KR101544022B1 - Method and apparatus for pose tracking - Google Patents

Abstract

A first processor for roughly matching a current frame with a reference frame based on a result of the refinement matching of a previous frame and performing a refinement matching between a current frame to which rough matching is applied and a reference frame; And a second processing unit for tracking the current posture of the object based on the result of the tablet matching.

Description

METHOD AND APPARATUS FOR POSE TRACKING [0002]

The present invention relates to an attitude tracking method and an attitude tracking apparatus, and more particularly, to an attitude tracking method and an attitude tracking apparatus using three-dimensional image matching.

Conventional posture tracking techniques mainly focus on techniques for analyzing two-dimensional image information acquired by a CCD camera to use for attitude recognition. Recently, low-cost RGB-D sensors such as Microsoft's Kinect have been widely used As a result, there is an increasing tendency to use 3D information in posture tracking. The conventional three-dimensional image-based attitude tracking method is a method based on machine learning. Since it is necessary to acquire the image or three-dimensional distance data of all attitudes that the object can take offline, There is a drawback that the side falls.

An object of the present invention is to provide a method of tracking an attitude of an object in real time using an image matching technique and an attitude tracking device using the same.

The technical problem of the present invention is not limited to the above-mentioned technical problems, and other technical problems which are not mentioned can be clearly understood by those skilled in the art from the following description.

According to an embodiment of the present invention, there is provided an orientation tracking apparatus including an image acquiring unit acquiring a current frame of an object, an outline matching unit that roughly matches the current frame with a reference frame based on a refinement matching result of a previous frame, A first processing unit for refining-matching the reference frame and a second processing unit for tracking the current posture of the object based on the refinement matching result.

In an embodiment, the coarse matching may be performed by applying a transformation matrix to the current frame to match the previous frame to the reference frame.

In an embodiment, the tablet matching may be performed using a modified IPP algorithm.

In an embodiment, the image acquiring unit may acquire a three-dimensional distance image.

In an embodiment, the reference frame may be obtained from an image of an object taken at an arbitrary point in time.

In an embodiment, the first processing unit may update the reference frame based on the refinement matching result.

In an embodiment, the second processing unit may remove the noise and background of the current frame acquired by the image acquisition unit.

The driver's face posture tracking apparatus according to an embodiment of the present invention includes an image acquiring unit that acquires a three-dimensional distance image of a driver's face, an outline matching unit that approximates a current frame and a reference frame of the three- A second processing unit for tracking the current posture of the driver's face on the basis of the result of the refinement matching, and a second processing unit for, based on the current face posture of the driver, And a warning sound generating unit for warning the driver of the forward watching neglect.

In an embodiment, the coarse matching may be performed by applying a transformation matrix to the current frame to match the previous frame to the reference frame.

In an embodiment, the tablet matching may be performed using a modified IPP algorithm.

In an embodiment, the reference frame may be a frame in which the driver views the front.

In an embodiment, the first processing unit may update the reference frame based on the refinement matching result.

In an embodiment, the second processing unit may remove the noise and background of the current frame acquired by the image acquisition unit.

In an embodiment, the warning sound generating unit may generate a warning sound to the driver when the current face posture of the driver tracked by the second processing unit is deviated by more than a predetermined angle in one direction.

In the embodiment, even if the tracked driver's face posture is shifted by a predetermined angle or more in one direction, the warning sound generating unit may detect that the direction indicator of the vehicle is turned on in one direction or that the gear of the vehicle is in the reverse gear, neutral gear, In the case of any one gear, it is possible not to generate a warning sound to the driver.

According to an embodiment of the present invention, there is provided a method for tracking an object, comprising the steps of acquiring a current frame of a target object, roughly matching the current frame and a reference frame based on a result of the refinement matching of a previous frame, And refraining-matching the reference frame and tracking the current posture of the object based on the refinement matching result.

