides a contour recognition method, comprising acquiring a scene image, wherein the scene image includes a target object; downsampling the scene image to obtain a downsampled image; performing semantic segmentation on the downsampled image to obtain a semantically segmented image; and performing guided upsampling filtering on the semantically segmented image based on the scene image to obtain a contour image containing the contour of the target object.

[0006] Secondly, this application provides a contour recognition device, which includes:

[0007] An acquisition module is used to acquire a scene image, wherein the scene image includes the target object;

[0008] The downsampling module is used to downsample the scene image to obtain a downsampled image;

[0009] A semantic segmentation module is used to perform semantic segmentation on the downsampled image to obtain a semantically segmented image;

[0010] The filtering module is used to perform guided upsampling filtering on the semantic segmentation image based on the scene image to obtain a contour image containing the contour of the target object.

[0011] Thirdly, this application provides a processing device, including a motor, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described contour recognition method.

[0012] Fourthly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described contour recognition method.

[0013] Fifthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described contour recognition method.

[0014] The contour recognition method, contour recognition device, processing equipment, electronic equipment, and computer-readable storage medium provided in this application acquire a scene image containing the target object to be recognized, then downsample the scene image to obtain a lower-resolution downsampled image, and then perform semantic segmentation on the downsampled image to reduce the computational load required for semantic segmentation and improve the efficiency of semantic segmentation, thereby quickly determining the semantics of each pixel in the downsampled image to obtain a semantically segmented image; then, based on the scene image containing complete contour information of the target object, guided upsampling filtering is performed to achieve upsampling while ensuring that the contour of the upsampled image is aligned with the scene image, thereby ensuring the accuracy of contour recognition, and the computational load required for guided upsampling filtering is low, with a low impact on recognition efficiency.

[0015] In this way, both the efficiency and accuracy of contour recognition can be taken into account, and the accuracy of contour recognition is also relatively high while ensuring recognition efficiency.

[0016] Additional aspects and advantages of embodiments of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of embodiments of this application. Attached Figure Description

[0017] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0018] Figure 1 This is a schematic diagram illustrating an application scenario of the contour recognition method provided in the embodiments of this application;

[0019] Figure 2 This is a first flowchart of the contour recognition method provided in the embodiments of this application;

[0020] Figure 3 This is a schematic diagram illustrating the principle of the contour recognition method provided in the embodiments of this application;

[0021] Figure 4 This is a second flowchart illustrating the contour recognition method provided in this application embodiment;

[0022] Figure 5This is a schematic diagram of the third process of the contour recognition method provided in the embodiments of this application;

[0023] Figure 6 This is a schematic diagram of the contour recognition device provided in the embodiments of this application;

[0024] Figure 7 This is a schematic diagram of the processing equipment provided in the embodiments of this application;

[0025] Figure 8 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application; and

[0026] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0027] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0028] To facilitate understanding, the technical background and application scenarios of this application will be introduced below:

[0029] 1. A 3D printer is a machine that can deposit material layer by layer according to a digital model to create a three-dimensional solid object. 3D printing technology (additive manufacturing technology) has developed rapidly in the past few decades and is widely used in many fields such as prototyping, manufacturing, medical, and education.

[0030] The following section details the components and working principle of 3D printing equipment.

[0031] (1) Components of 3D printing equipment

[0032] Printhead (Extruder) (including nozzles): responsible for heating and extruding printing consumables (such as plastic filaments, metal powder, etc.).

[0033] Motion System: Includes motion control mechanisms along the X, Y, and Z axes to ensure the printhead can move precisely to the designated position.

[0034] Control Unit: Typically controlled by an embedded system or computer, it is responsible for processing G-code (numerical control programming language) and controlling the coordinated operation of various components.

[0035] Heated Bed: Used to maintain the temperature of the printing platform, helping the first layer of material adhere better and reducing warping.

[0036] Frame: The structural foundation that supports all components and ensures the stability of the equipment.

[0037] Filament Feeder: Stores and supplies printing consumables.

[0038] (2) Working principle of 3D printing

[0039] Model design: Design a 3D model using CAD software.

[0040] Slicing: The 3D model is converted into a series of cross sections (slices), each slice representing a layer of material.

[0041] Print preparation: Convert the slice information into a machine-readable format (such as G-code) and load it into the 3D printer.

