Target three-dimensional reconstruction method and system based on omnibearing multiple views
By adopting target suspension and automatic rotation technology in the three-dimensional reconstruction system, combining stepper motors and image automatic shooting modules, the problems of low automation and incomplete reconstruction in the existing technology are solved, and efficient and comprehensive three-dimensional reconstruction of the fruit is achieved. .
Patent Information
- Application Number
- CN202410993708.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-11-19
AI Technical Summary
The existing three-dimensional reconstruction system has problems such as low automation, high cost and slow efficiency in fruit target reconstruction, especially the inability to collect fruit surface images in all aspects, resulting in incomplete reconstruction.
Adopt the target three-dimensional reconstruction method based on all-round multi-views, and realizes automatic image acquisition through target suspension and automatic rotation. Combined with the stepper motor timing rotation and image automatic shooting module, it automatically collects multi-angle visual images, avoid manual operations and ensures the full surface of the fruit. Position shooting.
It improves the automation level of the three-dimensional reconstruction system, reduces cost and time, and realizes all-round, automated three-dimensional reconstruction of the fruit, ensuring the integrity and accuracy of the reconstruction.
Smart Images

Figure CN118982618A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of target three-dimensional reconstruction, and in particular to a target three-dimensional reconstruction method and system based on omnidirectional multi-views. Background Art
[0002] With the continuous development of computer vision technology, 3D reconstruction has become increasingly mature and widely used, especially in modern agriculture. The 3D reconstruction system using 3D reconstruction technology can be used to reconstruct fruit targets and generate point cloud or 3D model data, solving the problem that traditional 2D image methods cannot obtain all-round fruit phenotypic information. Based on the 3D data of fruits, the morphological characteristics of crops can be obtained, which can be used for fruit industry construction and management, planting decision support, agricultural scientific research and innovation, and agricultural product marketing and display.
[0003] Existing methods for 3D reconstruction include: Currently used 3D reconstruction systems are mainly based on two methods: 3D scanning imaging based on LiDAR and 2D multi-view reconstruction imaging. LiDAR is expensive and difficult to promote and deploy effectively. Therefore, most research focuses on system design based on 2D multi-view reconstruction methods.
[0004] The principle of multi-view reconstruction includes: 2D multi-view reconstruction refers to recovering the 3D representation of an object or scene from a 2D image sequence. It is mainly divided into two steps: multi-view acquisition (using a camera to take multi-view images of the same target); 3D modeling (using relevant modeling software to extract image features, match and calculate the relationship between the camera's image coordinate system and the world coordinate system, and finally reconstruct the target's 3D model or point cloud).
[0005] There are problems with existing multi-view reconstruction methods:
[0006] ① For the reconstruction of fruit objects, most existing methods place the fruit on a platform for reconstruction, which makes it impossible to capture the bottom of the fruit and thus to reconstruct the entire surface of the fruit;
[0007] ② The existing 3D reconstruction system has a low degree of automation and often requires complicated manual operations. For example, when acquiring multiple views, a person needs to hold a camera and take images from various angles around the fruit; and after acquiring multiple views, the subsequent reconstruction software needs to be debugged and used. The steps are cumbersome, resulting in the entire process being costly, slow, and inefficient. Summary of the invention
[0008] In order to solve the problems existing in the prior art, a method and system for target three-dimensional reconstruction based on omnidirectional multi-view is provided, which realizes automatic omnidirectional image acquisition of the target through target suspension and automatic rotation of the target. At the same time, omnidirectional image acquisition of the specific target is realized through specific target hanging spikes, avoiding the problem of incomplete reconstruction of the target surface caused by failure to capture the bottom of the target due to placement angle problems.
[0009] On the one hand, the present invention provides a method for 3D reconstruction of a target based on omnidirectional multi-views, comprising:
[0010] S1, select the target for 3D reconstruction;
[0011] S2, automatically shooting multi-angle images based on a target multi-view acquisition program, thereby performing multi-camera multi-angle visual image sampling to obtain multiple views of the target;
[0012] S3, performing three-dimensional reconstruction based on the multiple views of the target.