In an embodiment, the method further includes a step of registering, as a reference frame of the target, an arbitrary frame obtained from an image of an object photographed at an arbitrary point in time prior to acquiring a current frame of the object .

In an embodiment, the object posture tracking method may further include updating the refinement result to the reference frame after the step of tracking the current posture of the object.

The object attitude tracking method may be implemented as a program that can be executed in a computer on a computer-readable recording medium.

According to an aspect of the present invention, a method of tracking an object's attitude in real time using a video matching technique and an attitude tracking device using the method can be provided.

According to an aspect of the present invention, it is possible to prevent the driver from looking ahead, by tracking the posture of the driver's face in real time.

FIG. 1 is a view for explaining a posture tracking method according to an embodiment of the present invention.
2 is a diagram for explaining the step of acquiring a current frame in more detail.
3 is a diagram for explaining a modified IPP algorithm according to an embodiment of the present invention.
4 is a diagram for explaining a step of updating a reference frame on the basis of a result of refinement matching according to an embodiment of the present invention.
5 is a view for explaining an attitude tracking apparatus according to an embodiment of the present invention.
6 is a view for explaining a driver's face posture tracking apparatus according to an embodiment of the present invention.

Other advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail with reference to the accompanying drawings. The present invention may, however, be embodied in many different forms and should not be construed as being limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the invention to those skilled in the art. Is provided to fully convey the scope of the invention to those skilled in the art, and the invention is only defined by the scope of the claims.

Unless defined otherwise, all terms (including technical or scientific terms) used herein have the same meaning as commonly accepted by the generic art in the prior art to which this invention belongs. Terms defined by generic dictionaries may be interpreted to have the same meaning as in the related art and / or in the text of this application, and may be conceptualized or overly formalized, even if not expressly defined herein I will not.

The terminology used herein is for the purpose of illustrating embodiments and is not intended to be limiting of the present invention. In the present specification, the singular form includes plural forms unless otherwise specified in the specification. As used herein, the terms' comprise 'and / or various forms of use of the verb include, for example,' including, '' including, '' including, '' including, Steps, operations, and / or elements do not preclude the presence or addition of one or more other compositions, components, components, steps, operations, and / or components.

The term 'and / or' as used herein refers to each of the listed configurations or various combinations thereof.

It should be noted that the terms such as '~', '~ period', '~ block', 'module', etc. used in the entire specification may mean a unit for processing at least one function or operation. For example, a hardware component, such as a software, FPGA, or ASIC. However, '~ part', '~ period', '~ block', '~ module' are not meant to be limited to software or hardware. Modules may be configured to be addressable storage media and may be configured to play one or more processors. ≪ RTI ID = 0.0 > Thus, by way of example, the terms 'to', 'to', 'to block', 'to module' may refer to components such as software components, object oriented software components, class components and task components Microcode, circuitry, data, databases, data structures, tables, arrays, and the like, as well as components, Variables. The functions provided in the components and in the sections ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ' , '~', '~', '~', '~', And '~' modules with additional components.

The present invention relates to a method of tracking a current position of a target by matching a reference frame serving as a position determination criterion of a target with a current frame of the target and an apparatus for tracking the current position of the target. Unlike the prior art in which a database containing various attitude information of an object is used to determine the current attitude of an object, in the present invention, any one frame is used as a reference frame for determining the current attitude of the object. For example, in the present invention, one frame viewed from the driver's front may be used as a reference frame to confirm whether or not the driver is looking forward.

In the present invention, when the object is photographed in real time, the posture of the object of the current frame is different from the posture of the object of the previous frame, It is used for rough matching. That is, coarse matching is performed by applying a transform matrix to the current frame to match the previous frame with the reference frame. Then, the current frame to which the coarse matching is applied and the reference frame are refinement-matched. The reference frame is updated by the refinement matching result of the current frame. Also, the refinement result of the current frame is used for coarse matching of the next frame. Hereinafter, the present invention will be described in more detail with reference to the accompanying drawings.