[0042] Material deposition: The print head deposits material layer by layer according to instructions. After each layer is completed, the platform descends a certain height, and this process is repeated until the model is completely printed.

[0043] 2. Engraving equipment is a tool used for engraving and processing various materials.

[0044] For example, engraving equipment includes mechanical engraving machines, laser engraving machines, jet engraving machines, and ultrasonic engraving machines. Engraving equipment uses contour recognition to accurately position the engraving area, thus assisting in the completion of the engraving process.

[0045] 3. Die-cutting equipment is an important piece of equipment for post-printing packaging processing and forming. It is also called a die-cutting machine, cutting machine, or CNC punching machine. It is mainly used for die-cutting, creasing, hot stamping, laminating, and automatic waste removal of some non-metallic materials, self-adhesive labels, EVA, double-sided tape, electronic and mobile phone pads, etc.

[0046] For example, die-cutting equipment includes rotary die-cutting machines, rotary-to-flat die-cutting machines, and flatbed die-cutting machines. Die-cutting equipment uses contour recognition to accurately position the die-cutting material, thus assisting in the completion of the die-cutting process.

[0047] Please see Figure 1 , Figure 1 This is an application scenario diagram of a contour recognition method provided in an embodiment of this application. The application scenario provided in this application includes a terminal device 101, a server 102, and a processing device 103. The contour recognition method provided in this application can be executed by at least one of the terminal device 101, the server 102, and the processing device 103.

[0048] The terminal devices may include, but are not limited to: smartphones (such as Android phones, iOS phones, etc.), tablets, laptops, desktop computers, smart speakers, smartwatches, portable personal computers, mobile internet devices (MIDs), smart voice interaction devices, smart home appliances, vehicle terminals, aircraft, wearable devices, etc., but this application does not limit them.

[0049] The terminal device may integrate a client, which can be a client with the function of displaying data information such as text, images, audio, and video, including but not limited to a device control client associated with the processing equipment, so as to realize communication and control with the processing equipment. This client can be a standalone client or an embedded sub-client integrated into a client (e.g., a social client), and there is no limitation on this.

[0050] The terminal device can communicate with the processing equipment to obtain scene images collected by the processing equipment, thereby realizing contour recognition.

[0051] The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. This application does not limit this.

[0052] The server can also communicate with the processing equipment to obtain scene images captured by the processing equipment, thereby achieving contour recognition.

[0053] The processing equipment can also integrate its own processing device and image acquisition device. The image acquisition device can acquire scene images collected by the processing equipment, and the processing device can realize contour recognition.

[0054] It should be noted that, Figure 1 The number of terminal devices and servers is for illustrative purposes only; the number of terminal devices and servers can be more or less, and there is no limitation herein. Terminal devices and servers can be connected directly or indirectly via wired or wireless communication, and this application does not impose any limitations on this.

[0055] The contour recognition method involved in this application can be implemented using cloud technology.

[0056] Cloud technology refers to a hosting technology that unifies a series of resources such as hardware, software, and networks within a wide area network or local area network to achieve data computing, storage, processing, and sharing.

[0057] Cloud technology is a collective term for network technologies, information technologies, integration technologies, management platform technologies, and application technologies applied to the cloud computing business model. It can form resource pools, providing flexible and convenient on-demand access. Cloud computing technology will become a crucial support. Backend services of technical network systems require substantial computing and storage resources, such as video websites, image websites, and many portal websites. With the rapid development and application of the internet industry, every item may have its own identification mark in the future, requiring transmission to backend systems for logical processing. Data at different levels will be processed separately, and various industry data will all require robust system support, which can only be achieved through cloud computing.

[0058] The contour recognition method in this application can be implemented based on cloud computing. Cloud computing is a computing model that distributes computing tasks across a resource pool composed of a large number of computers, enabling various application systems to obtain computing power, storage space, and information services as needed. The network providing these resources is called the "cloud." From the user's perspective, resources in the "cloud" are infinitely scalable, readily available, on-demand, expandable, and pay-as-you-go.

[0059] As a provider of fundamental cloud computing capabilities, a cloud resource pool (referred to as a cloud platform, generally called an IaaS (Infrastructure as a Service) platform) is established. Various types of virtual resources are deployed in the resource pool for external customers to choose from. The cloud resource pool mainly includes: computing devices (virtualized machines containing operating systems), storage devices, and network devices.

