KR20210097291A - System and method for recognizing gestures - Google Patents

Abstract

Disclosed are a gesture recognition system and a method thereof. According to the present invention, based on RANSAC, human gestures can be recognized using box recognition, person recognition, estimation of a height of a recognized person, and a human arm recognition technology. The gesture recognition system of the present invention comprises: a camera unit (110); a control unit (120); a data storage unit (130); and a communication unit (140).

Description

Gesture recognition system and method

The present invention relates to a gesture recognition system and method, and more particularly, a gesture recognition system and method using RANSAC-based box recognition, person recognition, and height estimation of a recognized person and human arm recognition technology. is about

In general, in the trunk constituting the rear part of the vehicle, general luggage, emergency work tools, spare tires, and the like are stored.

In addition, in case of an accident, a first aid kit is stored, and the thing covering the trunk room is called a trunk lid.

The trunk lid is controlled by the trunk lock device, and a hinge device is provided so that the trunk lid can be easily raised with little force as long as the lock device is released.

Recently, a power trunk/power tailgate automatic opening/closing device that can automatically open or close the trunk lid by simply operating a switch from the driver's seat has been developed.

In a general power tailgate control system, it automatically opens or closes the power tailgate through a tailgate button located outside or remote control of a wireless door lock (Remote Keyless Entry, RKE).

In addition, the conventional power tailgate control system may also perform locking and unlocking functions of the power tailgate.

However, in the power tailgate control system according to the prior art, when the user is holding an object or both hands are not free, the user must put down the object and press a button to open the trunk/tailgate, so this process is convenient from the user's point of view. There is a problem of lowering.

Korean Patent Laid-Open Publication No. 10-2015-0007407 (Title of the Invention: Power Trunk or Power Tailgate Control System)

In order to solve this problem, an object of the present invention is to provide a gesture recognition system and method using RANSAC-based box recognition, person recognition, and height estimation of a recognized person and human arm recognition technology. .

In order to achieve the above object, an embodiment of the present invention provides a gesture recognition system, comprising: a camera unit for outputting a photographed image around a vehicle including distance information to an object in a Time of Flight (ToF) method; Generates 3D point cloud data based on the location of 3D points using accumulated data of past captured images through data sampling of the captured images, extracts basic features based on RANSAC (Random Sample Consensus), and the RANSAC based on the recognition of a box possessed by a person, recognition of a person, estimation of a recognized height of a person, and a human arm from the photographed image to determine an open gesture according to the angle of the arm, and control the operation of the tailgate according to the determination result a control unit outputting a signal; a data storage unit for storing the generated 3D point cloud data and RANSAC-based operation information; and a communication unit configured to transmit an operation signal of the tailgate to a vehicle control unit.

In addition, the basic feature according to the embodiment is that the controller establishes a model satisfying randomly extracted sample data based on RANSAC, stores a model with a large number of data close to the established model, and repeats a certain number of times. It is characterized by returning the model with the largest number of supported data as the final result.

In addition, in box recognition according to the embodiment, the control unit randomly extracts three points in space based on RANSAC, and calculates the distance between the reference plane and arbitrary points calculated based on the extracted three points, It is characterized in that the center of gravity of the box is calculated through the three extracted points, and the box is recognized by using whether the centers match.

In addition, in the human recognition according to the embodiment, the control unit detects a circle having an arbitrary extraction point as a center point in two dimensions based on RANSAC, calculates the distance between the vertically projected point and the center point, and the distance is The closer it is, the more it is recognized as a human being.

In addition, the control unit according to the embodiment determines the highest point of the point extracted by RANSAC as the height of a person, measures the head width and the shoulder width based on the width from the highest point to a predetermined distance below, and the measured head width and If the shoulder width is within a certain range, it is characterized in that it is recognized as a person.

In addition, the human arm recognition according to the embodiment is performed by the controller randomly extracting two points in space based on RANSAC, and calculating the distance between the reference straight line calculated based on the two extracted points and arbitrary points. It is characterized in that it is recognized as an arm as the distance increases.