Position tracking method

FIG. 1 is a view for explaining a posture tracking method according to an embodiment of the present invention.

A method of tracing an attitude according to an embodiment of the present invention includes the steps of acquiring a current frame from a target image (S10), roughly matching a current frame and a reference frame based on a result of refinement matching of a previous frame (S20) (S40) of tracking the current posture of the object based on the result of the refinement matching, and a step (S50) of updating the reference frame based on the refinement matching result .

In step S10 of acquiring the current frame from the image of the object, the current frame can be obtained from the image captured by the camera in real time. The camera may be a camera for photographing a three-dimensional distance image. The current frame can be obtained from the image stored in the storage device. Step S10 of acquiring the current frame from the image of the object will be described in more detail with reference to Fig.

In step (S20) of roughly matching the current frame and the reference frame based on the result of the refinement of the previous frame, the coarse matching may be performed by applying a transform matrix to the current frame to match the previous frame with the reference frame. The coarse matching can be performed by the first processing unit. The first processing unit may be an image processing unit (GPU).

The refinement matching can be performed using the modified IPP algorithm in the refinement matching step S30 of the current frame to which the coarse matching is applied and the reference frame. The tablet matching can be performed by the first processing unit. The modified IPP algorithm will be described in detail with reference to FIG.

The current posture of the object can be tracked based on the transformation matrix that matches the current frame with the reference frame in step S40 of tracking the current posture of the object based on the refinement matching result. The current posture tracking of the object can be performed by the second processing unit. The second processing unit may be a central processing unit (CPU).

In step S50 of updating the reference frame based on the refinement matching result, the reference frame is updated using the current frame to which the refinement matching is applied. When the reference frame is updated while accumulating the refinement matching result every time the refinement matching is performed, the updated reference frame includes the region of the object which was not included in the first reference frame. As a result, in the subsequent refinement matching, the overlapping area between the current frame and the reference frame is increased, and the matching success rate can be improved. The update of the reference frame may be performed by the first processing unit. A method of updating the reference frame based on the refinement matching result will be described in detail with reference to FIG.

Although not shown in the figure, after step S40 of tracking the current posture of the object based on the result of the refinement matching, various post-processing can be performed based on the tracking result. In the embodiment, the object may be the face of the driver who drives the vehicle, and the driver's current face posture may be tracked by using the scene in which the driver strolls forward and gazes forward, as a reference frame, thereby preventing the driver from overlooking the forward direction. In an embodiment, the method may be utilized for rehabilitation of the patient. The operation range of the patient's joints can be confirmed by tracking the patient's posture using the arbitrary posture of the patient as a reference frame.

FIG. 2 is a diagram for explaining step (S10) of acquiring a current frame in more detail.

The step of acquiring a current frame S10 includes a step S12 of acquiring an original frame from an image of the object, a step S14 of preprocessing the acquired original frame, a step S16 of registering the preprocessed original frame as a current frame, . A source frame is a frame that is acquired directly from the target image and has not been subjected to any image processing.

As described with respect to step S10 in Fig. 1, the image of the object may be the image of the object photographed by the camera in real time. The image of the subject can be obtained through a method such as reading from the recording medium, receiving from an external device, etc. in addition to the image directly shot by the camera. The original frame can be obtained from the image of the object.

Step S14 of preprocessing the original frame may be performed by the second processing unit. Preprocessing of the original frame involves removing noise and background. The image matching can be performed by the object other than the background by removing the background from the original frame, and the accuracy of matching can be improved.

When the second processing unit transmits the preprocessed original frame to the first processing unit, the first processing unit acquires the preprocessed original frame as a current frame (S16) so as to match with the reference frame.

In the present invention, the reference frame is a frame that serves as a reference for attitude tracking of the object in the current frame. The reference frame can be obtained from an image of an object taken at an arbitrary point in time by a method similar to the acquisition of the current frame described in Fig. From the image of the object, one frame containing the reference posture of the object is selected and the reference frame is obtained by removing noise and background from the selected frame in the same manner as described in Fig. In the present invention, one frame selected from an image of an object is used as a reference frame, and the current posture of the object can be tracked through image matching without establishing a database of the posture of the object.

