KR20130062540A - Method and system for distinguishing multiple objects using external vision device - Google Patents

Abstract

PURPOSE: A method for distinguishing a plurality of objects using an external vision device and a system thereof are provided to supply each unique light emitting pattern to a light emitting unit, thereby distinguishing a position and a motion of an object. CONSTITUTION: A control unit provides each unique bitstream to light emitting units installed in one or more objects(S510). The control unit sets light emitting time of the bitstream(S530). The light emitting unit emits light as a light emitting pattern corresponding to the unique bitstream and the light emitting time(S550). The light emitting pattern is recognized and displayed(S570). [Reference numerals] (AA) Start; (BB) End; (S510) Control unit provides a unique bitstream B to light emitting units installed in objects; (S530) Control unit sets a light emitting time t_b of the light emitting unit; (S550) Light emitting unit emits light as a light emitting pattern corresponding to the unique bitstream B and the light emitting time t_b; (S570) Camera recognizes and displays the light emitting pattern; (S590) User distinguishes each object by the recognized light emitting pattern

Description

METHOD AND SYSTEM FOR DISTINGUISHING MULTIPLE OBJECTS USING EXTERNAL VISION DEVICE}

The present invention relates to a method and system for distinguishing a plurality of objects. More specifically, the present invention can distinguish the position and movement of objects such as a plurality of robots by assigning unique light emission patterns (unique IDs) to the light emitting units, respectively, and obtain the light emission time of the light emitting unit from the camera. A plurality of object discrimination methods and systems using an external vision device that can be adjusted to match time precisely, can significantly improve object discrimination errors.

In order to recognize and distinguish a large number of robots using a vision system, the existing research used unique color information or a certain pattern. This method has a very high success rate in a limited environment with constant lighting, but not in an unpredictably changing external environment. This led to the introduction of a stable vision system in an external environment with varying illuminance using the LED Active Marker. In this study, the active marker was recognized by using the on / off frequency operation of the LED active marker, and the stability was verified through experiments in various environments and applied to one robot. Such a technique is shown in FIG.

1 is a schematic diagram of an object discrimination system according to the prior art. Referring to FIG. 1, the object discrimination system according to the related art displays a robot 110, an active marker 130 installed on the robot, a camera 150 recognizing the robot and the active marker 130, and an image recognized by the camera. The display unit 170 is included. More specifically, the camera 150 recognizes the active marker 130 using, for example, a CCD camera and a lens filter. Here, as shown in FIG. 2, the active marker 130 includes four bright red colored 4x4 LED arrays 210 and operates at a constant on / off frequency, but the number of colors and LEDs is not limited thereto. Does not. In this way, in the active marker 130, the emitted light emitting pattern is driven by the camera 150 by using the on / off frequency operation of the camera 150, and the camera 150 turns on / off the active marker 130. By recognizing the light emission pattern and recognizing the recognized active marker 150 as a robot, the position and movement of the robot can be distinguished through the display unit 170.

However, this conventional technique is limited to one robot, and as of present, there is a limit in distinguishing a plurality of robots or another object. That is, when using the same on / off pattern, even if correctly recognized, there is a disadvantage that can not distinguish a plurality of objects. In particular, the active marker may operate at a constant frequency through the microcontroller, but since the image acquisition time of the general camera is not constant, an error may occur in that it does not properly recognize the actual light emission pattern of the active marker. This problem will be described in detail with reference to FIG. 3.

3 is a graph illustrating an example of a recognition error occurring in the object discrimination system according to the prior art. Referring to FIG. 3, the desired result of the user by operating the active marker at a predetermined frequency should be recognized one of four times and the remaining three times the active marker should not be recognized. However, an error that may be recognized may occur. For example, in the portion indicated by the rectangular border of the pattern graph at the top of the figure, the active marker should be recognized twice, once in the portion indicated by 1. However, since the camera image acquisition time is irregular, two image acquisition pulses occur at the portion where the active marker emits the second light, thereby recognizing the emission of the second active marker twice. As a result, as shown in the lower part of the figure, although the active marker actually emits twice, an error occurs that the acquired image result is recognized three times in total.

Thus, it can be seen that, with the prior art alone, it is not possible to accurately distinguish a large number of objects (eg robots).

R. Cassinis, F., F Tampalini, and R. Fedrigotii, “Active markers for outdoor and indoor robot localization,” Proc.OfDEA-Unibs, 2005 R. Cassinis, F., F. Tampalini. “AMIRoLoS an active marker internet-based robot localization system,” Robotics and Autonomous Systems, 55 (4), pp.306-315, 2007.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem, and an object of the present invention is to assign a unique light emission pattern (unique ID) to each light emitting unit, thereby distinguishing the position and movement of objects such as a plurality of robots. In addition, the present invention provides a method and system for distinguishing a plurality of objects using an external vision device that may adjust the light emission time of the light emitter to exactly match the image acquisition time of the camera, thereby remarkably improving the error of distinguishing the object.