In addition, the control unit according to the embodiment calculates an angle with the Y-axis by calculating the direction vector of the straight line with the vertical Z-axis, and the most recent angle of the calculated angle is in the range of 0° to 60°, calculated in the past It is characterized in that the number of straight lines in the range of -45° to 0° is counted and if the relative angle is greater than a certain value, it is determined as an open gesture.

In addition, the control unit according to the embodiment is characterized in that, when a predetermined number or more of points located below a preset angle are detected while the tailgate is being opened, it is determined as an obstacle and additionally outputs an operation stop signal of the tailgate.

In addition, in the gesture recognition method according to an embodiment of the present invention, a) a control unit is input through the camera unit 110 that outputs a photographed image around the vehicle including distance information to an object in a Time of Flight (ToF) method. detecting a tailgate opening gesture signal by analyzing data of the captured image; and b) the control unit detects the open/closed state of the tailgate, and performs an obstacle recognition step if the tailgate is being opened, and performs a gesture and box recognition step if the tailgate is in a closed state.

In addition, the step of recognizing the obstacle according to the embodiment may include: receiving, by the controller, a photographed image around the vehicle including distance information to an object in a Time of Flight (ToF) method; generating, by the controller, 3D point cloud data based on the location of a 3D point using accumulated data of past captured images through data sampling of captured images; setting, by the controller, an upper region of the tail gate as a region of interest (ROI) and detecting an obstacle within the ROI; when an obstacle is detected, calculating, by the controller, an angle between the detected obstacle and the lower end of the tailgate and comparing it with a preset reference value; and outputting, by the controller, an operation stop signal of the tail gator when the calculated angle is less than or equal to a preset reference value.

In addition, the step of recognizing the gesture and the box according to the embodiment may include: receiving, by the controller, a photographed image around the vehicle including distance information to an object in a Time of Flight (ToF) method; generating, by the controller, 3D point cloud data based on the location of a 3D point using accumulated data of past captured images through data sampling of captured images; The control unit randomly extracts a basic feature and three points in space based on RANSAC (Random Sample Consensus), and calculates the distance between the reference plane and arbitrary points calculated based on the extracted three points, searching for a box by using the center point of the box calculated through the extracted three points and whether the center point coincides with the center point; and outputting, by the controller, an open operation signal of the tailgate when the box is recognized through the search and determined to be in a stopped state.

In addition, in the above embodiment, if the box is not recognized or is not in a stationary state through the search, the control unit detects a circle having an arbitrary extraction point as the center point on the two-dimensional surface based on RANSAC, and the vertically projected point and searching for a person using points having a close distance between the center points; when a person is found through the search, calculating, by the controller, the highest point of the points extracted by RANSAC as the height of the person; The control unit randomly extracts two points in space based on RANSAC, calculates the distance between the reference straight line calculated based on the two extracted points and arbitrary points, and searches for the arm based on the detected straight line. ; and when an arm is found as a result of the search, tracking a change in inclination of the arm to recognize an open gesture, and outputting, by the controller, an opening/closing operation signal of the tailgate according to the recognition result.

In addition, the control unit according to the embodiment calculates an angle with the Y-axis by calculating the direction vector of the detected straight line with the vertical Z-axis, and the most recent angle of the calculated angle is in the range of 0° to 60°, It is characterized in that the calculated relative angle counts the number of straight lines in the range of -45° to 0°, and determines as an open gesture if it is greater than or equal to a certain value.

The present invention has the advantage of recognizing a human gesture by recognizing a box possessed by a user, recognizing a person, estimating a recognized person's height, and recognizing a human arm based on RANSAC.

In addition, the present invention has the advantage of controlling the opening and closing of the tailgate of the vehicle through the user's gesture recognition and obstacle recognition.