Hereinafter, approximate matching and refinement matching according to an embodiment of the present invention will be described.

outline coordination

The coarse matching according to the embodiment of the present invention can be performed by applying a transformation matrix to the current frame to match the previous frame with the reference frame. The coarse matching is performed by the following equation (1).

는 k번째 프레임,

는 개략 정합이 적용된 k번째 프레임, 변환 행렬

은 이전 프레임

을 기준 프레임(

)로 정합하는 상대 변환을 의미한다. 즉, k번째 프레임(

)을 현재 프레임이라고 하면, 현재 프레임(

)에 대해 이전 프레임을 기준 프레임으로 정합하는 변환 행렬을 적용함으로써 개략 정합이 수행된다. 그 결과 개략 정합이 적용된 현재 프레임(

)을 얻을 수 있다. 이전 프레임을 기준 프레임으로 정합하는 변환 행렬

은 이전 프레임의 정제 정합 결과로부터 구해질 수 있다.In Equation (1)

Th frame,

K < / RTI > frame to which coarse matching is applied,

The previous frame

To a reference frame (

). ≪ / RTI > That is, the kth frame (

) Is the current frame, the current frame

), A coarse matching is performed by applying a transformation matrix that matches the previous frame with the reference frame. As a result, the current frame to which the coarse matching is applied

) Can be obtained. A transformation matrix that matches the previous frame with the reference frame

Can be obtained from the refinement matching result of the previous frame.

According to an embodiment of the present invention, the tablet matching can be performed by a modified IPP algorithm. Prior to describing the refinement matching, the modified IPP algorithm will be described in more detail with reference to FIG.

transform IPP  algorithm

3 is a diagram for explaining a modified IPP algorithm according to an embodiment of the present invention.

To match two images, a matching point between two images must be found. The existing IPP algorithm repeatedly executes the algorithm and selects the convergent 3D point as a matching point for matching. On the other hand, the modified IPP algorithm according to the embodiment of the present invention can shorten the time required for matching point estimation by determining a matching point through only one estimation process while using the advantages of the existing IPP as it is . The purpose of the modified IPP algorithm is to find the point Q ', which is the matching point of the point P, to match the S and D surfaces. The modified IPP algorithm is briefly described with reference to Fig.

First, a two-dimensional image coordinate (P _D ) is obtained by projecting a three-dimensional point P on the S curved surface to a two-dimensional image of the D curved surface. And D on the three-dimensional curved surface projection point (Q) corresponding to the two-dimensional image coordinates (P _D) is estimated using the distance image (I _D) and the two-dimensional image coordinates (P _D) of the D-form surface. After constructing the tangent plane of the estimated three-dimensional projection point Q ', a point at which the point P and the tangent plane meet vertically is selected as the matching point.

If the k matching points {P} and {Q}} are determined on the S curved surface and the D curved surface, a three-dimensional transformation matrix T minimizing the matching error (e) Can be obtained using Singular Value Decomposition (SVD).

refine coordination

)은 이어서 기준 프레임과 정제 정합된다. 기준 프레임(

)과 개략 정합이 적용된 현재 프레임(

)이 주어지면, 변형 IPP 알고리즘을 이용하여 다음의 수학식 3을 만족하는 변환 행렬(

)를 계산할 수 있다.Current frame to which coarse matching is applied (

) Is then refined to the reference frame. Reference frame (

) And the current frame to which the coarse matching is applied

) Is given, a transform matrix (i, j) that satisfies the following equation (3) using the modified IPP algorithm

) Can be calculated.

