Description MOTION CAPTURE DEVICE USING MOVING CAMERA

Technical Field

[1] The present invention relates to motion capture device, and more particularly, to motion capture device by moving cameras attached on the moving object like human body. Background Art

[2] The motion capture device is used to digitize the information of complicated motion of object like human body to produce the animation of digital character of game or special effect of movie. The conventional motion capture devices can be classified into 3 categories: mechanical, magnetic and optical one.

[3] In order to use mechanical motion capture device, actor must wear mechanical sensors which are uncomfortable. The mechanical parts restrict the range of action.

[4] Magnetic motion capture system is so sensitive to the environment that the work place must be selected very carefully.

[5] Optical capture device use light source(LS) , self reflective mark(SM) on actor and camera(CA) to take the image as shown in Fig.l. The optical mark is more comfortable to wear than the mechanical sensors and It is not sensitive to the environment. The self reflective film is well known product of 3M company which reflects the light back to the light source.

[6] The optical motion capture requires complicated post processing if the actor rotate.

In other words, because each marks is identical, each marks must be identified by human operator. It is very difficult to capture the motion of many person at one time. If two person approach, overlap and go back to their original position then the the marks on two persons will overlap and they must be identified by human operator in postprocessing stage. Another example of drawback of the optical motion capture system is that it can not distinguish the folding motions of flag like the flag in Fig.2 and Fig.3.

[7]

[8]

Disclosure of Invention

Technical Problem

[9] To solve the above problem, it is an object of the present invention to provide a motion capture device of moving camera which can capture the folding motion of flag and rotating motion of actor.

[10]

[H]

Technical Solution

[12] To achieve the object, there is provided a motion capture device of moving camera comprising: mark portion for providing reference position; camera portion for capturing image of mark; and image processing portion for analyzing the captured image of mark and producing the position information between camera and mark.

Advantageous Effects

[13] By using the motion capture device of the present invention, it is possible to capture the folding motion of flag and rotating motion of actor without complicated postprocessing of professional operator. Therefore, It can be used as motion capture device for real time video game. For example, gamer can control the action of character of game. It can be also used for robot to capture the motion and measure the position of robot.

[14]

Brief Description of the Drawings

[15] Fig.l is the conventional optical motion capture system DDD

[16] Fig.2 is the flag folded in forward direction.

[17] Fig.3 is the flag folded in backward direction.

[18] Fig.4 is the motion capture system of present invention.

[19] Fig.5 is the example of rectangle marks.

[20] Fig.6 is the example of rotated rectangle marks whose angle difference is 180 degree.

[21] <description of symbols in drawings >

[22] SM : self reflective mark LS : light source

[23] CA : camera MO: monitor

[24] dm : direction mark im : id mark

[25]

Best Mode for Carrying Out the Invention

[26] Fig.1 shows the conventional optical motion capture system. Self reflective marks(SM) are attached on the human body or moving object and light source(LS) is positioned around the camera(CA). The light from the light source(LS) reflected back to the camera(CA) by self reflective mark(SM) and results in the bright spot in the captured image. In order to trace the bright spot in the captured image, it is recommended to use infrared light source and infrared sensitive camera. The motion information of human is produced by the image processing program which analyze the image and trace the bright spot.

[27] The motion capture device of present invention in Fig.4 includes the cameras(CA) attached on the human body or object and marks of light source(LS). The light

sources(LS) in Fig.4 are arranged as rectangle mark and positioned at computer monitor(MO). Such an arrangement of light sources is identical to the mark in embodiment of Korean patent 10-0532525-0000. Each of image captured by camera is transferred to computer or digital signal processor by image grabber and analyzed by image processing program which produces the relative 3 dimensional position between camera and light sources and motion data. In order not to restrict the action of human whose motion is to be captured, it is recommended to use the wireless camera. The algorithm to calculate the relative 3 dimensional position between camera and rectangle mark by capturing the rectangle mark and analyzing the shape of rectangle mark is well known as 'Perspective n point problem' in image processing technology. Details of the algorithm can be found in http://homepages.inf.ed.ac.uk/rbf/CVonline/LOC AL_COPIES/MARBLE/high/pia/solv ing.htm. The more intuitive method to calculate the relative position between camera and rectangle mark can be found in Korean patent 10-0532525-0000. the mark of light source in Fig.4 is rectangle and the number of feature point n in 'perspective n point problem' is 4. The present invention does not restrict the method to calculate the relative position between mark and camera. The algorithm of perspective n point problem is only an example of such method. In order to distinguish the light sources from the background, it is recommended to use the infrared light source and infrared sensitive camera. And in order to reduce the number of camera, it is recommended to use the camera of wide angle of field of view like fisheye lens camera. Such a case, the distortion correction must be included in the image processing portion. It does not matter to use the rectangle mark without light source as in Fig.5. In Fig.5, the black rectangle marks are attached on the white background. If the human does not rotate or spin then the motion capture can be achieved with only one rectangle mark and cameras on the font side of human body as in Fig.4 but if the action of human is very wild and if he spin then it is recommended to use many rectangle marks in all possible direction as in Fig.5, and it is recommended to insert the ID mark(im) like 1,2,3,4,5,6,.. in order to distinguish the each rectangle marks. The white circles (dm) in the black rectangle marks in Fig.5 are the examples of direction mark. The direction marks can be used to determine whether the rectangle is not rotate or rotate in 180 degree as Fig.6. The ID marks in captured image can be recognized by optical character recognition technology. It does not matter to use the 2 dimensional barcode mark or any kind of distinguishable shape rather than numeric character as the ID marks. It is recommended to attach the cameras not only on the front side of human but also on the back of human in order to capture the motion of spin or tumbling. The motion capture device of present invention can distinguish the folding state and shape of Fig.2 and Fig.3 by attaching the cameras on the front and back surface of flag. It is recommended

to attach the cameras in row and column like matrix. [28]