In order to solve the above-described problem, a method of distinguishing a plurality of objects using an external vision apparatus according to an exemplary embodiment of the present invention includes: Giving; (b) the control unit setting an emission time t _b of the bit stream; (c) the light emitting unit emitting light in a light emission pattern according to the inherent bit stream (B) and the light emission time (t _b ); And (d) recognizing and displaying the emitted light emission pattern in the camera unit.

In this case, the bit stream B is set to B = [log ₂ N] +2 (where N is the number of light emitting portions), and all start bits and end bits can be fixed.

In addition, the light emission time t _b may be set to t _b = n α t _f (n is an integer, α t _f is an image acquisition period of the camera, α ≧ 1).

In addition, it is preferable to set said n to n> (alpha) ((2- (alpha)) (1 <(alpha) <= 2).

On the other hand, a plurality of object discrimination system using an external vision device according to another embodiment of the present invention, the light emitting unit is installed on each of one or more objects to emit light; A camera unit for recognizing and displaying a light emission pattern of each light emitting unit; And a control unit for assigning a unique bit stream (B) to each of the light emitting units, and setting a light emission time (t _b ) of the bit stream, wherein the light emitting unit includes the unique bit stream (B) and the light emitting unit. It characterized in that the light emission in the light emission pattern according to the time (t _b ).

In this case, the bit stream B is set to B = [log ₂ N] +2 (where N is the number of light emitting portions), and all start bits and end bits can be fixed.

In addition, the light emission time t _b may be set to t _b = n α t _f (n is an integer, α t _f is an image acquisition period of the camera, α ≧ 1).

In addition, it is preferable that said n is set to n> (alpha) / (2- (alpha)) (1 <= (alpha) <= 2).

According to the present invention, by assigning unique light emission patterns (unique IDs) to the light emitting units, it is possible to distinguish the positions and movements of objects such as a plurality of robots. In addition, by adjusting the light emission time of the light emitting unit to exactly match the image acquisition time of the camera, it is possible to remarkably improve the error of distinguishing the object.

1 is a schematic diagram of an object discrimination system according to the prior art;
2 is a photograph of an active marker used in the prior art.
3 is a graph showing an example of a recognition error occurring in the object discrimination system according to the prior art.
4 is a structural diagram of a plurality of object discrimination system according to an embodiment of the present invention.
5 is a flow diagram of an algorithm for distinguishing a plurality of objects in accordance with the present invention.
6 is a graph showing the definition of emission time t _b , image acquisition time t _f , and n in accordance with the present invention;
7 is a diagram illustrating an example in which a bit stream is provided to four light emitting units according to an exemplary embodiment of the present invention.
8 is a graph for distinguishing between continuous light emission and single light emission of the light emitting unit according to an embodiment of the present invention.
9 is an exemplary graph for setting a light emission time of a light emitting unit according to an embodiment of the present invention.

Hereinafter, a method and a system for distinguishing a plurality of objects using an external vision device according to a preferred embodiment will be described in detail with reference to the accompanying drawings. In the following description, well-known functions or constructions are not described in detail to avoid unnecessarily obscuring the subject matter of the present invention.

In addition, the size of each component in the drawings may be exaggerated for the sake of explanation and does not mean a size actually applied.

4 is a structural diagram of a plurality of object discrimination system 400 according to an embodiment of the present invention. Referring to FIG. 4, a plurality of object discrimination systems 400 according to an exemplary embodiment of the present invention include a controller 410, a light emitting unit 430, and a camera unit 450. More specifically, the light emitting unit 430 may use the above-described active marker 130, but is not necessarily limited thereto, and includes other conventional controllable light emitting means. In addition, one light emitting unit 430 is installed in each of a plurality of objects. The controller 410 assigns a unique ID to the light emitting unit 430. The term “unique ID” herein refers to a signal that is encoded / decoded to operate in a different light emission pattern for each light emitter 430 to distinguish each light emitter 430, and is herein referred to as “unique bit stream (B)”. Has the same meaning as The control unit 410 assigns a unique bit stream B to the light emitting unit 430, and then emits each of the light emitting units 430. The camera unit 450 recognizes the light emission pattern of each light emitting unit 430 that emits light. The camera unit 450 generates an image acquisition pulse when the emission pattern is recognized. As the camera unit 450, a CCD camera may be used, for example. In this case, in order to prevent the error of image acquisition described above in the background art, the controller 450 predicts the image acquisition period of the actual image acquisition pulse, and accordingly, sets the emission time of the emission pattern of the light emitting unit 430. Adjust Details thereof will be described below. The camera unit 450 recognizes the emitted light emission pattern and displays the same to the user. The user may look at the light emission patterns displayed by the camera unit 450 and compare the light emission patterns according to the unique bit streams assigned by the control unit 410 (that is, set by the user) to identify the corresponding objects if they match. have. In addition, the display function may be integrated with the camera unit 450, or may be independent as a separate device to the outside. This can be changed as appropriate according to the user's design convenience. Hereinafter, an algorithm according to the present invention that can distinguish a plurality of objects will be described in more detail.