1 is a block diagram illustrating a gesture recognition system according to an embodiment of the present invention.
2 is a flowchart illustrating a recognition process using a gesture recognition system according to an embodiment of the present invention.
3 is a flowchart illustrating an obstacle recognition process in the gesture recognition process according to the embodiment of FIG. 2 .
4 is a flowchart illustrating a box recognition process in the gesture recognition process according to the embodiment of FIG. 2 .
5 is a flowchart illustrating a human recognition process in the gesture recognition process according to the embodiment of FIG. 2 .
6 is an exemplary diagram illustrating a data pre-processing process in the recognition process according to the embodiments of FIGS. 3 and 4 .
7 is an exemplary diagram illustrating a box recognition process in the embodiment of FIG. 4 .
FIG. 8 is an exemplary diagram illustrating a human recognition process in the embodiment of FIG. 5 .
9 is an exemplary view illustrating an arm recognition process in the embodiment of FIG. 5 .

Hereinafter, the present invention will be described in detail with reference to preferred embodiments of the present invention and the accompanying drawings.

Prior to describing the specific contents for carrying out the present invention, it should be noted that components not directly related to the technical gist of the present invention are omitted within the scope of not disturbing the technical gist of the present invention.

In addition, the terms or words used in the present specification and claims have meanings and concepts consistent with the technical idea of the invention based on the principle that the inventor can define the concept of an appropriate term to best describe his invention. should be interpreted as

In the present specification, the expression that a part "includes" a certain element does not exclude other elements, but means that other elements may be further included.

Also, terms such as “… unit”, “… group”, and “… module” mean a unit that processes at least one function or operation, which may be divided into hardware, software, or a combination of the two.

In addition, the term "at least one" is defined as a term including the singular and the plural, and even if the term "at least one" does not exist, each element may exist in the singular or plural, and may mean the singular or plural. will be self-evident.

In addition, that each component is provided in singular or plural may be changed according to an embodiment.

Hereinafter, a preferred embodiment of a gesture recognition system and method according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing a gesture recognition system 100 according to an embodiment of the present invention. As shown in FIG. 1 , the gesture recognition system 100 recognizes a box possessed by a user based on RANSAC, recognizes a person, A camera unit 110 , a control unit 120 , a data storage unit 130 , a communication unit 140 , and a power supply unit 150 to recognize a human gesture by estimating the recognized height of a person and recognizing a human arm ) is included.

The camera unit 110 is configured to output a photographed image around the vehicle including distance information to the object in a Time of Flight (ToF) method. The distance information is calculated by the TOF method for calculating .

Also, the camera unit 110 may acquire depth information in units of pixels of a CCD camera image, and may preferably be configured as a depth camera.

The control unit 120 generates 3D point cloud data based on the location of 3D points using accumulated data of past captured images through data sampling of captured images output from the camera unit 110, and RANSAC (Random Sample) Consensus) and extracts basic features based on the RANSAC, and recognizes a box possessed by a person from the photographed image, recognizes a person, estimates the height of a recognized person, recognizes a person's arm , and outputs an operation control signal of the tailgate according to the determination result.

Through a pre-processing process of generating 3D point cloud data based on the position of 3D points using accumulated data of past captured images through data sampling of the captured images, for example, 320 × 240 resolution to 160 × 120 resolution As 16-bit depth data is sampled and stored, and data of the previous image frame is accumulated and stored in the data storage unit 130 .

In addition, when the 16-bit depth data is within the effective range, the controller 120 calculates the 3D position by multiplying the light stem unit vector of the pixel position, and finally generates 3D point cloud data as shown in FIG. 6 .

In addition, the control unit 120 takes the end (end) of the rear bumper of the vehicle as the origin of the coordinate system (right-hand coordinate system), +X is the right direction of the vehicle, +Y is the front direction of the vehicle, +Z is the height direction of the vehicle do it with

In addition, the controller 120 establishes a model that satisfies randomly extracted sample data based on RANSAC in order to extract basic features.

