3D stereoscopic vision multi-normal-form visual angle control method and visual angle control system
By using a 3D stereoscopic vision multi-paradigm perspective control method, combined with head tracking devices and an interactive pen, automatic perspective following and smooth switching are achieved, solving the problem of the single perspective control method in existing virtual simulation teaching systems and improving the efficiency and immersion of practical training.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FXB CO LTD
- Filing Date
- 2026-04-20
- Publication Date
- 2026-05-19
AI Technical Summary
In existing automotive virtual simulation teaching systems, the viewpoint control method is simplistic and lacks intelligent assistance. Students need to frequently adjust the viewpoint manually, which affects their learning focus and training efficiency.
Employing a 3D stereoscopic vision multi-paradigm perspective control method, combined with head tracking devices and an interactive pen, it achieves automatic perspective following and smooth switching through automatic perspective optimization, head tracking fine-tuning, and keyboard navigation. It also integrates AI intelligent guidance to provide the best observation perspective.
It improves the efficiency and immersion of 3D spatial observation, reduces visual fatigue, and enhances the intelligence and ease of operation of practical training.
Smart Images
Figure CN122069339A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of virtual simulation teaching, and in particular to a 3D stereoscopic vision multi-paradigm view control method and view control system. Background Technology
[0002] Current automotive virtual simulation teaching software generally adopts a fixed planar perspective or a single roaming mode, with camera positions and observation directions preset and fixed, preventing users from flexibly switching angles according to their observation needs in 3D space. In complex training scenarios such as observing the internal structure of power battery packs, inspecting high-voltage wiring harness connection points in hybrid vehicles, and reading data from fault diagnostic instruments, trainees need to observe component details from different angles and distances, but existing systems struggle to support this, resulting in incomplete structural understanding and difficulty in locating fault points.
[0003] In terms of interaction methods, existing systems mostly rely on mouse and keyboard or touchscreen for operation. Mouse and keyboard operation requires simultaneous coordination of cursor movement and button presses, resulting in long operation paths and high learning costs; touchscreen operation lacks precision and is prone to accidental touches when precise selection of small parts is required. In the hardware form of all-in-one machines and interactive pens, existing interaction methods fail to fully utilize the high-precision positioning advantages of interactive pens and the spatial perception potential of large screens in all-in-one machines, resulting in a disconnect between the operating experience and the immersive feeling of 3D stereoscopic vision.
[0004] Furthermore, the existing system does not provide intelligent perspective optimization functionality in conjunction with target switching and changes in the observation range. When trainees switch the object of observation or zoom in and out on local structures, they need to frequently and manually adjust the camera position, angle, and focal length. This operation is cumbersome, easily interrupts the learning process, distracts attention, and affects the efficiency of practical training and the continuity of teaching. Summary of the Invention
[0005] The main objective of this invention is to provide a 3D stereoscopic vision multi-paradigm view control method and view control system, which aims to solve the technical problems of existing virtual simulation teaching systems having a single view control method, lacking intelligent assistance, and requiring students to frequently manually adjust the view, thus interrupting their learning focus.
[0006] To achieve the above objectives, this invention proposes a 3D stereoscopic vision multi-paradigm viewpoint control method, comprising: In response to a target switching command or observation range zoom command, the camera calculates the target relative position and target orientation using the current observation target as a reference point, drives the camera to move to the target relative position and target orientation, and adjusts the camera parameters to bring the target into the field of view; Acquire user head pose data and generate camera view offset based on the head pose data; Responding to user input control commands, the camera is rotated or moved around the observed target based on the camera's field of view offset.
[0007] Furthermore, the methods also include: Detect the angle between the camera's current upward vector and the reference upward vector, as well as the angle between the camera's viewing direction and the direction of the center point of the observed target; When the angle between the upper vectors exceeds the first threshold, calculate the first rotation amount to rotate the current upper vector of the camera to align with the reference upper vector; When the angle between the observation directions exceeds the second threshold, calculate the second rotation amount to rotate the camera's observation direction to point towards the center point of the observed target. The first rotation amount and the second rotation amount are superimposed to form the target rotation amount, and the target rotation amount is decomposed into multi-frame step amounts, so as to drive the camera to return to the corrected posture frame by frame according to the multi-frame step amounts.