[0060] The contour recognition method in this application embodiment can be executed by an electronic device, which can be at least one of a server and a terminal device. That is, the method can be executed by the server or the terminal device alone, or by both the server and the terminal device. Therefore, the executing entity of each step will not be described again below.

[0061] It should be noted that the contour recognition method examples in the following text are based on the example of the processing equipment performing the contour recognition method on its own. Based on the understanding of the following text, those skilled in the art can apply the contour recognition method provided in the embodiments of this application to other types of scenarios (such as the terminal device or server performing the contour recognition method, or the terminal device cooperating with the server to jointly perform the contour recognition method, or the processing equipment cooperating with the terminal device to jointly perform the contour recognition method, or the processing equipment cooperating with the terminal device and the server to jointly perform the contour recognition method).

[0062] Recognizing object contours has important applications in laser engraving, die-cutting, and 3D printing tasks. By recognizing object contours through camera images, objects can be precisely located and segmented. This technology is used in tasks such as object recognition, target localization, engraved pattern replication, automatic plating, and foreign object detection.

[0063] Current methods for recognizing object contours mainly fall into the following categories:

[0064] a. Use general software tools (Halcon, OpenCV) for contour finding. The principle is to use classic computer vision algorithms (binarization, filtering, edge extraction, contour finding, etc.) to obtain contour information in the scene.

[0065] b. Reconstruct the object's surface using line laser scanning or multiple exposure methods with a projector, and distinguish the object by its shape or material.

[0066] c. Artificial neural networks can also be used to transform the problem into semantic segmentation and instance segmentation problems.

[0067] Current methods for recognizing object contours have the following problems:

[0068] a. Traditional computer vision methods are difficult to be stable and robust in complex scenes and are easily affected by interference. Laser engraving equipment using such methods usually has higher requirements for the object being engraved, lighting, background, object outline clarity, and imaging accuracy. The object being engraved needs to be significantly different from the foreground, making it difficult to adapt to complex working environments. When anomalies occur, user intervention is usually required for interaction and correction.

[0069] b. Methods using line lasers or projectors for fine reconstruction also place higher demands on system positioning accuracy, imaging, and lighting, and require the introduction of measurement equipment. To ensure measurement and reconstruction accuracy, the system typically needs to be calibrated and standardized before use, and the processing speed is slower.

[0070] c. Using artificial neural networks usually results in low recognition efficiency when the computing power of the device is limited, requiring the use of GPU devices with high computing power, which is costly.

[0071] Based on the above description, this application provides a contour recognition method, which will be described in detail below:

[0072] Please see Figure 2 and Figure 3 The contour recognition method provided in this application embodiment is implemented by steps 011 to 014, which are described in detail below.

[0073] Step 011: Obtain scene image P1, which includes the target object.

[0074] The target object is the object to be processed by the processing equipment.

[0075] Alternatively, the processing equipment can be a 3D printing device, an engraving device, or a die-cutting device. The object to be processed can be an object to be printed, an object to be engraved, or an object to be die-cut.

[0076] Optionally, the processing equipment includes a stage for carrying the object to be processed and an image acquisition device capable of acquiring images of the stage. The image acquisition device acquires images of the stage to obtain a scene image P1 containing the stage and the target object. The background object in the scene image P1 is the stage, and the foreground object is the target object.

[0077] Step 012: Downsample the scene image P1 to obtain the downsampled image P2.

[0078] Generally, the scene image P1 acquired by the image acquisition device has a high resolution and can contain more contour details of the target object. If contour recognition is performed directly on the scene image P1, such as using classic computer vision algorithms (binarization, filtering, edge extraction, contour finding, etc.) to obtain the contour information in the scene, the accuracy of contour information recognition is difficult to guarantee and the robustness is poor when the sampling conditions (such as the type of stage (e.g., background pattern), the type of object to be processed (e.g., complexity), the lighting conditions of the stage, and the type of debris generated during processing) change.

[0079] Therefore, this application uses a neural network model to achieve contour recognition. By adaptively setting the training data of the neural network model, it can be ensured that the neural network model adapts to various sampling conditions, thereby ensuring the accuracy of the neural network model in recognizing target objects in scene images P1 collected under different sampling conditions.

[0080] However, neural network models require a large amount of computation for contour recognition, and when the computing power of processing equipment is limited, the contour recognition efficiency of neural network models will become slow.