[0013] Preferably, the target multi-view acquisition procedure includes:
[0014] S21, stepper motor timing rotation subroutine; and
[0015] S22, automatic image shooting subroutine.
[0016] Preferably, the stepper motor timed rotation subroutine is implemented based on a driving module, and the driving module is a stepper motor timed rotation module; the stepper motor timed rotation module is used to control the target to rotate at a uniform speed, and the stepper motor timed rotation module includes a stepper motor, a stepper motor driver and an Arduino development board; the program in the stepper motor timed rotation module is used to control the stepper motor to perform normally according to the set working mode; the set working mode includes speed and direction.
[0017] Preferably, the stepper motor timing rotation subroutine is used to control the stepper motor to perform normally according to the set working mode, including:
[0018] (1) Connecting the reconstruction target to the suspension device;
[0019] (2) When the hardware components corresponding to the stepper motor timing rotation module are powered on, the Arduino development board is connected to the stepper motor driver, and a pulse signal is provided to the stepper motor driver according to the programmed program;
[0020] (3) The stepper motor driver drives the stepper motor to operate according to the received pulse signal; the stepper motor is driven by the stepper motor driver to drive the suspension device to rotate at a constant speed in a set direction, and the control target rotates at a constant speed at a fixed time;
[0021] (4) All hardware components of the stepper motor timing rotation module are powered off and stop running;
[0022] The automatic image shooting subroutine is implemented through an automatic image shooting module and a computer control module, including a computer control module software unit and an automatic image shooting module software unit, wherein the computer control module software unit is used to implement the relevant control functions of the computer on the automatic image shooting module, and the automatic image shooting module software unit is used to implement the function of the Raspberry Pi calling the camera to take pictures; the automatic image shooting module debugs and programs the Raspberry Pi, so that the Raspberry Pi controls the camera to take pictures at a designed time interval, and uses cameras located at different pitch angles to take pictures of the target in uniform rotation at equal time intervals; the automatic image shooting module includes a Raspberry Pi and a camera; the computer control module includes a computer, which is used to control the camera to take pictures at a designed time interval.
[0023] Preferably, the workflow of the computer control module software unit includes:
[0024] (1) The computer is started and connected to the Internet, and the automatic image shooting module is remotely connected;
[0025] (2) The computer sets corresponding shooting parameters, such as shooting time interval, number of shots, etc., and sends shooting instructions;
[0026] (3) After confirming that the photo is taken, send an image transmission command and use the remote control function to control the Raspberry Pi to start transmitting the image;
[0027] (4) Receive the image sent back by the Raspberry Pi through the file transfer function;
[0028] (5) Receive the return end information sent by the Raspberry Pi through the return information function and confirm that the image has been received;
[0029] (6) Turn off the remote control function and shut down the computer.
[0030] Preferably, the automatic image shooting module software unit performs all-round image acquisition of the target by target suspension and target automatic rotation, and realizes all-round image acquisition of the specific target by specific target suspension. The workflow of the automatic image shooting module software unit includes:
[0031] (1) The Raspberry Pi is started and connected to the Internet, and the computer remote control program for the Raspberry Pi is started;
[0032] (2) The computer sets the corresponding shooting parameters and shooting instructions and sends them to the Raspberry Pi via remote control;
[0033] (3) The Raspberry Pi controls the camera to shoot according to the specified parameters, runs the photo-taking program and saves the acquired images;
[0034] (4) After the camera takes the photo, the Raspberry Pi receives the image transmission command and transmits the captured image back to the computer through the remote control function;
[0035] (5) After confirming that the image is uploaded, a message indicating the end of the upload is sent and the Raspberry Pi is shut down.