[0008] Furthermore, in response to a target switching command or observation range zoom command, using the current observation target as a reference point, the camera's relative position and target orientation are calculated, the camera is driven to move to the target's relative position and target orientation, and camera parameters are adjusted to bring the target into the field of view. Specifically, this includes: When the 3D virtual simulation software starts and the scene resources are loaded, the preset initial observation position and angle parameters are read, and the camera is driven to move to the initial observation position and angle. When a target switching command is received, the camera acquires the 3D bounding box of the new target model, calculates the center point of the bounding box and the radius of the circumscribed sphere, determines the optimal observation distance based on the radius of the circumscribed sphere, and calculates the camera's target relative position and target orientation based on the center point of the bounding box and the optimal observation distance, generates a movement path, and drives the camera to move along the movement path. Upon receiving a zoom command for the observation range, the system obtains the screen percentage of the target model within the current field of view, calculates the target camera distance based on a preset screen percentage threshold, and adjusts the camera's field of view angle.
[0009] Furthermore, user head pose data is acquired, and camera view offset is generated based on the head pose data, specifically including: The head six-degree-of-freedom pose data collected by the head tracking device is acquired at a preset frequency. The six-degree-of-freedom pose data includes the rotation angle around the three axes and the displacement along the three axes. Multiply the rotation angle by a preset first sensitivity coefficient to obtain the camera rotation offset, and multiply the displacement by a preset second sensitivity coefficient to obtain the camera displacement offset.
[0010] Furthermore, in response to user input control commands, the camera is controlled to rotate or move around the currently observed target, specifically including: Store the correspondence between keyboard keys and camera control commands, including horizontal rotation around the target point, vertical rotation around the target point, zooming in along the line of sight, and zooming out along the line of sight. A spherical coordinate system is constructed based on the current camera position and the current observation target position. The azimuth, pitch, and radius values in the spherical coordinate system are updated based on the incremental control commands input by the buttons and the correspondence between the stored keyboard keys and camera control commands. The camera's target position coordinates are calculated based on the updated azimuth, pitch, and radius values. The camera is then moved to the target position coordinates while maintaining the camera's observation direction pointing towards the current observation target.
[0011] Furthermore, the methods also include: In response to a click on the screen with the interactive pen, a ray is emitted from the click location, the 3D model intersecting with the ray is detected, and the first intersecting model is set as the current observation target; In response to the dragging operation of the interactive pen on the screen, the screen coordinates of the starting point of the drag and the screen coordinates of the current point are recorded. The horizontal offset and vertical offset between the two points are calculated. The horizontal offset is multiplied by a preset rotation coefficient to obtain the rotation angle of the camera around the Y-axis of the world coordinate system. The vertical offset is multiplied by a preset rotation coefficient to obtain the rotation angle of the camera around its own horizontal rightward vector. The two rotation angles are superimposed to drive the camera to rotate, while keeping the camera's observation direction pointing towards the current observation target.
[0012] Furthermore, it also includes: when the current operation step signal is received, reading the preset camera view parameters corresponding to the operation step, and driving the camera to move to the position and angle corresponding to the preset camera view parameters.
[0013] A perspective control system, comprising: An all-in-one machine, which has a built-in processor and memory, with the memory storing 3D virtual simulation software; The interactive pen communicates and connects with the all-in-one machine. The head tracking device communicates with the all-in-one machine. When the processor executes the program in memory, it performs the following functions: in response to the operation of the interactive pen, it selects the observation target in the 3D scene or controls the camera rotation; in response to the head posture data collected by the head tracking device, it generates the camera pose offset and superimposes it on the current camera pose; in response to the switching of the observation target, it automatically calculates the target observation position and drives the camera to move to the target observation position; it detects the camera pose, and when the pose deviation exceeds the threshold, it generates a corrected pose and drives the camera to return.
[0014] Furthermore, when the processor executes a program in memory, it also performs the following functions: In response to user input control commands, the camera is controlled to rotate or move around the currently observed target.
[0015] Furthermore, when the processor executes a program in memory, it also performs the following functions: According to the current operation steps, move the camera to the preset viewpoint corresponding to the operation steps.
[0016] This invention achieves automatic viewpoint following and smooth switching through multi-paradigm control of automatic viewpoint optimization, head tracking fine-tuning, and keyboard navigation, reducing the number of manual operations; combined with real-time viewpoint posture correction, it ensures a stable and comfortable field of view; and integrated AI intelligent guidance automatically recommends the best viewpoint based on the operation steps, which can improve the efficiency and immersion of 3D spatial observation, reduce visual fatigue, and empower the intelligentization of automotive training and teaching. Attached Figure Description
[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.