In addition, if the number of data adjacent to the established model is greater than that of the previous model, the controller 120 stores a large number of generated models in the data storage unit 130 .

In addition, the control unit 120 repeats the establishment of a model satisfying the randomly extracted sample data based on the RANSAC a predetermined number of times, and then returns the model with the largest number or the largest number of supported data as a final result.

In addition, the control unit 120 randomly extracts three points in space based on RANSAC to perform box recognition.

In addition, the control unit 120 calculates an arbitrary plane equation using the extracted three points, and calculates the distance between all points and the plane by using the plane calculated through the plane equation as a reference plane.

In addition, as shown in FIG. 7A , the controller 120 counts +1 when the distance to the reference plane is less than or equal to the In Area, and counts -2 when the distance to the reference plane is less than or equal to the Out Area.

In addition, the control unit 120 calculates the center of gravity of the box through the extracted three points, and recognizes the box as a box by using whether the center points match.

That is, the control unit 120 calculates the center of gravity and the center of the minimum/maximum coordinates for each X, Y, and Z direction, and when the centers are almost identical, it can be recognized as a box as indicated in red in FIG. 7(b). .

Also, when it is determined that the box is a box, the controller 120 may remove the points determined as the box from the list so that the recognition accuracy of a person can be increased.

In addition, the control unit 120 detects a circle having an arbitrary extraction point as a center point in two dimensions based on RANSAC to perform human recognition, and calculates a distance between a vertically projected point and the center point, , the closer the distance, the more recognizable as a person.

That is, assuming that the Z coordinates of all points are '0', using a circle centered on an arbitrary extracted point as a model, calculate the distance from the center point for all points projected on the XY plane, and Fig. 8(a) ), if the distance from the center point is IN or less, +1 is counted, and if the distance from the center point is OUT or less, -2 is counted. can be judged as

In addition, the control unit 120 determines the highest point of the point extracted by RANSAC as the height of the person, measures the head width and shoulder width based on the width from the highest point to a predetermined distance below, and the measured head width and shoulder width If the width is within a certain range, it can be recognized as a human being.

In addition, when the radius of the circle used for recognizing a person is large, the control unit 120 may remove the arm by processing it as a body, so that only a range narrower than the radius used in clustering may be recognized as a body and removed.

In addition, the controller 120 randomly extracts two points in space based on RANSAC to perform human arm recognition.

In addition, the control unit 120 calculates the distance between the reference straight line calculated by deriving a straight line equation based on the two extracted points and all points.

That is, the control unit 120 counts +1 when the distance to the straight line is less than IN, and counts -2 when the distance to the straight line is less than or equal to OUT, as shown in FIG. Recognize the arm in the green area.

In addition, when the direction vector of the straight line includes the -Y direction, the control unit 120 reverses the direction to unify it in the +Y direction, and removes the points in the arm area for recognition of both arms to perform arm recognition once again. can be done

In addition, in order to recognize the gesture, the controller 120 calculates an angle with the Y axis by performing a dot operation on the direction vector of a straight line of the arm with respect to a predetermined number of image frames in the past with the Z axis in the vertical direction, and the calculated angle If the most recent angle of is in the range of 0° to 60°, the number of straight lines with a relative angle in the range of -45° to 0° among the frames calculated in the past is counted. do.

In addition, the control unit 120 controls a preset angle while the tailgate is being opened, for example, in a range of 0.9 m to the left and right, and a point located below the angle of 50° in the range of tailgate length×Cos 50° to the rear side. When a predetermined number or more is detected, it is determined as an obstacle and an operation stop signal of the tailgate may be output.

The data storage unit 130 is configured to store the generated 3D point cloud data and RANSAC-based operation information, and may be configured as a storage means such as a memory.

The communication unit 140 transmits the tailgate operation signal output from the control unit 120 to the vehicle control unit (not shown), and may use a vehicle communication protocol such as CAN-Bus or Ethernet.