[0036] Preferably, the S3 performs three-dimensional reconstruction based on the multiple views of the target by writing a script of the automated operation software based on the reconstruction software to form a three-dimensional modeling program, and runs the three-dimensional modeling program based on the computer control module to achieve the multiple views of the target. Figure 3 3D reconstruction function; wherein the 3D modeling part in the computer control module is based on the reconstruction software, a one-key operation reconstruction software is written, a script for executing multiple reconstruction steps is executed, and a multi-view reconstruction of the fruit target is realized; the 3D modeling program includes:
[0037] (1) Add a multi-view image sequence of the reconstruction target and automatically run the next step;
[0038] (2) registering the multi-view image sequence and generating a sparse point cloud of the target, and judging whether the registration is successful based on the registration algorithm, judging whether the sparse point cloud generation is successful, and automatically running the next step if successful, and recording and feeding back the problem if failed, analyzing, modifying the 3D modeling program or adjusting the shooting plan;
[0039] (3) generating a dense point cloud based on the target sparse point cloud, and asking the user whether to generate a three-dimensional model of the target or directly export the dense point cloud;
[0040] (4) If the user determines to generate a three-dimensional model of the target, a three-dimensional model and texture are generated based on the dense point cloud, and the dense point cloud and the three-dimensional model are exported.
[0041] A second aspect of the present invention is to provide a target three-dimensional reconstruction system based on omnidirectional multi-views, comprising:
[0042] A target selection module, used for selecting a target for three-dimensional reconstruction;
[0043] A multi-view acquisition module, used to automatically shoot multi-angle images based on a target multi-view acquisition program, thereby sampling visual images from multiple camera positions and multiple angles to acquire multiple views of the target;
[0044] The three-dimensional reconstruction module is used to perform three-dimensional reconstruction based on multiple views of the target.
[0045] A third aspect of the present invention provides an electronic device, comprising a processor and a memory, wherein the memory stores a plurality of instructions, and the processor is configured to read the instructions and execute the method described in the first aspect.
[0046] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a plurality of instructions, and the plurality of instructions can be read by a processor and execute the method described in the first aspect.
[0047] The method, system, electronic device, and computer-readable storage medium provided by the present invention have the following beneficial technical effects:
[0048] (1) The three-dimensional reconstruction process based on the designed reconstruction system includes two main steps: obtaining multiple views of the reconstructed target and using modeling software to perform three-dimensional modeling; the method and system implement the process-based, automated, and convenient execution of some related operations involved in these two main steps, including the control program design of two hardware modules and the automatic execution script design of a modeling software; automatic all-round image acquisition of the target is achieved through target suspension and automatic target rotation; at the same time, all-round image acquisition of the specific target is achieved through specific target suspension spikes, avoiding the problem of incomplete reconstruction of the specific target surface due to the inability to capture the bottom of the specific target due to placement.
[0049] (2) Improve the automation level of the 3D reconstruction system, eliminating the need for complicated manual operations and processing steps, thereby reducing the cost of the entire process and improving processing speed and efficiency.
[0050] (3) For the multi-view acquisition part of the reconstructed target, the system can automatically sample the visual images of the target from multiple positions and angles, without the need for manual operation to hold the camera and adjust the angle for shooting.
[0051] (4) For the 3D modeling part, the system can automatically use the sampled images and call the 3D reconstruction software to perform high-precision 3D modeling, thus achieving all-round and automated 3D reconstruction of the fruit. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 This is a flow chart of the target three-dimensional reconstruction method based on omnidirectional multi-views according to the present invention;
[0053] Figure 2 This is a software flow diagram of the stepping motor timing rotation module of the present invention;
[0054] Figure 3 A schematic diagram of the workflow of the computer control module software unit of the present invention;
[0055] Figure 4It is a schematic diagram of the three-dimensional modeling program flow of the present invention;
[0056] FIG5( a ) shows the set image shooting angle and phase position of the three-dimensional imaging of grape fruit according to the present invention;
[0057] FIG5( b ) shows the omnidirectional image of the fruit obtained according to the present invention;
[0058] FIG5(c) shows the final three-dimensional model of the fruit according to the present invention:
[0059] Figure 6 It is a schematic diagram of the structure of the target three-dimensional reconstruction system based on omnidirectional multi-view according to the present invention;
[0060] Figure 7 It is a schematic diagram of the structure of the electronic device described in the present invention. DETAILED DESCRIPTION
[0061] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.