[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] Figure 1 This is a flowchart illustrating a 3D stereoscopic vision multi-paradigm viewpoint control method according to the present invention. Figure 2 This is a schematic diagram of the view control system structure of the present invention.
[0020] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of the present invention and are not intended to limit the present invention.
[0022] To better understand the technical solution of the present invention, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments.
[0023] like Figure 1 As shown, Figure 1 This is a flowchart illustrating a 3D stereoscopic vision multi-paradigm perspective control method according to the present invention.
[0024] A 3D stereoscopic vision multi-paradigm viewpoint control method includes: S10, in response to the observation target switching command or the observation range zoom command, using the current observation target as a reference point, calculates the camera's target relative position and target orientation, drives the camera to move to the target relative position and target orientation, and adjusts the camera parameters to make the target within the field of view; S20: Acquire user head pose data and generate camera view offset based on head pose data; S30 responds to user input control commands and controls the camera to rotate or move around the observed target based on the camera's field of view offset.
[0025] The invention will now be described in further detail using a virtual simulation training scenario of an automotive power battery pack. Those skilled in the art should understand that this scenario is merely illustrative, and the invention is equally applicable to other automotive virtual simulation training scenarios such as hybrid vehicle fault diagnosis and motor controller assembly / disassembly.
[0026] In one specific embodiment of the present invention, the system first executes an initialization process upon startup. Specifically, the processor reads preset initial camera pose parameters from the memory, including the camera's position coordinates in the world coordinate system, its orientation angle, and its field of view angle. The camera is then moved to the initial observation position and angle, at which point, after the trainee enters the training scenario, the camera is at the optimal observation angle for the entire power battery assembly.
[0027] This initialization process directly drives the camera to the target pose through preset parameters, avoiding the tedious manual adjustments required by trainees, enabling them to quickly enter a learning state and improving the efficiency of training startup.
[0028] S10, in response to the observation target switching command or the observation range zoom command, using the current observation target as a reference point, calculates the camera's target relative position and target orientation, drives the camera to move to the target relative position and target orientation, and adjusts the camera parameters to make the target within the field of view; Furthermore, in response to a target switching command or observation range zoom command, using the current observation target as a reference point, the camera's relative position and target orientation are calculated, the camera is driven to move to the target's relative position and target orientation, and camera parameters are adjusted to bring the target into the field of view. Specifically, this includes: When the method starts, it reads the preset initial observation position and angle parameters and drives the camera to move to the initial observation position and angle. When a target switching command is received, the camera acquires the 3D bounding box of the new target model, calculates the center point of the bounding box and the radius of the circumscribed sphere of the bounding box, determines the optimal observation distance based on the radius of the circumscribed sphere, and calculates the camera's target relative position and target orientation based on the center point and the optimal observation distance, generates a movement path, and drives the camera to move along the movement path. Upon receiving a zoom command for the observation range, the system obtains the screen percentage of the target model within the current field of view, calculates the target camera distance based on a preset screen percentage threshold, and adjusts the camera's field of view angle.
[0029] In this embodiment, when trainees need to switch their observation target during practical training, such as switching from the entire power battery assembly to a specific battery module, the system automatically optimizes the viewing angle through the following steps: First, the system receives a target switching command, triggered by the student clicking on the target model with an interactive pen. The processor acquires the 3D mesh data of the new target model, calculates its minimum axis-aligned bounding box, and then obtains the coordinates of the bounding box's center point and the radius of its circumscribed sphere. Based on the circumscribed sphere radius multiplied by a preset viewing distance coefficient, the optimal viewing distance is determined to ensure that the target model occupies a reasonable proportion in the field of view without exceeding the boundary.
[0030] Secondly, based on the center point of the bounding box and the optimal viewing distance, the camera's relative position to the target and the target's orientation are calculated. Specifically, using the target's center point as a reference point, the camera's target position is obtained by offsetting the optimal viewing distance in the opposite direction of the target's orientation, with the camera pointing towards the target's center point.
[0031] The processor then generates a Bézier curve interpolation path from the current camera position to the target position, discretizes the path into a series of path point sequences, drives the camera to move smoothly along the path, and updates the camera orientation in real time to ensure it always points towards the target center point. The entire roaming process uses a gradual curve to control the speed, accelerating at the beginning and decelerating at the end to avoid visual jumps.