와

는 각각 개략 정합이 적용된 현재 프레임(

)과 기준 프레임(

) 사이에 변형 IPP 알고리즘으로 매칭되는 3차원 점들을 말하며,

는

에 대응하는 법선 벡터를 의미한다.In Equation 3,

Wow

Is a current frame ("

) And a reference frame

) Is a three-dimensional point matched by a modified IPP algorithm,

The

Quot; normal vector ".

)는 아래 수학식 4에 의해 구해진다.Since the current posture means how much conversion has occurred based on the reference frame, the current posture

) Is obtained by the following equation (4).

)를 이용하여, 상술한 바와 같은 운전자 얼굴 자세 추적 등의 후처리를 할 수 있다.
Tracked current posture (

), It is possible to perform post-processing such as the above-described driver's face posture tracking and the like.

4 is a diagram for explaining a step of updating a reference frame on the basis of a result of refinement matching according to an embodiment of the present invention.

The more matching points between the reference frame and the current frame, the better the match success rate. Therefore, in the embodiment of the present invention, the reference frame is updated based on the result of the refinement matching between the current frame to which the coarse matching is applied and the reference frame.

Referring to FIG. 4, in step 1, the reference frame and the frame 1 are matched. Referring to the one-step matching result in which the reference frame and the frame 1 are matched, a portion U1 that was not included in the reference frame appears in the one-step matching result. According to the embodiment of the present invention, the one-step matching result is updated to the reference frame so that the one-step matching result can be used as the two-step matching reference frame.

In the same manner, step 2 shows the result of matching the updated reference frame and frame 2. Referring to the two-step matching result, the portion U2 that was not included in the reference frame appears in the two-step matching result. When the matching result of each step is updated to the reference frame, an area not included in the first reference frame is included in the reference frame. Accordingly, the number of regions where the current frame and the reference frame overlap each other can be increased, thereby improving the matching success rate.

Attitude tracking device

5 is a view for explaining an attitude tracking apparatus 100 according to an embodiment of the present invention.

In the embodiment of the present invention, the posture tracking apparatus 100 may include an image acquisition unit 110, a first processing unit 130, and a second processing unit 150.

In an embodiment of the present invention, the image acquisition unit 110 may be a camera. The camera can acquire a three-dimensional distance image. As an example, the image acquisition unit 110 may be a kinect of Microsoft Corporation. The image obtaining unit 110 according to the embodiment of the present invention may be a device for directly reading an image recorded on a recording medium, Lt; / RTI >

According to the embodiment of the present invention, the second processing unit 150 can pre-process an image acquired by the image acquiring unit 110. [ The preprocessing involves removing noise from the image and removing the background. The second processing unit 150 may transmit the image from which noise and background are removed to the first processing unit 130. [

According to the embodiment of the present invention, the second processing unit 150 can receive the processing result of the first processing unit 130 and track the current posture of the object. The processing result of the first processing unit 130 includes the matching result of the reference frame and the current frame. The second processing unit 150 can perform various post-processing using the current posture of the tracked object. In the embodiment, the object may be the face of the driver, and the driver's face posture may be tracked to confirm whether or not the driver is looking forward.

According to the embodiment of the present invention, the first processing unit 130 can process the image preprocessed by the second processing unit 150. The processing of the image includes the coarse matching and the refinement matching of the reference frame and the current frame of the image.

According to the embodiment of the present invention, the first processing unit 130 can roughly match the current frame and the reference frame based on the result of the refinement matching of the previous frame. Coarse matching may be performed by applying a transformation matrix to the current frame to match the previous frame to the reference frame.

According to an embodiment of the present invention, the first processing unit 130 may refine-match the current frame to which the coarse matching is applied and the reference frame. Tablet matching may be performed using a modified IPP algorithm.

According to the embodiment of the present invention, the first processing unit 130 can acquire a reference frame. The reference frame can be obtained from an image of an object taken at an arbitrary point in time. The first processing unit 130 may set any one frame selected from the image as a reference frame.

According to the embodiment of the present invention, the first processing unit 130 may update the reference frame using the current frame to which the refinement matching is applied. The updated reference frame is used as a reference frame when matching the next frame.