[0062] The method provided by the present invention can be implemented in the following terminal environment, and the terminal may include one or more of the following components: a processor, a memory, and a display screen. The memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the method described in the following embodiment.
[0063] The processor may include one or more processing cores. The processor uses various interfaces and lines to connect various parts of the entire terminal, and executes various functions of the terminal and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory, and calling data stored in the memory.
[0064] The memory may include a random access memory (RAM) or a read-only memory (ROM). The memory may be used to store instructions, programs, codes, code sets or instructions.
[0065] The display screen is used to display the user interface of each application.
[0066] In addition, those skilled in the art can understand that the structure of the above terminal does not constitute a limitation on the terminal, and the terminal may include more or fewer components, or combine certain components, or arrange the components differently. For example, the terminal also includes components such as a radio frequency circuit, an input unit, a sensor, an audio circuit, and a power supply, which will not be described in detail here.
[0067] Embodiment 1
[0068] like Figure 1 As shown, this embodiment provides an all-round automated target three-dimensional reconstruction method, including:
[0069] S1, select the target for 3D reconstruction;
[0070] In this embodiment, the reconstruction fruit target will be selected.
[0071] S2, automatically shooting multi-angle images based on a target multi-view acquisition program, thereby performing multi-camera multi-angle visual image sampling to obtain multiple views of the target;
[0072] S3, performing three-dimensional reconstruction based on the multiple views of the target.
[0073] As a preferred implementation, the target multi-view acquisition procedure includes:
[0074] S21, stepper motor timing rotation subroutine; and
[0075] S22, automatic image shooting subroutine.
[0076] In this embodiment, the stepper motor timed rotation subroutine is implemented as a burned program, which is mainly for the programming implementation of the uniform speed rotation function, involving the initialization of the development board pin function, setting the rotation direction, speed, etc.
[0077] In this embodiment, the target hanging module has multiple modes, including a large target hanging mode for hanging larger targets (including a paddle and a hook, wherein the paddle is connected to the hook and drives the hook to rotate, and the hook is used to hook and fix the large target) and a small target hanging mode for hanging smaller targets (including a needle disk and a turntable, wherein the needle disk is used to pierce and fix the small target, and the needle disk is distributed on the turntable).
[0078] The target (fruit) rotates automatically to achieve automatic all-round image acquisition of the target (fruit); at the same time, through the specific target (fruit grain) hanging thorns, all-round image acquisition of the specific target (fruit grain) is achieved, avoiding the problem of incomplete surface reconstruction of the target (fruit or fruit grain) due to the inability to capture the bottom of the target (fruit or fruit grain) due to placement. Among them, fruit hanging is used for the reconstruction of large fruits that can be hung, such as apples, pears, citrus, and grape bunches; fruit grain hanging thorns are used for the reconstruction of smaller fruit grains such as grape grains.
[0079] like Figure 2As shown, as a preferred implementation, the stepper motor timing rotation subroutine is implemented based on the driving module. The driving module of this embodiment is a stepper motor timing rotation module; the stepper motor timing rotation module is used to control the target to rotate at a constant speed, and the stepper motor timing rotation module includes a stepper motor, a stepper motor driver and an Arduino development board; the program in the stepper motor timing rotation module is used to control the stepper motor to perform normally according to the set working mode; the set working mode includes speed and direction; the stepper motor timing rotation subroutine in the stepper motor timing rotation module is used to control the stepper motor to perform normally according to the set working mode and includes:
[0080] (1) Connecting the reconstruction target to the suspension device;
[0081] (2) When the hardware components corresponding to the stepper motor timing rotation module are powered on, the Arduino Uno development board is connected to the stepper motor driver, and a pulse signal is provided to the stepper motor driver according to the programmed program;
[0082] (3) The stepper motor driver drives the stepper motor to operate according to the received pulse signal; the stepper motor is driven by the stepper motor driver to drive the suspension device to rotate at a constant speed in a set direction, and the control target rotates at a constant speed at a fixed time;
[0083] (4) All hardware components of the stepper motor timing rotation module are powered off and stop running.