[0032] When a user uses the zoom function to observe a local structure, such as zooming in on the terminal connections of a battery module, the system receives a zoom command for the observation range. The processor obtains the pixel percentage of the target model on the screen within the current field of view, compares it with a preset optimal percentage threshold, calculates the target camera distance in reverse, and simultaneously adjusts the camera's field of view angle to ensure the target remains centered and clearly visible.
[0033] This automatic perspective optimization module calculates the optimal observation position based on the geometric features of the target bounding box and uses Bézier curves to achieve smooth roaming. This reduces the number of times trainees manually adjust the perspective, allowing them to focus on structural observation and fault analysis, and significantly improves training efficiency.
[0034] S20: Acquire user head pose data and generate camera view offset based on head pose data; Furthermore, user head pose data is acquired, and camera view offset is generated based on the head pose data, specifically including: The head six-degree-of-freedom pose data collected by the head tracking device is acquired at a preset frequency. The six-degree-of-freedom pose data includes the rotation angle around the three axes and the displacement along the three axes. Multiply the rotation angle by a preset first sensitivity coefficient to obtain the camera rotation offset, and multiply the displacement by a preset second sensitivity coefficient to obtain the camera displacement offset.
[0035] In this embodiment, the trainee wears head-tracking glasses, and the system achieves head-tracking micro-control through the following steps: The pose acquisition subunit acquires six-degree-of-freedom head pose data from the head tracking device at a preset frequency of 60Hz. The six-degree-of-freedom pose data includes rotation angles (i.e., pitch angle, yaw angle, and roll angle) around the X-axis, Y-axis, and Z-axis, as well as displacement along the three axes.
[0036] The offset generation subunit multiplies the rotation angle by a preset first sensitivity coefficient to obtain the camera rotation offset; it then multiplies the displacement by a preset second sensitivity coefficient to obtain the camera displacement offset. The sensitivity coefficient can be adjusted according to the user's comfort preferences; the higher the sensitivity, the greater the change in viewing angle caused by even slight head movements.
[0037] The pose overlay subunit overlays the camera rotation offset and camera displacement offset onto the current camera's target rotation angle and target position coordinates. Specifically, the final camera pose = base target pose plus head tracking offset.
[0038] In this way, when the trainee's head turns or moves slightly, the camera synchronously produces a slight displacement and angle change. For example, when the trainee looks down to observe the side terminals of the battery module, the camera tilts downward and moves closer. This fine-tuning mechanism is ergonomic, giving trainees a near-realistic physical observation experience and reducing visual fatigue during long training sessions.
[0039] S30 responds to user input control commands and controls the camera to rotate or move around the observed target based on the camera's field of view offset.
[0040] Furthermore, in response to user input control commands, the camera is controlled to rotate or move around the currently observed target, specifically including: Store the correspondence between keyboard keys and camera control commands, including horizontal rotation around the target point, vertical rotation around the target point, zooming in along the line of sight, and zooming out along the line of sight. A spherical coordinate system is constructed based on the current camera position and the current observation target position. The azimuth, pitch, and radius values in the spherical coordinate system are updated incrementally based on the control commands input by the buttons. The camera's target position coordinates are calculated based on the updated azimuth, pitch, and radius values. The camera is then moved to the target position coordinates while maintaining the camera's observation direction pointing towards the current observation target.
[0041] In this embodiment, when a student needs to quickly switch the overall viewing perspective, they can do so via the keyboard. The system implementation is as follows: The key mapping subunit stores the correspondence between keyboard keys and camera control commands. For example, the W key corresponds to zooming in along the line of sight, the S key corresponds to zooming out, the A key corresponds to rotating horizontally counterclockwise around the target point, the D key corresponds to rotating horizontally clockwise, the Q key corresponds to rotating vertically upward around the target point, and the E key corresponds to rotating vertically downward.
[0042] The orbit calculation subunit constructs a spherical coordinate system based on the current camera position, camera view offset, and the current observation target position, with the observation target as the center of the sphere. The camera position is uniquely determined by three parameters: azimuth, pitch, and radius. When a button input is received, the azimuth, pitch, and radius values are incrementally updated according to the control command corresponding to the button.