According to the embodiment of the present invention, the first processing unit 130 transmits the result of the tablet matching to the second processing unit 150, and allows the second processing unit 150 to track the current posture of the object.

According to an embodiment of the present invention, the first processing unit 130 may be an image processing unit (GPU).

5, the first processing unit 130 and the second processing unit 150 are shown as separate devices. However, the first processing unit 130 and the second processing unit 150 may be implemented by a single processing unit .

Driver's face posture tracking device

FIG. 6 is a view for explaining a driver's face posture tracking apparatus 200 according to an embodiment of the present invention.

The driver's face posture tracking apparatus 200 may include an image obtaining unit 210, a first processing unit 230, and a second processing unit 250. In the present embodiment,

In the embodiment of the present invention, the image obtaining unit 210 may be a camera. The camera can acquire a three-dimensional distance image. As an embodiment, the image acquisition unit 210 may be a kinect of Microsoft Corporation. The image acquisition unit 210 according to an embodiment of the present invention may be a device that directly reads an image recorded on a recording medium in addition to an apparatus that acquires an image by directly shooting an object, Lt; / RTI > The image acquisition unit 210 may be disposed at a position where the face of the driver can be photographed from the front.

According to an embodiment of the present invention, the second processing unit 250 can pre-process an image acquired by the image acquisition unit 210. [ The preprocessing involves removing noise from the image and removing the background except for the driver's face. The second processing unit 150 can transmit the image to which the background except for the noise and the driver's face is removed to the first processing unit 230. [

According to the embodiment of the present invention, the second processing unit 250 can receive the processing result of the first processing unit 230 and track the current posture of the driver's face. The processing result of the first processing unit 230 includes the matching result of the reference frame and the current frame.

According to an embodiment of the present invention, the first processing unit 230 may process the image that has been pre-processed by the second processing unit 250. The processing of the image includes the coarse matching and the refinement matching of the reference frame and the current frame of the image.

According to the embodiment of the present invention, the first processing unit 230 can roughly match the current frame and the reference frame based on the refinement matching result of the previous frame. Coarse matching may be performed by applying a transformation matrix to the current frame to match the previous frame to the reference frame.

According to the embodiment of the present invention, the first processing unit 230 may refine-match the reference frame with the current frame to which the coarse matching is applied. Tablet matching may be performed using a modified IPP algorithm.

According to an embodiment of the present invention, the first processing unit 230 may acquire a reference frame. The reference frame can be obtained from an image of an object taken at an arbitrary point in time. The first processing unit 130 may set any one frame selected from the image as a reference frame. In an embodiment, the reference frame may be a frame in which the driver is looking at the front.

According to the embodiment of the present invention, the first processing unit 230 may update the reference frame using the current frame to which the refinement matching is applied. The updated reference frame is used as a reference frame when matching the next frame.

According to the embodiment of the present invention, the first processing unit 230 transmits the result of the tablet matching to the second processing unit 250, and allows the second processing unit 250 to track the current posture of the driver's face.

According to an embodiment of the present invention, the first processing unit 230 may be an image processing unit (GPU).

6, the first processing unit 230 and the second processing unit 250 are shown as separate devices. However, the first processing unit 230 and the second processing unit 250 may be implemented by a single processing unit .

According to the embodiment of the present invention, when the current posture of the driver's face tracked by the second processing unit 250 is shifted by more than a predetermined angle in one direction, the warning sound generating unit 270 may generate a warning sound informing the driver of the forward- have. The predetermined angle may be 45 degrees and the predetermined angle may be changed as needed.