[0084] In this embodiment, for example, the pseudo code for implementing the forward and reverse rotation of the motor using Arduino IDE includes:
[0085]
[0086]
[0087]
[0088]
[0089] As a preferred embodiment, the automatic image shooting subroutine is implemented by an automatic image shooting module and a computer control module, including a computer control module software unit and an automatic image shooting module software unit, wherein the computer control module software unit is used to realize the relevant control functions of the computer to the automatic image shooting module, and the automatic image shooting module software unit is used to realize the function of the Raspberry Pi calling the camera to take pictures; the automatic image shooting module debugs and programs the Raspberry Pi, so that the Raspberry Pi controls the camera to take pictures at a designed time interval, and uses cameras located at different pitch angles to take pictures of the target in uniform rotation at equal time intervals; the automatic image shooting module includes a Raspberry Pi and a camera; the computer control module includes a computer, which is used to control the camera to take pictures at a designed time interval.
[0090] like Figure 3 As shown, as a preferred embodiment, the workflow of the computer control module software unit includes:
[0091] (1) The computer is started and connected to the Internet, and the automatic image shooting module is remotely connected;
[0092] (2) The computer sets corresponding shooting parameters, such as shooting time interval, number of shots, etc., and sends shooting instructions;
[0093] (3) After confirming that the photo is taken, send an image transmission command and use the remote control function to control the Raspberry Pi to start transmitting the image;
[0094] (4) Receive the image sent back by the Raspberry Pi through the file transfer function;
[0095] (5) Receive the return end information sent by the Raspberry Pi through the return information function and confirm that the image has been received;
[0096] (6) Turn off the remote control function and shut down the computer.
[0097] As a preferred implementation, the workflow of the automatic image capture module software unit includes:
[0098] (1) The Raspberry Pi is started and connected to the Internet, and the computer remote control program for the Raspberry Pi is started;
[0099] (2) The computer sets the corresponding shooting parameters and shooting instructions and sends them to the Raspberry Pi via remote control;
[0100] (3) The Raspberry Pi controls the camera to shoot according to the specified parameters, runs the photo-taking program and saves the acquired images;
[0101] (4) After the camera takes the photo, the Raspberry Pi receives the image transmission command and transmits the captured image back to the computer through the remote control function;
[0102] (5) After confirming that the image is uploaded, a message indicating the end of the upload is sent and the Raspberry Pi is shut down.
[0103] As a preferred embodiment, the startup of the automatic image shooting module and the feedback of the results obtained by automatic shooting of multi-angle images after target fixation and target timed rotation are realized based on computer programming of the computer control module, wherein the computer control module is connected to the automatic image shooting module.
[0104] As a preferred embodiment, the S3 performs three-dimensional reconstruction based on multiple views of the target by writing a script of the automated operation software based on the reconstruction software to form a three-dimensional modeling program, and running the three-dimensional modeling program based on the computer control module to achieve the multiple views of the target. Figure 3 Reconstruction function.
[0105] In this embodiment, the 3D modeling part in the computer control module is based on the reconstruction software, and a script is written to operate the reconstruction software with one key, execute multiple reconstruction steps, and realize the multi-view reconstruction of the fruit target. Figure 4 The figure shows the Agisoff Metashape reconstruction software as an example, and uses Python to write scripts to automate the operation process in the reconstruction software.
[0106] like Figure 4 As shown, as a preferred embodiment, the three-dimensional modeling program includes:
[0107] (1) Add a multi-view image sequence of the reconstruction target and automatically run the next step;
[0108] (2) Align the multi-view image sequence and generate a sparse point cloud of the target. At the same time, based on the registration algorithm, determine whether the registration is successful and whether the sparse point cloud is generated successfully. If successful, the next step will be automatically executed. If failed, the problem will be recorded and fed back for analysis, and the 3D modeling program will be modified or the shooting plan will be adjusted. Among them, the registration algorithm is responsible for extracting the associated feature points between the images, thereby restoring the 3D structure from the disordered 2D image to generate the target point cloud. The specific algorithm can be any open source image registration algorithm such as the Scale-invariant feature transform (SIFT) algorithm, the Speeded Up Robust Features (SURF) algorithm, and the Features from Accelerated Segment Test (FAST) algorithm.