[0081] Therefore, in order to reduce the computational cost required for contour recognition by the neural network model, the scene image P1 can be downsampled to reduce the resolution of the scene image P1 while preserving the contour of the target object in the scene image P1.

[0082] For example, if the resolution of scene image P1 is 1000*1000, then by downsampling, a downsampled image P2 with a resolution of 100*100 can be obtained, which greatly reduces the resolution of the image that the neural network model needs to process, reduces the amount of computation required for the neural network model to perform contour recognition, and the neural network model for contour recognition of low-resolution images is not only easier to train, but also more efficient.

[0083] It is understandable that, based on the downsampling ratio setting, downsampling images P2 with different resolutions can be obtained. The higher the downsampling ratio, the lower the resolution of the downsampling image P2, and the lower the computational load required for contour recognition. However, the lower the resolution of the downsampling image P2, the less information it contains about the target object, and the lower the accuracy of the identified target object contour.

[0084] Therefore, by reasonably setting the downsampling factor, the accuracy of contour recognition can be improved as much as possible while reducing the computational load of contour recognition.

[0085] Step 013: Perform semantic segmentation on the downsampled image P2 to obtain the semantically segmented image P3.

[0086] Semantic segmentation is achieved through semantic segmentation networks. Semantic segmentation networks are neural network models in the field of computer vision used to classify each pixel in an image or video into a specific category.

[0087] By performing semantic segmentation on the downsampled image P2 using a semantic segmentation network, the classification of each pixel in the downsampled image P2 (such as whether it is a target object) can be obtained, thereby realizing the contour recognition of the target object.

[0088] Optionally, the semantic segmentation network includes at least one of instance segmentation network, foreground segmentation network, and interactive segmentation network. Depending on the actual needs, an appropriate semantic segmentation network can be selected.

[0089] Optionally, the training set for training the semantic segmentation network to convergence includes multiple training images. The training images are obtained by downsampling scene images P1 acquired under different sampling conditions. The sampling conditions include at least one of the following: lighting conditions, stage type, object type to be processed, and type of debris generated during processing.

[0090] In this way, the semantic segmentation network can accurately identify the contours of the scene image P1 sampled under different sampling conditions, ensuring the robustness of contour recognition.

[0091] Optionally, please refer to Figure 4 Step 013 includes:

[0092] Step 0131: Perform semantic segmentation on the downsampled image P2 to obtain the semantic segmentation results corresponding to each pixel of the downsampled image P2. The semantic segmentation results include at least one of whether it is a target object, background type, and object type.

[0093] Step 0132: Generate a semantic segmentation image P3 based on the semantic segmentation results corresponding to each pixel of the downsampled image P2.

[0094] Specifically, semantic segmentation of the downsampled image P2 is performed based on a semantic segmentation network, thus obtaining the semantic segmentation results corresponding to each pixel of the downsampled image P2. The semantic segmentation results can be the classification results of the object to which each pixel of the downsampled image P2 belongs, such as at least one of the following classification results: whether it is a target object, background type (such as the type of the stage), and object type (the type of the object to be processed).

[0095] Given the semantic segmentation results of the object to which each pixel of the downsampled image P2 belongs, the semantic segmentation image P3 can be generated based on the semantic segmentation results corresponding to each pixel of the downsampled image P2.

[0096] For example, pixels with the same semantic segmentation result in the downsampled image P2 are labeled with the same color, thereby quickly generating a semantic segmentation image P3 composed of different color blocks, each color block corresponding to an object (such as a stage, target object, etc.).

[0097] Optionally, since only the outline of the target object needs to be identified, in order to reduce the amount of computation required for identification, the semantic segmentation result only needs to include whether it is the target object, thereby improving the efficiency of semantic segmentation and quickly generating the semantic segmentation image P3.

[0098] Optionally, step 0132: Based on the semantic segmentation results corresponding to each pixel of the downsampled image P2, generate a semantic segmentation image P3, including:

[0099] Step 01321: Binarize the semantic segmentation results corresponding to each pixel of the downsampled image P2 to generate a semantic segmentation image P3.

[0100] Specifically, taking the semantic segmentation result as only including whether it is the target object as an example, the pixels whose semantic segmentation result is the target object are set to the first color (such as black and white), and the pixels whose semantic segmentation result is not the target object are set to the second color (such as black and white), thereby generating a binarized semantic segmentation image P3, which helps to reduce the difficulty of subsequent processing of the semantic segmentation image P3.