[0043] The motion execution subunit calculates the camera target position coordinates based on the updated azimuth, pitch, and radius values. The calculation is completed using the formula for converting spherical coordinates to rectangular coordinates. The camera is then driven to move linearly to the target position coordinates while maintaining the camera's observation direction pointing towards the current observation target.
[0044] This keyboard navigation mechanism provides a wide range of perspectives that can be quickly switched. Students can quickly switch from the local structure of the battery module to the overall layout of the power battery assembly, or view the connection relationship between adjacent components, meeting the need for rapid switching between different observation scales.
[0045] Furthermore, the methods also include: Detect the angle between the camera's current upward vector and the reference upward vector, as well as the angle between the camera's viewing direction and the direction of the center point of the observed target; When the angle between the upper vectors exceeds the first threshold, calculate the first rotation amount to rotate the current upper vector of the camera to align with the reference upper vector; When the angle between the observation directions exceeds the second threshold, calculate the second rotation amount to rotate the camera's observation direction to point towards the center point of the observed target. The first rotation amount and the second rotation amount are superimposed to form the target rotation amount, which is then decomposed into multi-frame step amounts to drive the camera to return to the corrected posture frame by frame.
[0046] In this embodiment, to prevent trainees from experiencing uncomfortable viewing angles and postures during operation, the system performs real-time viewing angle detection and correction: The attitude anomaly detection subunit calculates in real time the angle between the camera's current top vector and the top vector in the absolute world coordinate system, as well as the angle between the camera's viewing direction vector and the direction vector of the center point of the observed target. The former is detected to prevent excessive camera tilt (e.g., tilt exceeding 15 degrees), while the latter is detected to prevent the camera from deviating from the target (e.g., the viewing direction deviates from the target center by more than 30 degrees).
[0047] When the angle between the camera's top vector and the top vector in the absolute world coordinate system exceeds a first threshold (e.g., 15 degrees), the attitude correction calculation subunit calculates a correction rotation amount to align the camera's top vector with the top vector in the absolute world coordinate system. This correction rotation amount is calculated using quaternion interpolation to ensure the shortest possible rotation path.
[0048] When the angle between the camera's observation direction vector and the observation target's center point direction vector exceeds a second threshold (e.g., 30 degrees), a correction rotation amount is calculated to rotate the camera's observation direction to point towards the observation target's center point.
[0049] The smooth regression subunit decomposes the corrected rotation amount into multi-frame step rotation amounts, driving the camera frame by frame to return to the corrected target posture using a preset easing curve. The entire process is smooth and natural, and the trainee hardly notices it. For example, if the trainee accidentally rotates the camera to an inverted position, the system automatically corrects the camera to a horizontal and stable posture within 0.5 seconds to avoid visual dizziness.
[0050] This perspective correction mechanism ensures the stability and comfort of the observation field, effectively reducing the risk of visual fatigue during long-term training.
[0051] Furthermore, the methods also include: In response to a click on the screen with the interactive pen, a ray is emitted from the click location, the 3D model intersecting with the ray is detected, and the first intersecting model is set as the current observation target; In response to the dragging operation of the interactive pen on the screen, the screen coordinates of the starting point of the drag and the screen coordinates of the current point are recorded. The horizontal offset and vertical offset between the two points are calculated. The horizontal offset is multiplied by a preset rotation coefficient to obtain the rotation angle of the camera around the Y-axis of the world coordinate system. The vertical offset is multiplied by a preset rotation coefficient to obtain the rotation angle of the camera around its own horizontal rightward vector. The two rotation angles are superimposed to drive the camera to rotate, while keeping the camera's observation direction pointing towards the current observation target.
[0052] In this embodiment, when a student uses an interactive pen to click on a 3D model on the screen, the system responds to the click operation by emitting a ray inward from the click location along the screen, detecting the 3D model that intersects with the ray, setting the first intersecting model as the current observation target, highlighting the target, and triggering the aforementioned automatic view optimization module to complete the view switching.
[0053] When a student drags on the screen using the interactive pen, the system records the screen coordinates of the starting point and the current point, and calculates the horizontal and vertical offsets between the two points. The horizontal offset is multiplied by a preset rotation coefficient to obtain the camera's rotation angle around the Y-axis of the world coordinate system, enabling horizontal panning. The vertical offset is multiplied by the preset rotation coefficient to obtain the camera's rotation angle around its own horizontal rightward vector, enabling pitch observation. The two rotation angles are then superimposed to drive the camera's rotation while maintaining the camera's viewing direction pointing towards the current target.