[0109] (3) generating a dense point cloud based on the sparse point cloud of the target, and asking the user whether to generate a three-dimensional model of the target or directly export the dense point cloud;
[0110] (4) If the user determines to generate a three-dimensional model of the target, a three-dimensional model and texture are generated based on the dense point cloud, and the dense point cloud and the three-dimensional model are exported.
[0111] Application Examples
[0112] As shown in Figure 5(a), the image shooting angle and phase pose are set for the three-dimensional imaging of grapes. Figure 5(b) shows the acquired omnidirectional image of the fruit. It can be seen that the bottom of the grape is not blocked. Figure 5(c) shows the final three-dimensional model of the fruit.
[0113] Embodiment 2
[0114] like Figure 6 As shown, this embodiment provides an all-round automated target three-dimensional reconstruction system, including: a target selection module 101, used to select a target for three-dimensional reconstruction; a multi-view acquisition module 102, used to automatically shoot multi-angle images based on a target multi-view acquisition program, thereby performing multi-camera multi-angle visual image sampling to obtain multiple views of the target; a three-dimensional reconstruction module 103, used to perform three-dimensional reconstruction based on the multiple views of the target.
[0115] The present invention also provides a memory storing a plurality of instructions, wherein the instructions are used to implement the method described in the first embodiment.
[0116] like Figure 7 As shown, the present invention also provides an electronic device, including a processor 301 and a memory 302 connected to the processor 301, wherein the memory 302 stores a plurality of instructions, which can be loaded and executed by the processor so that the processor can execute the method described in Example 1.
[0117] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A method for 3D reconstruction of a target based on omnidirectional multi-views, characterized in that: include: S1, select the target for 3D reconstruction; S2, automatically shooting multi-angle images based on a target multi-view acquisition program, thereby performing multi-camera multi-angle visual image sampling to obtain multiple views of the target; S3, performing three-dimensional reconstruction based on the multiple views of the target.
2. The method for 3D reconstruction of a target based on omnidirectional multi-view according to claim 1, characterized in that: The target multi-view acquisition procedure includes: S21, stepper motor timing rotation subroutine; and S22, automatic image shooting subroutine.
3. The method for 3D reconstruction of a target based on omnidirectional multi-view according to claim 2, characterized in that: The stepper motor timed rotation subroutine is implemented based on a driving module, and the driving module is a stepper motor timed rotation module; the stepper motor timed rotation module is used to control the target to rotate at a uniform speed, and the stepper motor timed rotation module includes a stepper motor, a stepper motor driver and an Arduino development board; the program in the stepper motor timed rotation module is used to control the stepper motor to perform normally according to the set working mode; the set working mode includes speed and direction.
4. The method for 3D reconstruction of a target based on omnidirectional multi-view according to claim 3, characterized in that: The stepper motor timing rotation subroutine is used to control the stepper motor to perform normally according to the set working mode, including: (1) Connecting the reconstruction target to the suspension device; (2) When the hardware components corresponding to the stepper motor timing rotation module are powered on, the Arduino development board is connected to the stepper motor driver, and a pulse signal is provided to the stepper motor driver according to the programmed program; (3) The stepper motor driver drives the stepper motor to operate according to the received pulse signal; the stepper motor is driven by the stepper motor driver to drive the suspension device to rotate at a constant speed in a set direction, and the control target rotates at a constant speed at a fixed time; (4) All hardware components of the stepper motor timing rotation module are powered off and stop running; The automatic image shooting subroutine is implemented through an automatic image shooting module and a computer control module, including a computer control module software unit and an automatic image shooting module software unit, wherein the computer control module software unit is used to implement the relevant control functions of the computer on the automatic image shooting module, and the automatic image shooting module software unit is used to implement the function of the Raspberry Pi calling the camera to take pictures; the automatic image shooting module debugs and programs the Raspberry Pi, so that the Raspberry Pi controls the camera to take pictures at a designed time interval, and uses cameras located at different pitch angles to take pictures of the target in uniform rotation at equal time intervals; the automatic image shooting module includes a Raspberry Pi and a camera; the computer control module includes a computer, which is used to control the camera to take pictures at a designed time interval.