The power supply unit 150 supplies power for operation to the gesture recognition system 100 .

The following describes a recognition method using a gesture recognition system according to an embodiment of the present invention.

2 is a flowchart illustrating a recognition process using a gesture recognition system according to an embodiment of the present invention, FIG. 3 is a flowchart illustrating an obstacle recognition process in the gesture recognition process according to the embodiment of FIG. 2 , and FIG. 4 is FIG. is a flowchart illustrating a box recognition process in a gesture recognition process according to an embodiment, and FIG. 5 is a flowchart illustrating a human recognition process in the gesture recognition process according to the embodiment of FIG. 2 .

1 to 5 , the control unit 120 controls a photographed image input through the camera unit 110 that outputs a photographed image around the vehicle including distance information to an object in a Time of Flight (ToF) method. By analyzing the data, a tailgate opening gesture signal is detected ( S100 ).

In step S100, the control unit 120 detects the open/closed state of the tailgate (S200) and determines whether the tailgate is being opened (S300).

If the tailgate is being opened in step S300, the control unit 120 performs an obstacle recognizing step (S400), and when the tailgate is closed, performs a gesture and box recognizing step (S500).

In step S400, the control unit 120 receives (S410) a photographed image around the vehicle including distance information from the camera unit 110 of the Time of Flight (ToF) method to the object.

In addition, the control unit 120 performs a data pre-processing process (S420) of sampling the data of the captured image received in step S410, and uses the accumulated data of the past captured images to obtain 3D point cloud data based on the position of the 3D point. Performs a three-dimensional reconstruction step (S430) to generate a.

Subsequently, the controller 120 sets the upper region of the tailgate as a region of interest (ROI) (S440), and detects an obstacle within the ROI (S450).

When an obstacle is detected in step S450, the controller 120 calculates an angle between the detected obstacle and the lower end of the tailgate (S460), and compares the calculated angle with a preset reference value (S470).

As a result of the comparison in step S470 , if the calculated angle is less than or equal to a preset reference value, the control unit 120 outputs a stop signal of the tail gator to the vehicle control unit through the communication unit 140 ( S480 ).

In step S500, the control unit 120 receives (S510) a photographed image around the vehicle including distance information from the camera unit 110 of the Time of Flight (ToF) method to the object.

In addition, the control unit 120 performs a data preprocessing process (S520) of sampling data of the captured image received in step S510, and 3D point cloud data based on the position of a 3D point using accumulated data of past captured images. Performs a three-dimensional reconstruction step (S530) to generate.

Subsequently, the control unit 120 randomly extracts a basic feature and three points in space based on RANSAC (Random Sample Consensus), a reference plane calculated based on the extracted three points, and random points A box is searched using whether the center of gravity of the box calculated through the three extracted points by calculating the distance of and the center point coincides with each other (S540).

If box recognition is successful through the search in step S540 (S550), the control unit 120 determines whether it is in a stopped state (S560).

If the determination result of step S560 is a stop state, the control unit 120 outputs a tailgate opening operation signal through the communication unit 140 ( S570 ).

On the other hand, if the box is not recognized through the search in step S540 or is not in a stationary state in step S560, the control unit 120 detects a circle having an arbitrary extraction point as the center point on the two-dimensional surface based on the RANSAC, and vertical A person is searched for ( S510 ′) by using points that are close in distance between the projected point in the direction and the central point.

If the human recognition is successful through the search in step S510' (S520'), the controller 120 calculates the highest point of the points extracted by the RANSAC as the height of the person (S530').

In addition, the controller 120 may remove the body region when recognizing a person ( S540 ′).

Subsequently, the control unit 120 randomly extracts two points in space based on RANSAC, calculates the distance between the reference straight line calculated based on the two extracted points and arbitrary points, and based on the detected straight line to search for the arm (S550').

As a result of the search in step S550', it is determined whether arm recognition is successful (S560'), and if the arm recognition is successful, the arm area is removed (S561'), and the step S550' is repeated a predetermined number of times.