[0101] Step 014: Perform guided upsampling filtering on the semantic segmentation image P3 based on the scene image P1 to obtain a contour image P4 containing the contour of the target object.

[0102] Among them, guided upsampling filtering can be implemented based on a preset guided upsampling algorithm.

[0103] Alternatively, guided upsampling is a technique used in image or data processing to improve resolution or restore details. It uses additional guiding information to direct the upsampling process to obtain more accurate and higher-quality results. For example, guided upsampling can be an upsampling algorithm based on guided filtering.

[0104] Specifically, after obtaining the semantic segmentation image P3, since the semantic segmentation image P3 is obtained by semantic segmentation based on the lower resolution downsampled image P2, the contour detail of the target object in the semantic segmentation image P3 may not be high, and it cannot meet the accuracy requirements of contour recognition.

[0105] Therefore, by introducing a scene image P1 containing complete contour information of the target object and performing guided upsampling filtering, the contour of the upsampled image is aligned with the scene image P1 while achieving upsampling, thereby improving the accuracy of contour recognition. Furthermore, the computational cost required for guided upsampling filtering is relatively low, and its impact on recognition efficiency is also relatively low.

[0106] Therefore, by combining semantic segmentation networks and guided upsampling filtering, the accuracy of contour recognition is guaranteed while ensuring its robustness and efficiency, thus achieving a balance between efficiency and accuracy.

[0107] Optionally, please refer to Figure 4 Step 014 includes:

[0108] Step 0141: Upsample the semantic segmentation image P3 to obtain an upsampled image;

[0109] Step 0142: Perform guided filtering on the upsampled image based on the scene image P1 to obtain the contour image P4.

[0110] Specifically, when performing guided upsampling filtering, the semantic segmentation image P3 can be upsampled first to obtain an upsampled image.

[0111] Optionally, the resolution of the upsampled image can be the same as that of the scene image P1, thereby making full use of the contour information of the target object in the scene image P1.

[0112] It is understandable that guided filtering and upsampling can be performed asynchronously. That is, after upsampling, guided filtering is then applied to the upsampled image to obtain the contour image P4.

[0113] Optionally, guided filtering and upsampling can be performed simultaneously, that is, the edge information of the target object in the scene image P1 is used to guide the semantic segmentation image P3 to be upsampled, thereby obtaining the contour image P4.

[0114] For example, when guiding filtering is applied to a semantic segmentation image P3 based on a scene image P1 to obtain a contour image P4, the guided filtering uses a local linear model to guide the upsampling process of a low-resolution mask (such as the semantic segmentation image P3) using edge information from the original image. When the output image and the original image satisfy a linear relationship within a local window, the edge structure of the original image can be preserved while smoothing noise.

[0115] In some embodiments, please refer to Figure 5 Contour recognition methods also include:

[0116] Step 015: Display the contour image P4 and receive confirmation information, including whether the contour is accurate;

[0117] Step 016: If the contour is accurate, output the contour image P4;

[0118] Step 017: If the contour is inaccurate, perform contour recognition again.

[0119] Specifically, after contour recognition is completed, in order to further determine the accuracy of contour recognition, contour image P4 can be displayed (such as the display screen of the processing equipment or the display screen of the electronic device connected to the processing equipment). Then, the user can manually confirm it. When displaying contour image P4, a confirmation option (such as accurate or inaccurate) is also displayed. By selecting the confirmation option, the user can confirm the accuracy of contour image P4. After the user selects the confirmation option, the processing equipment receives the confirmation information, thereby determining whether contour image P4 is accurate.

[0120] If the contour image P4 is accurate, it can be directly output for subsequent processing (such as 3D printing, carving, die cutting, etc.). If the contour image P4 is inaccurate, contour recognition is performed again (such as re-executing steps 011 to 014) until the contour image P4 is confirmed to be accurate by the user.

[0121] In some embodiments, scene image P1 may include one or more images. One or more scene images P1 may be continuously acquired by an image acquisition device, or one or more scene images P1 may be acquired separately by one or more image acquisition devices (e.g., two image acquisition devices acquire two scene images P1 respectively).