[0054] The interactive pen operation combines the precision of click selection with the intuitiveness of drag and rotate, complementing head tracking and keyboard navigation. Students can freely choose the interaction method according to their operating habits and scenario needs, thus improving the ease of operation.
[0055] Furthermore, it also includes: when the current operation step signal is received, reading the preset camera view parameters corresponding to the operation step, and driving the camera to move to the position and angle corresponding to the preset camera view parameters.
[0056] Furthermore, this invention also integrates AI intelligent guidance functionality, specifically implemented as follows: The system stores a database mapping training operation steps to preset camera viewpoints. For example, in the training of disassembling a power battery module, step one, "disconnecting the high-voltage harness," corresponds to the viewpoint parameters of the camera focusing on the harness connection point; step two, "removing the fixing bolts," corresponds to the viewpoint parameters of the camera focusing on the bolt position.
[0057] When a student performs a certain operation step, the system receives the signal of the current operation step, reads the preset camera view parameters corresponding to the operation step, triggers the aforementioned automatic view optimization module to move the camera to the position and angle corresponding to the preset camera view parameters, and highlights the operation area on the screen to help the student accurately locate the operation position.
[0058] This AI-powered intelligent guidance function deeply integrates perspective control with the training process, automatically providing trainees with the best observation angle, reducing the time trainees spend searching for operation points, and improving the standardization of training operations and learning efficiency.
[0059] like Figure 2 As shown, Figure 2 This is a schematic diagram of the view control system structure of the present invention.
[0060] Reference Figure 2 This invention proposes a perspective control system, comprising: The all-in-one machine 10 has a built-in processor and memory, and the memory stores 3D virtual simulation software. The interactive pen 20 is connected to the all-in-one machine 10 for communication. The head tracking device 30 is communicatively connected to the all-in-one machine 10; When the processor executes the program in memory, it performs the following functions: in response to the operation of the interactive pen 20, it selects the observation target in the 3D scene or controls the camera rotation; in response to the head posture data collected by the head tracking device 30, it generates the camera pose offset and superimposes it onto the current camera pose; in response to the switching of the observation target, it automatically calculates the target observation position and drives the camera to move to the target observation position; it detects the camera pose, and when the pose deviation exceeds the threshold, it generates a corrected pose and drives the camera to return.
[0061] Furthermore, when the processor executes a program in memory, it also performs the following functions: In response to user input control commands, the camera is controlled to rotate or move around the currently observed target.
[0062] Furthermore, when the processor executes a program in memory, it also performs the following functions: According to the current operation steps, move the camera to the preset viewpoint corresponding to the operation steps.
[0063] The above are only some embodiments of the present invention and do not limit the patent scope of the present invention. All equivalent structural transformations made under the technical concept of the present invention using the contents of the present invention specification and drawings, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.
Claims
1. A 3D stereoscopic vision multi-paradigm viewpoint control method, characterized in that, include: In response to a target switching command or a view range zoom command, the camera calculates the target relative position and target orientation using the current target as a reference point, drives the camera to move to the target relative position and target orientation, and adjusts the camera parameters to bring the target into the field of view; Acquire user head pose data and generate camera view offset based on the head pose data; Responding to user input control commands, the camera is rotated or moved around the observed target based on the camera's field of view offset.
2. The 3D stereoscopic vision multi-paradigm viewpoint control method according to claim 1, characterized in that, The method further includes: Detect the angle between the camera's current upward vector and the reference upward vector, as well as the angle between the camera's viewing direction and the direction of the center point of the observed target; When the angle between the upper vectors exceeds the first threshold, calculate a first rotation amount to rotate the current upper vector of the camera to align with the reference upper vector; When the included angle of the observation direction exceeds the second threshold, calculate the second rotation amount to rotate the camera's observation direction to point to the center point of the observed target; The first rotation amount and the second rotation amount are superimposed to form the target rotation amount, and the target rotation amount is decomposed into multi-frame step amounts, so as to drive the camera to return to the corrected posture frame by frame according to the multi-frame step amounts.