In addition, if the arm is not recognized in step S560', the controller 120 recognizes the open gesture by tracking the change in inclination of the arm (S570'), and determines whether the gesture recognition is successful (S580').

The open gesture calculates an angle with the Y axis by calculating the direction vector of the detected straight line with the vertical Z axis, and the most recent angle of the calculated angle is, for example, in the range of 0° to 60°, If the calculated relative angle counts the number of straight lines in the range of -45° to 0° and exceeds a certain value, it is determined as an open gesture.

As a result of the recognition in step S580', if the gesture recognition is successful, the control unit 120 outputs a tailgate opening/closing operation signal to the vehicle control unit (not shown) through the communication unit 140 (S590') so that the tailgate is opened. do.

Therefore, based on RANSAC, it is possible to recognize a human gesture by recognizing a box possessed by the user, recognizing a person, estimating the height of a recognized person, and recognizing a human arm. It becomes possible to control the opening and closing of the gate.

As described above, although described with reference to the preferred embodiment of the present invention, those skilled in the art can variously modify and change the present invention within the scope without departing from the spirit and scope of the present invention described in the claims below. You will understand that it can be done.

In addition, the reference numbers described in the claims of the present invention are only described for clarity and convenience of description, and are not limited thereto, and in the process of describing the embodiment, the thickness of the lines shown in the drawings or the size of components, etc. may be exaggerated for clarity and convenience of explanation.

In addition, the above-mentioned terms are terms defined in consideration of functions in the present invention, which may vary depending on the intention or custom of the user or operator, so the interpretation of these terms should be made based on the content throughout this specification. .

In addition, even if it is not explicitly shown or described, a person of ordinary skill in the art to which the present invention pertains can make various modifications including the technical idea according to the present invention from the description of the present invention. Obviously, this still falls within the scope of the present invention.

In addition, the above embodiments described with reference to the accompanying drawings have been described for the purpose of explaining the present invention, and the scope of the present invention is not limited to these embodiments.

100: Gesture recognition system.
110: camera unit
120: control unit
130: data storage unit
140: communication department
150: power unit

Claims (13)

