当前位置：网站首页>Principle of motion capture system

Motion capture is the process of recording the motion of objects or people , Also known as motion capture or motion capture . The technology also involves dimensional measurement 、 Data that can be directly processed by computer, such as the positioning and orientation determination of objects in physical space . This technology is widely used in robots 、 Unmanned aerial vehicle (uav) 、 Multi agent cluster collaborative control 、 military 、 Medical care 、 sports 、 Scientific research and film 、 Animation 、 game 、 Virtual live broadcasting . Distinguish from technical principles , Mobile fishing can be divided into ： Optical type 、 Inertial 、 mechanical 、 Acoustic and electromagnetic . Among them, mechanical 、 Acoustic 、 Electromagnetic motion capture system , With the development of the times , Limited by their own shortcomings , It has been phased out of the market , Now infrared optical motion capture or inertial motion capture is widely used in the industry .

1. Inertial motion capture

Inertial navigation sensor 、 The signal receiver and data processing system constitute the inertial sensor type motion capture system . Wear the posture sensor on the main body parts , The attitude signal can be wirelessly transmitted to the data processing system for motion calculation . The attitude sensor integrates inertial sensor 、 Gravity sensors 、 Accelerometer 、 Magnetic induction meter 、 Micro gyroscope and other elements , Get the posture information of each part of the limbs , Then combine the length information of bones and the hierarchical connection relationship of bones , Calculate the spatial location information of the clearance node .

advantage ：
Portable and easy to operate , It can be used outdoors .

shortcoming ：

Inaccurate positioning —— The spatial position information obtained by integrating the limb posture information has different degrees of integral drift , And some sensors will also be affected by the surrounding magnetic field, which will affect the accuracy ;
Poor endurance —— The sustainable working time is limited after charging once , Not suitable for long-term use ;
Data drift —— The principle itself is based on the assumption of single leg support and ground constraint , It is impossible to solve the motion positioning of feet off the ground ;
Limited movement —— The sensor's own weight and cable connection will still impose certain constraints on the actor's action performance ;
Double the cost —— Especially when the number of captured objects increases , The cost of equipment directly doubled , Lose price advantage .

2. Optical motion capture

Through a plurality of optical lenses arranged in the site, it is captured and fixed on the human body from different angles / Location information of reflective identification points on the object surface , Capture its posture . According to the catch / People have different characteristics , Optical motion capture can also be divided into unmarked dot optical motion capture system 、 Marking point optical motion capture system .

（1） The principle of unmarked dot optical motion capture

The first is motion capture based on ordinary video images , By extracting the coordinates of human joint points in the two-dimensional image , Then calculate the three-dimensional space coordinates of the joint according to the three-dimensional measurement of multi camera vision .
The second is motion capture of infrared camera image based on active heat source illumination to separate foreground and background information , Also known as thermal motion capture , The principle is similar to the first one .
The third is motion capture of 3D depth information , The system obtains the 3D depth information of the object in the field of view in real time based on structured light coding projection , Human shape detection based on three-dimensional morphology , Extract joint trajectory .

Overall speaking , Unmarked dot motion capture still has low accuracy 、 Movement deformation and other problems .

（2） Marking point optical motion capture principle

The marking point optical motion capture system uses multiple optical cameras to capture and stick to the captured object from all angles / Human key nodes Marker spot , And real-time transmission to the data processing workstation , The precise spatial coordinates of the point are calculated by using the principle of triangulation , Then calculate the six degrees of freedom movement of the bone . Both active and passive optical mobile capture are classified according to the type of marked points .

a. Active

Use LED As an identification point , Stick it on the joints of human body LED Then connect through cable , Power supply .

advantage ：

The tracking accuracy is high , Good recognition robustness . And it has its own light source , Therefore, it can be indoors 、 Outdoor use , Not limited by the site .