5. The method for 3D reconstruction of an object based on omnidirectional multi-view according to claim 4, characterized in that: The workflow of the computer control module software unit includes: (1) The computer is started and connected to the Internet, and the automatic image shooting module is remotely connected; (2) The computer sets corresponding shooting parameters and sends shooting instructions; (3) After confirming that the photo is taken, send an image transmission command and use the remote control function to control the Raspberry Pi to start transmitting the image; (4) Receive the image sent back by the Raspberry Pi through the file transfer function; (5) Receive the return end message sent by the Raspberry Pi through the return message function and confirm that the image has been received; (6) Turn off the remote control function and shut down the computer.
6. The method for 3D reconstruction of an object based on omnidirectional multi-view according to claim 5, characterized in that: The automatic image shooting module software unit performs all-round image acquisition of the target through target suspension and automatic rotation of the target, and realizes all-round image acquisition of the specific target through specific target suspension. The workflow of the automatic image shooting module software unit includes: (1) The Raspberry Pi is started and connected to the Internet, and the computer remote control program for the Raspberry Pi is started; (2) The computer sets the corresponding shooting parameters and shooting instructions and sends them to the Raspberry Pi via remote control; (3) The Raspberry Pi controls the camera to shoot according to the specified parameters, runs the photo-taking program and saves the acquired images; (4) After the camera takes the photo, the Raspberry Pi receives the image transmission command and transmits the captured image back to the computer through the remote control function; (5) After confirming that the image is uploaded, a message indicating the end of the upload is sent and the Raspberry Pi is shut down.
7. The method for 3D reconstruction of an object based on omnidirectional multi-views according to claim 6, characterized in that: The S3 performs 3D reconstruction based on multiple views of the target by writing a script of the automated operation software based on the reconstruction software to form a 3D modeling program, and running the 3D modeling program based on the computer control module to achieve the multi-view 3D reconstruction function of the target; wherein the 3D modeling part in the computer control module is based on the reconstruction software, and a one-key operation reconstruction software is written to execute a script of multiple reconstruction steps to achieve the multi-view reconstruction of the target; the 3D modeling program includes: (1) Add a multi-view image sequence of the reconstruction target and automatically run the next step; (2) registering the multi-view image sequence and generating a sparse point cloud of the target, and judging whether the registration is successful based on the registration algorithm, judging whether the sparse point cloud generation is successful, and automatically running the next step if successful, and recording and feeding back the problem if failed, analyzing, modifying the 3D modeling program or adjusting the shooting plan; (3) generating a dense point cloud based on the sparse point cloud of the target, and asking the user whether to generate a three-dimensional model of the target or directly export the dense point cloud; (4) If the user determines to generate a three-dimensional model of the target, a three-dimensional model and texture are generated based on the dense point cloud, and the dense point cloud and the three-dimensional model are exported.
8. A target three-dimensional reconstruction system based on omnidirectional multi-view, used to implement any method of claims 1-7, characterized in that: include: A target selection module (101), used for selecting a target for three-dimensional reconstruction; A multi-view acquisition module (102) is used to automatically shoot multi-angle images based on a target multi-view acquisition program, thereby sampling visual images from multiple camera positions and multiple angles to acquire multiple views of the target; A three-dimensional reconstruction module (103) is used to perform three-dimensional reconstruction based on multiple views of the target.
9. An electronic device comprising a processor and a memory, characterized in that: The memory stores a plurality of instructions, and the processor is configured to read the instructions and execute the method according to claims 1-7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the plurality of instructions can be read by a processor to execute the method according to claims 1-7.