[0122] Then, contour recognition is performed on one or more scene images P1 simultaneously to obtain contour images P4 corresponding to each scene image P1. When displaying contour images P4, contour images P4 corresponding to each scene image P1 can be displayed simultaneously. Then, based on the user's confirmation information (such as the confirmation information including the contour image P4 selected by the user), the final contour image P4 is output, thereby improving the recognition efficiency of contour images P4.

[0123] In some embodiments, after obtaining the contour image, further tasks such as object recognition (e.g., target object type recognition, stage recognition, etc.), fine positioning and segmentation of objects, automatic plating, and foreign object detection can be performed based on the contour image.

[0124] Based on the method described in the above embodiments, this application also provides a contour recognition device 300 for performing the steps in the above contour recognition method. Please refer to... Figure 6 , Figure 6 This is a schematic diagram of a contour recognition device 300 provided in an embodiment of this application. The contour recognition device 300 includes:

[0125] The acquisition module 301 is used to acquire a scene image, which includes the target object;

[0126] The downsampling module 302 is used to downsample the scene image to obtain a downsampled image;

[0127] The semantic segmentation module 303 is used to perform semantic segmentation on the downsampled image to obtain a semantically segmented image;

[0128] The filtering module 304 is used to perform guided upsampling filtering on the semantic segmentation image based on the scene image to obtain a contour image containing the contour of the target object.

[0129] It should be noted that the specific details of each module unit in the above contour recognition device have been described in detail in the embodiments of the above contour recognition method, and will not be repeated here.

[0130] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0131] In some embodiments, the contour recognition device in this application can be implemented in hardware, such as an electronic device or a component in an electronic device, such as an integrated circuit or a chip; the contour recognition device can also be implemented in software, such as as an application installed in an electronic device.

[0132] In some embodiments, please refer to Figure 7 The processing equipment 400 of this application includes a processor 402 and a memory 403. The memory 403 stores a computer program 404 that can run on the processor 402. When the processor 402 executes the program 404, it implements the various processes of the above-described contour recognition method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here.

[0133] Optionally, the processing equipment 400 is a 3D printing equipment. The print head of the 3D printing equipment includes a nozzle 405, a gripping wheel 406, and a feeding device 407. Driven by the motor 401, the gripping wheel 406 grips the printing material provided by the feeding device 407 and feeds it to the nozzle 405. The nozzle 405 is used to heat the printing material at a preset heating temperature and eject it.

[0134] In some embodiments, please refer to Figure 8 , Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device 500 includes a processor 501 and a memory 502. The memory 502 stores a computer program 503 that can run on the processor 501. When the processor 501 executes the program 503, it implements the various processes of the above-described contour recognition method embodiment and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0135] Please see Figure 9 , Figure 9This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this application. The electronic device can be a terminal or a server. Exemplarily, the electronic device 700 includes a central processing unit (CPU) 701, a system memory 704 including random access memory (RAM) 702 and read-only memory (ROM) 703, and a system bus 705 connecting the system memory 704 and the central processing unit 701.

[0136] In some embodiments, the electronic device 700 may also include a basic input / output system 706 that helps transmit information between various devices within the computer, and a mass storage device 707 for storing the operating system 713, the client 714, and other program modules 715.

[0137] In some embodiments, the basic input / output system 706 includes a display 708 for displaying information and an input device 709 for user input, such as a touch panel and other input devices. A touch panel is also called a touchscreen. A touch panel may include both a touch device and a touch controller. Other input devices may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be described further here.

[0138] Both the display 708 and the input device 709 are connected to the central processing unit 701 via an input / output controller 710 connected to the system bus 705. The basic input / output system 706 may also include the input / output controller 710 for receiving and processing input from touch panels, other input devices, etc. Similarly, the input / output system 706 also includes output devices such as displays, printers, or other types of output devices.

[0139] Mass storage device 707 is connected to central processing unit 701 via a mass storage controller (not shown) connected to system bus 705. Mass storage device 707 and its associated computer-readable media provide non-volatile storage for electronic device 700. That is, mass storage device 707 may include computer-readable media (not shown) such as hard disk or compact disc read-only memory (CD-ROM) drive.

[0140] According to various embodiments of this application, the electronic device 700 can also be connected to a remote computer on a network, such as the Internet. That is, the electronic device 700 can be connected to a network 717 via a network interface unit 716 connected to the system bus 705, or the network interface unit 716 can be used to connect to other types of networks or remote computer systems (not shown).