shortcoming ：

Motion deformation —— Time sequence encoded LED The recognition principle is to rely on the camera to collect imaging marks for multiple marking points at each time ID, The synchronization is broken , And the sampling rate of effective action frames is low , It is not suitable for capturing fast motion and data analysis ;
Low accuracy ——LED The visual angle of the identification point is small , The interior of an optical moving capture lens is often integrated with two cameras for close range acquisition , This kind of narrow baseline structure directly affects the accuracy of visual three-dimensional measurement ;
Double the cost —— Same question , Motion occlusion is inevitable during movement, resulting in data loss , Want to solve the problem of blocking blind areas , Only multiply the number of lenses , And the cost of equipment also rises ;
The number of people is limited —— It is still limited by the principle of sequential coding , The system can only carry a certain number of identification points , Usually no more than at the same time 2 Individuals conduct mobile capture collection .

b. Passive

In the passive optical motion capture system, reflective identification points are pasted on the captured object / Key nodes on human body , The infrared light emitted by the optical motion capture camera is generated by marker The surface reflective coating is reflected into the motion capture camera to locate the point in three-dimensional space .

advantage ：

Mature technology , High precision 、 High sampling rate 、 Low delay and accurate positioning , The movement performance is barrier free , And the cost of marking points is low, which can be flexibly increased 、 Distribution , Underwater environment can also be applied .

shortcoming ：

Environmental constraints —— Passive marker The dot itself does not glow , Collect by adjusting the threshold of the optical camera , Direct sunlight or interference from other reflectors will affect the capture effect ;
No anti occlusion —— Once the reflective marking point is completely covered during movement , Will cause movement dislocation , Data needs to be manually repaired later , Or use the way of light habitual mixing to avoid blocking .

performance	Inertial	Pointless type	Active	Passive
Positioning accuracy	low	low	in	high
sampling frequency	high	low	low	high
Data quality	in	low	in	high
Fast capture capability	high	low	low	high
Number of catches	in	low	in	high
Range of motion	Big	Small	in	in
Subject to interference	Ferromagnet	The sun 、 heat source	Strong light source	The sun
Convenience	low	high	low	in
Applicability	in	low	in	high

Two 、 Composition of optical motion capture system

1. Infrared optical lens

As the protagonist of optical dynamic capture system , Resolution of optical camera 、 Frequency and field of view angle directly affect the accuracy and effect of the whole dynamic capture system .

（1） Lens resolution

Reflective marker points are presented in the form of pixels in the picture of the lens , The center point of the pixel circle is used as the two-dimensional coordinate of the point . thus it can be seen , As long as this pixel shape is more rounded , The more accurate the coordinates of the center point .

therefore , Under the same conditions ,1200 The accuracy of the data collected by the ten thousand pixel lens is much higher than the resolution 130 A ten megapixel lens .

（2） Lens frequency

Lens frequency , Or frame rate , It refers to the number of frames per second that the picture is refreshed , That is, the number of times the graphics processor refreshes in one second . Its unit is usually fps Express , Or use Hz To express . The higher the frame rate of the optical lens , It means that the more image information is captured , Then the captured person / The accuracy of three-dimensional spatial data of objects is higher , Move more smoothly .

（3） Lens field angle

The field angle refers to the optical lens as the fixed point , The included angle formed by the two edges of the maximum range that the captured object can pass through the lens . This angle represents the field of vision of the optical lens . According to the lens classification , There are two types of ： wide-angle lens 、 Non wide angle lens . The differences are as follows ：

Lens type	Visual distance	Field angle range
wide-angle lens	near	Big
Non wide angle lens	far	Small

As shown in the figure , The wide-angle lens has a wider field of vision , However, the number of pixels occupied by a reflective marker point is relatively small . In contrast, the field angle on the right is narrow , But the accuracy of the collected data is higher than that on the left .

More Than This , There is often picture distortion at the edge of wide-angle lens , If the captured object is at the edge , The extraction accuracy of two-dimensional coordinates will also be affected .