3. The 3D stereoscopic vision multi-paradigm viewpoint control method according to claim 1, characterized in that, The step of responding to an observation target switching command or an observation range zoom command, using the current observation target as a reference point, calculates the camera's relative position and target orientation, drives the camera to move to the target's relative position and target orientation, and adjusts the camera parameters to bring the target into the field of view, specifically including: When the 3D virtual simulation software is started and the scene resources are loaded, the preset initial observation position and angle parameters are read, and the camera is driven to move to the initial observation position and angle. Upon receiving a target switching command, the camera acquires the 3D bounding box of the new target model, calculates the center point of the bounding box and the radius of the circumscribed sphere of the bounding box, determines the optimal observation distance based on the radius of the circumscribed sphere, calculates the camera's target relative position and target orientation based on the center point of the bounding box and the optimal observation distance, generates a movement path, and drives the camera to move along the movement path. Upon receiving a zoom command for the observation range, the system obtains the screen percentage of the target model within the current field of view, calculates the target camera distance based on a preset screen percentage threshold, and adjusts the camera's field of view angle.
4. The 3D stereoscopic vision multi-paradigm viewpoint control method according to claim 1, characterized in that, The step of acquiring user head pose data and generating camera viewpoint offset based on the head pose data specifically includes: The head six-degree-of-freedom pose data collected by the head tracking device is acquired at a preset frequency. The six-degree-of-freedom pose data includes the rotation angle around the three axes and the displacement along the three axes. The camera rotation offset is obtained by multiplying the rotation angle by a preset first sensitivity coefficient, and the camera displacement offset is obtained by multiplying the displacement by a preset second sensitivity coefficient.
5. The 3D stereoscopic vision multi-paradigm viewpoint control method according to claim 1, characterized in that, The control of the camera to rotate or move around the currently observed target in response to user input control commands specifically includes: The control commands include horizontal rotation around the target point, vertical rotation around the target point, zooming in along the line of sight, and zooming out along the line of sight. A spherical coordinate system is constructed based on the current camera position and the current observation target position. The azimuth, pitch, and radius values in the spherical coordinate system are updated based on the incremental control commands input by the buttons and the correspondence between the stored keyboard keys and camera control commands. The camera target position coordinates are calculated based on the updated azimuth, pitch, and radius values. The camera is then driven to move to the target position coordinates while maintaining the camera's observation direction pointing towards the current observation target.
6. The 3D stereoscopic vision multi-paradigm viewpoint control method according to claim 1, characterized in that, The method further includes: In response to a click on the screen with the interactive pen, a ray is emitted from the click location, the 3D model intersecting with the ray is detected, and the first intersecting model is set as the current observation target; In response to the dragging operation of the interactive pen on the screen, the screen coordinates of the starting point of the drag and the screen coordinates of the current point are recorded. The horizontal offset and vertical offset between the two points are calculated. The horizontal offset is multiplied by a preset rotation coefficient to obtain the rotation angle of the camera around the Y-axis of the world coordinate system. The vertical offset is multiplied by a preset rotation coefficient to obtain the rotation angle of the camera around its own horizontal rightward vector. The two rotation angles are superimposed to drive the camera to rotate, while keeping the camera's observation direction pointing towards the current observation target.
7. The 3D stereoscopic vision multi-paradigm viewpoint control method according to claim 1, characterized in that, Also includes: When the current operation step signal is received, the preset camera view parameters corresponding to the operation step are read, and the camera is driven to move to the position and angle corresponding to the preset camera view parameters.
8. A perspective control system, characterized in that, include: An all-in-one machine, wherein the all-in-one machine has a built-in processor and memory, and the memory stores three-dimensional virtual simulation software; An interactive pen, which communicates with the all-in-one machine; A head tracking device is communicatively connected to the integrated machine; When the processor executes the program in the memory, it performs the following functions: in response to the operation of the interactive pen, it selects an observation target in the three-dimensional scene or controls the camera rotation; in response to the head posture data collected by the head tracking device, it generates a camera pose offset and superimposes it on the current camera pose. In response to a change in the observation target, the system automatically calculates the target observation position and drives the camera to move to the target observation position; The camera pose is detected, and when the pose deviation exceeds a threshold, a corrected pose is generated and the camera is driven to return to its original position.
9. The system according to claim 8, characterized in that, When the processor executes the program in the memory, it also performs the following functions: In response to user input control commands, the camera is controlled to rotate or move around the currently observed target.
10. The system according to claim 8 or 9, characterized in that, When the processor executes the program in the memory, it also performs the following functions: According to the current operation steps, move the camera to the preset viewpoint corresponding to the operation steps.