[0141] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described contour recognition method embodiments and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0142] The processor can be the processor in the electronic device described in the above embodiments. The computer-readable storage medium can be a computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, etc.

[0143] Computer-readable media can include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented using any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer storage media include RAM, ROM, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other solid-state storage technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape cassettes, magnetic tape, disk storage, or other magnetic storage devices. Of course, those skilled in the art will recognize that computer storage media are not limited to the above-mentioned types.

[0144] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described contour recognition method. The processor may be a processor in the electronic device described in the above embodiments. When executed by the processor, the computer program implements the various processes of the embodiments of the contour recognition method and achieves the same technical effects; therefore, to avoid repetition, further details are omitted here.

[0145] It is understood that in the specific implementation of this application, data related to user identity or characteristics is involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

Claims

1. A contour recognition method, characterized in that, include: Acquire a scene image, wherein the scene image includes the target object; The scene image is downsampled to obtain a downsampled image; The downsampled image is semantically segmented to obtain a semantically segmented image; Guided upsampling filtering is performed on the semantic segmentation image based on the scene image to obtain a contour image containing the contour of the target object.

2. The contour recognition method according to claim 1, characterized in that, The step of performing semantic segmentation on the downsampled image to obtain a semantically segmented image includes: Based on a preset semantic segmentation network, the downsampled image is semantically segmented to obtain a semantically segmented image.

3. The contour recognition method according to claim 1, characterized in that, The method of performing semantic segmentation on the downsampled image based on a preset semantic segmentation network to obtain a semantically segmented image includes: The downsampled image is semantically segmented to obtain semantic segmentation results corresponding to each pixel of the downsampled image. The semantic segmentation results include at least one of whether it is a target object, background type, and object type. The semantic segmentation image is generated based on the semantic segmentation results corresponding to each pixel of the downsampled image.

4. The contour recognition method according to claim 3, characterized in that, The step of generating the semantic segmentation image based on the semantic segmentation results corresponding to each pixel of the downsampled image includes: The semantic segmentation image is generated by binarizing the semantic segmentation results corresponding to each pixel of the downsampled image.

5. The contour recognition method according to claim 2, characterized in that, The semantic segmentation network includes at least one of instance segmentation network, foreground segmentation network, and interactive segmentation network.

6. The contour recognition method according to claim 2, characterized in that, The training set for training the semantic segmentation network to convergence includes multiple training images. The training images are obtained by downsampling scene images acquired under different sampling conditions. The sampling conditions include at least one of the following: lighting conditions, stage type, object type to be processed, and type of debris generated during processing.

7. The contour recognition method according to claim 1, characterized in that, The guided upsampling filtering of the semantic segmentation image based on the scene image includes: Based on a preset guided upsampling algorithm, the scene image is subjected to guided upsampling filtering on the semantic segmentation image.

8. The contour recognition method according to claim 7, characterized in that, The pre-defined guided upsampling algorithm performs guided upsampling filtering on the semantic segmentation image using the scene image, including: The semantic segmentation image is upsampled to obtain an upsampled image; The upsampled image is guided by filtering based on the scene image to obtain the contour image.

9. The contour recognition method according to claim 1, characterized in that, The scene image also includes a background object, which includes a stage, and the target object is located on the stage.

10. The contour recognition method according to claim 1, characterized in that, Also includes: Display the outline image and receive confirmation information, including whether the outline is accurate; If the contour is accurate, output the contour image; If the contour is inaccurate, contour recognition is performed again.

11. A contour recognition device, characterized in that, include: An acquisition module is used to acquire a scene image, wherein the scene image includes the target object; The downsampling module is used to downsample the scene image to obtain a downsampled image; A semantic segmentation module is used to perform semantic segmentation on the downsampled image to obtain a semantically segmented image; The filtering module is used to perform guided upsampling filtering on the semantic segmentation image based on the scene image to obtain a contour image containing the contour of the target object.

12. A processing equipment, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the contour recognition method as described in any one of claims 1-10.

13. The processing equipment according to claim 12, characterized in that, The processing equipment includes 3D printing equipment, engraving equipment, or die-cutting equipment.

14. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the contour recognition method as described in any one of claims 1-10.

15. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the contour recognition method as described in any one of claims 1-10.

Images