2. Reflective markers

The reflective marking point is a solid ball , There are no electronic components inside , There is no need for power supply or connecting wire , Their surfaces are covered with reflective materials , To reflect the light produced by the optical lens . When in use, only reflective identification points are sampled by adjusting the threshold of the camera itself , Omit other objects around .

3. Software and accessories

Including motion capture operation and analysis and processing software 、POE Switches, fixtures and other accessories .

3、 ... and 、 Application scenarios

NOKOV Metric optical 3D motion capture system , High precision 、 Low delay infrared camera captures reflective identification points , Collect and generate accurate 、 Real time action information , It can be widely used in ：

1. robot & Project direction

Infrared optical motion capture system is often used as a reliable indoor positioning means , For all kinds of agents , Like drones 、 Car 、 Mobile robots 、 Mechanical arms provide high accuracy 、 Low latency 3D spatial location information , It can also verify the accuracy of truth values for machine learning

Case study ： Unmanned car clusters can cross obstacles autonomously
Customer ： Institute of automation, Chinese Academy of Sciences

【 Classic cases of motion capture 】 Unmanned car clusters can cross obstacles autonomously

Formation and autonomous obstacle avoidance of multi-agent cooperative control platform of Institute of automation of Chinese Academy of Sciences The unmanned trunking system of the Institute of automation of the Chinese Academy of Sciences is divided into three subsystems , Positioning subsystem 、 Communication subsystem and control subsystem , It can realize the control of single unmanned vehicle and unmanned aerial vehicle 、 Ground air coordination 、 Cluster combat and unmanned vehicles 、 UAV formation performance and other functions https://www.nokov.com/support/case_studies_detail/Multi-Agent-Formation-to-Avoid-Obstacles-Based-on-Motion-Capture-System.html

2. The movie 、CG Animation 、 Game direction

In film production and video game development , Optical dynamic capture system can be used to record the action of dynamic capture actors , And use this information to drive two-dimensional 、 The three dimensional CG Animated characters . In addition to body movements , It also shows facial expressions to actors 、 Capture subtle movements of fingers .

Case study ： Game making basketball action capture

Game making basketball action capture NOKOV Mobile capture equipment is widely used in game production , Or film and television animation 、 Virtual shooting 、 Virtual variety 、 Virtual performance 、TVC Other fields .https://www.nokov.com/support/case_studies_detail/motion-capture-basketball-game-demo.html

3. Virtual anchor 、 Virtual live direction

Comparison game 、CG Animation requires high-precision real-time capture of human motion data , Virtual live broadcast or virtual roadshow pay attention to real-time and stability , To ensure that it will be broadcast for a long time , Virtual anchor 、 Virtual idols still maintain a good image and smooth movements .

Case study ：Vtuber Virtual live broadcast application instance

Virtual digital human trainer fitness virtual live playback

Vtuber Virtual live broadcast application instance NOKOV Optical motion capture supports virtual anchors 24 Live for hours https://www.nokov.com/support/case_studies_detail/real-time-virtual-humans-with-optical-motion-capture.html

4. Motion analysis 、 Direction of sports rehabilitation

The direction of motion analysis and rehabilitation is usually related to the dynamometer 、 EMG and other equipment data synchronization , To obtain kinematic data and combine mechanics 、 Anatomic analysis .

Case study ： Ice and Snow Sports Science Laboratory
Customer ： Physical Education College of Northeast Normal University

Ice and Snow Sports Science Laboratory - Physical Education College of Northeast Normal University

5. Virtual reality direction

In virtual reality application scenarios , The optical mobile capture system will locate the helmet or handle and corresponding accessories , And between VR There are huge differences between directions and scenes , It also requires rich product lines and customized solutions to meet various needs .

More cases related to motion capture ：

Classic case | NOKOV Measure motion capture Classic case https://www.nokov.com/support/case_studies.html

原网站

版权声明
本文为[MocapLeader]所创，转载请带上原文链接，感谢
https://yzsam.com/2022/182/202207011550042477.html