5 is a flowchart of an algorithm for distinguishing a plurality of objects in accordance with the present invention. Referring to FIG. 5, the controller assigns a unique bit stream B to the light emitting unit installed in the object (S510). That is, each ID is assigned to each object, for example, a robot. For explanation of the ID assignment and distinction to the robot, first, the terms are summarized with reference to FIG.

6 is a graph showing definitions of light emission time t _b , image acquisition time t _f , and n according to the present invention. The term used in Figure 6,

t _f : 1 / frame rate (image acquisition period)

t _b : LED light emitting time

n: an integer multiple of t _f to t _b, that is, _b t = nt _f

N: number of light emitting parts

B: number of bit streams

to be.

Referring to the left figure of FIG. 6, when the period of the image acquisition pulse of the intermediate camera is t _f , and n is 2, the upper LED emission time t _b = 2t _f . As a result, the camera acquires two image acquisition pulses corresponding to the LED emission time, and only two pulses become 1 as shown in the lower figure.

Referring to the right figure of FIG. 6, it is shown that t _b is determined according to the n value. For example, when n is 2, the upper portion shows one light emission time and the lower portion shows two consecutive light emission times.

On the other hand, in order to distinguish the ID of the active marker (ie, the bit stream B), the accurate frame for second (FPS) of the camera and the exact time of turning on / off the active marker must be guaranteed. However, in a real environment, it is not possible to guarantee a constant t _f '(image acquisition cycle in a real situation) due to a change in execution time of an image processing algorithm and characteristics of a camera. In general, the image acquisition period (t _f ') of this actual situation is equal to or greater than the ideal state and exists within the boundary of α times, and α can be defined by Equation 1.

{Equation 1}

_{t f ≤t f '≤αt f (} α> 1)

The present invention encodes / decodes an operation of the light emitter into B bit streams according to Equation 2 below by predicting an α value to determine an ID of an object in which the light emitter is installed.

{Equation 2}

nt _f ≤nt _f '≤nαt _f

That is, given the value of α, by estimating the value of n, the light emission time of the light emitting portion can be set according to Equation 3 below.

{Equation 3}

t _b = nf _t

Further, given the number N of light emitting portions, the unique bit stream B for detecting the light emitting portions can be estimated.

Therefore, the following assumptions must be made for the algorithm of the present invention.

(1) The image acquisition cycle is slowed down due to the computation of the algorithm in the real environment rather than the ideal state (see Equation 1).

(2) The operation of the light emitting portion (LED array of active markers) is user adjustable.

(3) t _f 'is not constant in the actual environment, but t _f ' with a large difference is rare, and in this case, reset processing is possible.

Returning to step S501 of Fig. 5 again, the unique bit stream B is given to the light emitting part (active marker) installed in the object in detail. In order to distinguish each of the bit streams B assigned to the N active markers, it is necessary to give a different bit stream B to each active marker. A signal encoded in each of the given bit streams B is transmitted to the light emitting units, and each light emitting unit decodes the received signal to emit light in the light emission pattern corresponding to the unique bit stream B. FIG. As described above, in order to encode / decode the N active markers into the bit stream (B), the bit stream (B) is required according to Equation 4 below. Explain.

{Equation 4}

B = [log ₂ N] +2

here,

7 is a diagram illustrating an example in which a bit stream is provided to four light emitting units according to an embodiment of the present invention. Referring to the upper part of FIG. 7, when the number N of light emitting units is N = 4, B = 4 according to Equation 4. In Equation 4, the log ₂ N portion is defined by N to distinguish the bit stream, and the color emitting is red. In addition, the start bit and the end bit are always fixed. For example, referring to the lower part of FIG. 7, the bit streams B of four light emitting parts are divided into 0001, 0011, 0101, and 0111. As such, the start bits are all fixed to zero and the end bits are all fixed to one. Meanwhile, the color shown in the table of FIG. 7 is not a color emitted from the active marker. This is indicated in different colors to distinguish between the fixed start and end bits and the bits used for ID identification in the table, and the colors emitted from the actual active markers are all red.