a camera unit 110 for outputting a photographed image around a vehicle including distance information to an object in a Time of Flight (ToF) method;
Generates 3D point cloud data based on the location of 3D points using accumulated data of past captured images through data sampling of the captured images, extracts basic features based on RANSAC (Random Sample Consensus), and the RANSAC based on the recognition of a box possessed by a person, recognition of a person, estimation of a recognized height of a person, and a human arm from the photographed image to determine an open gesture according to the angle of the arm, and control the operation of the tailgate according to the determination result a control unit 120 for outputting a signal;
a data storage unit 130 for storing the generated three-dimensional point cloud data and RANSAC-based operation information; and
and a communication unit 140 that transmits an operation signal of the tailgate to a vehicle control unit. The method of claim 1,
The basic feature is that the controller 120 establishes a model satisfying randomly extracted sample data based on RANSAC, stores a model with a large number of data close to the established model, and repeats a predetermined number of times and then supports data A gesture recognition system, characterized in that the model with the largest number is returned as the final result. 3. The method of claim 2,
In the box recognition, the control unit 120 randomly extracts three points in space based on RANSAC, calculates the distance between the reference plane and arbitrary points calculated based on the extracted three points, and the extracted A gesture recognition system, characterized in that by calculating the center of gravity of the box through three points, and recognizing it as a box using whether the centers match. 4. The method of claim 3,
In the recognition of the person, the control unit 120 detects a circle having an arbitrary extraction point as a center point in two dimensions based on RANSAC, and calculates the distance between the vertically projected point and the center point. A gesture recognition system, characterized in that it recognizes a person. 5. The method of claim 4,
The control unit 120 determines the highest point of the point extracted by RANSAC as the height of the person, measures the head width and shoulder width based on the width from the highest point to a certain distance below, and the measured head width and shoulder width are Gesture recognition system, characterized in that the recognition as a human within a certain range. 6. The method of claim 5,
In the recognition of the human arm, the controller 120 randomly extracts two points in space based on RANSAC, and calculates the distance between the reference straight line calculated based on the two extracted points and arbitrary points, so that the distance is close. Gesture recognition system, characterized in that it recognizes the recorded arm. 7. The method of claim 6,
The control unit 120 calculates the angle with the Y axis by calculating the direction vector of the straight line with the Z axis in the vertical direction, and the most recent angle of the calculated angle is in the range of 0° to 60°, and the relative angle calculated in the past A gesture recognition system, characterized in that by counting the number of straight lines in the range of -45° to 0°, and determining it as an open gesture if it exceeds a certain value. 8. The method of claim 7,
The control unit 120 determines that an obstacle is an obstacle and additionally outputs an operation stop signal of the tailgate when a predetermined number of points located below a preset angle are detected while the tailgate is being opened. a) The control unit 120 analyzes data of a captured image input through the camera unit 110 that outputs a captured image around the vehicle including distance information to an object in a Time of Flight (ToF) method to open the tailgate detecting a gesture signal; and
b) the control unit 120 detects the open/closed state of the tailgate and performs an obstacle recognition step if the tailgate is being opened, and a gesture and box recognition step if the tailgate is closed. 10. The method of claim 9,
The step of recognizing the obstacle may include: receiving, by the controller 120, a photographed image around the vehicle including distance information to an object in a Time of Flight (ToF) method;
generating, by the controller 120, 3D point cloud data based on the location of a 3D point using accumulated data of past captured images through data sampling of captured images;
setting, by the controller 120, an upper region of the tailgate as a region of interest (ROI), and detecting an obstacle within the ROI;
when an obstacle is detected, the controller 120 calculating an angle between the detected obstacle and the lower end of the tailgate and comparing it with a preset reference value; and
and outputting, by the controller 120, an operation stop signal of the tail gator when the calculated angle is less than or equal to a preset reference value. 10. The method of claim 9,
The gesture and box recognition step may include: receiving, by the controller 120, a photographed image around the vehicle including distance information to an object in a Time of Flight (ToF) method;
generating, by the controller 120, 3D point cloud data based on the location of a 3D point using accumulated data of past captured images through data sampling of captured images;
The control unit 120 randomly extracts three points in space and a basic feature based on RANSAC (Random Sample Consensus), and calculates the distance between the reference plane and arbitrary points calculated based on the extracted three points. retrieving a box using the center of gravity of the box calculated through the three extracted points and whether the center point coincides with each other; and
and outputting, by the control unit 120, a tailgate opening operation signal when the box is recognized through the search and determined to be in a stationary state. 12. The method of claim 11,
If the box is not recognized or is not in a stationary state through the search, the controller 120 detects a circle having an arbitrary extraction point as a center point in two dimensions based on RANSAC, and the vertically projected point and the center point searching for a person using points having a close distance therebetween;
when a person is found through the search, calculating, by the controller 120, the highest point of the points extracted by RANSAC as the height of the person;
The control unit 120 randomly extracts two points in space based on RANSAC, calculates the distance between the reference straight line calculated based on the two extracted points and arbitrary points, and moves the arm based on the detected straight line. searching; and
and, if the arm is found as a result of the search, tracking a change in inclination of the arm to recognize an opening gesture, and outputting, by the controller 120, a tailgate opening/closing operation signal according to the recognition result; Recognition method. 13. The method of claim 12,
The control unit 120 calculates an angle with the Y axis by calculating the direction vector of the detected straight line with the vertical Z axis, and the most recent angle of the calculated angle is in the range of 0° to 60°, A gesture recognition method, characterized in that by counting the number of straight lines with a relative angle in the range of -45° to 0°, and determining as an open gesture if it is greater than a certain value.

Images