In this manner, it is important to determine the time of light emission among the light emission patterns that emit light according to the bit stream B after the unique bit stream B is assigned. This is to obtain an image corresponding to the bit stream without error by the camera as described above. In particular, in order to distinguish between when the light emitting unit is continuously turned on and when it is not, the active marker must be recognized at least once more when the light emitting unit is continuously turned on to be distinguishable. This example is shown in FIG.

8 is a graph for distinguishing between continuous light emission and single light emission of a light emitting unit according to an exemplary embodiment of the present invention. Referring to FIG. 8, three image acquisition pulses of the camera correspond to each other when the active markers are continuously emitted, and two image acquisition pulses of the camera correspond to the emission of the active marker once. Therefore, when the active marker emits light twice, rather than when the active marker emits light once, since there are at least one corresponding image acquisition pulse, continuous light emission and single emission of the active marker can be distinguished. Hereinafter, in order to enable such a distinction, the light emission time setting of the active marker will be described in detail with reference to FIG. 9.

9 is an exemplary graph for setting the light emission time of the light emitting unit according to the embodiment of the present invention. In the case where the light emission time is set in Equation 3, the minimum (n + 1) should be detected when the maximum number of active markers is detected once when turned on and the second time when the emission markers are turned on twice consecutively. Considering the actual image acquisition time of the camera must satisfy the following equation (5).

{Equation 5}

(n + 1) αt _f <2nt _f

This equation can be simplified to obtain the value n when α is given, and the light emission time can be set using Equation 6 below.

{Equation 6}

n> α / (2-α) (1 ≦ α ≦ 2)

In this way, the controller sets the light emission time t _b of the light emitting unit (S530). Thereafter, the light emitting unit emits light in the light emission pattern according to the unique bit stream B and the set light emission time t _b (S550). The camera recognizes and displays such a light emission pattern (S570). The user can visually distinguish each object, i.e. each robot, by looking at the displayed luminescence pattern and comparing it to the original uniquely given bit stream.

Meanwhile, the distinguishing result of each object according to the algorithm of the present invention will be described. The simulation result table in the case where the light emission time t _b of the light emitting unit is set according to the above algorithm is as follows.

Experiment parameter Average success rate LED Array ID Detection Time alpha Dispersion One 1.3 0.0 100% 1.17 seconds 2 1.3 0.5 100% 1.17 seconds 3 1.3 1.0 100% 1.17 seconds 4 1.5 0.5 100% 1.77 sec 5 1.5 1.0 100% 1.77 sec 6 1.6 0.5 100% 2.97 seconds 7 1.7 0.5 100% 3.57 seconds

Referring to the table, when the α value is given as the table above and the error of the image acquisition time of the camera is given as a variance, the result obtained by repeating the experiment for 10 seconds for 30 seconds was 100% and the actual success rate was 100%. In the implementation, it can be seen that as the value of α increases, the time for distinguishing IDs increases.

In the foregoing detailed description of the present invention, specific examples have been described. However, various modifications are possible within the scope of the present invention. The technical spirit of the present invention should not be limited to the above-described embodiments of the present invention, but should be determined by the claims and equivalents thereof.

110: robot
130: active marker
210: LED
410: control unit
430: light emitting unit
450: camera unit

Claims (8)

(a) the control unit assigning unique bit streams (B) to the light emitting units respectively installed in the one or more objects;
(b) the control unit setting an emission time t _b of the bit stream;
(c) the light emitting unit emitting light in a light emission pattern according to the inherent bit stream (B) and the light emission time (t _b ); And
and (d) recognizing and displaying the emitted light emission pattern in the camera unit.
The method of claim 1,
The bit stream (B),
B = [log ₂ N] +2 (where N is the number of light emitters) and all start and end bits are fixed.
The method of claim 2,
The light emission time t _b is,
t _b = n α t _f (n is an integer, α t _f is the image acquisition period of the camera, α> 1).
The method of claim 3, wherein
N is set to n> α / (2-α) (1 <α ≦ 2).
Light emitting units installed on each of the one or more objects to emit light;
A camera unit for recognizing and displaying a light emission pattern of each light emitting unit; And
A control unit for assigning a unique bit stream (B) to each of the light emitting units and setting a light emission time (t _b ) of the bit stream,
And the light emitting unit emits light with a light emission pattern according to the unique bit stream (B) and the light emission time (t _b ).
6. The method of claim 5,
The bit stream (B),
B = [log ₂ N] +2 (where N is the number of light emitters), wherein all start and end bits are fixed.
The method according to claim 6,
The light emission time t _b is,
t _b = n α t _f (n is an integer, α t _f is the image acquisition period of the camera, α> 1).
8. The method of claim 7,
N is set to n> α / (2-α) (1 <α ≦ 2).

Images