According to the embodiment of the present invention, the warning sound generating unit 270 may not generate a warning sound in consideration of other conditions even when the driver's current face posture is deviated by more than a predetermined angle in one direction. The warning sound generating unit 270 may be connected to the control unit of the vehicle, and may receive the speed, the gear, the state of the turn signal lamp, and the like. Even when the face of the driver is turned to the left, the warning sound generating unit 270 may determine that the driver sees the left side mirror when the left turn indicator lamp of the vehicle is turned on, and may not generate a warning sound. The right side can be applied in the same way. In an embodiment, the warning sound generating unit 270 may not generate a warning sound when it is determined that the vehicle's rear face is looking for a backward gear when the gear of the vehicle is a reverse gear, even if the driver's face posture is changed in one direction, In the case of a neutral gear or a parking gear, it is determined that the vehicle is stopped and no warning sound is generated. The case in which the above-mentioned warning sound is not generated is an exemplary one, and the warning sound generating section 270 can operate in consideration of other external environments.

The posture tracking method according to the embodiment of the present invention described above may be implemented as a program to be executed in a computer and stored in a computer-readable recording medium. The computer-readable recording medium includes all kinds of storage devices in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like.

It is to be understood that the above-described embodiments are provided to facilitate understanding of the present invention, and do not limit the scope of the present invention, and it is to be understood that various modifications may be made within the scope of the present invention. For example, each component shown in the embodiment of the present invention may be distributed and implemented, and conversely, a plurality of distributed components may be combined. Therefore, the technical protection scope of the present invention should be determined by the technical idea of the claims, and the technical protection scope of the present invention is not limited to the literary description of the claims, The invention of a category.

100: Attitude tracking device
110:
130:
150:
200: Facial posture tracking device
210:
230: first processing section
250: second processing section
270:

Claims (19)

An image acquiring unit acquiring a current frame of an object;
A first processor for roughly matching the current frame and the reference frame based on the result of the refinement of the previous frame, and refinement-matching the current frame and the reference frame to which the rough matching is applied; And
A second processing unit for tracking the current posture of the object based on the result of the tablet matching in the first processing unit;
And a posture tracking device.
The method according to claim 1,
Wherein the coarse matching is performed by applying a transformation matrix to the current frame to match the previous frame with the reference frame.
The method according to claim 1,
Wherein the tablet matching is performed using a modified IPP algorithm.
The method according to claim 1,
Wherein the image acquiring unit is capable of acquiring a three-dimensional distance image.
The method according to claim 1,
Wherein the reference frame is obtained from an image of an object at an arbitrary point in time.
The method according to claim 1,
Wherein the first processing unit updates the reference frame based on a result of the refinement matching in the first processing unit.
The method according to claim 1,
And the second processing unit removes noise and background of a current frame acquired by the image acquisition unit.
An image obtaining unit for obtaining a three-dimensional distance image of a driver's face;
A first processor for roughly matching the current frame and the reference frame of the 3D distance image based on the result of the refinement of the previous frame and performing refinement matching of the current frame and the reference frame to which the rough matching is applied;
A second processing unit for tracking the current posture of the driver's face based on the result of the tablet matching in the first processing unit;
A warning sound generating unit for warning the driver of a forward sighting error based on the current face posture of the driver;
The driver's face posture tracking device.
9. The method of claim 8,
Wherein the coarse matching is performed by applying a transformation matrix to the current frame to match the previous frame with the reference frame.
9. The method of claim 8,
Wherein the tablet matching is performed using a modified IPP algorithm.
9. The method of claim 8,
Wherein the reference frame is a frame in which the driver looks in front of the driver.
9. The method of claim 8,
Wherein the first processing unit updates the reference frame based on a result of the refinement matching in the first processing unit.
9. The method of claim 8,
And the second processing unit removes noise and background of a current frame acquired by the image acquisition unit.
9. The method of claim 8,
Wherein the warning sound generating unit generates a warning sound to the driver when the current face posture of the driver tracked by the second processing unit is deviated by more than a predetermined angle in one direction.
15. The method of claim 14,
Even if the tracked driver's face posture is shifted by a predetermined angle or more in one direction, the warning sound generating unit may be configured such that the direction indicator of the vehicle is turned on in one direction, or that the gear of the vehicle is in any one of gears of reverse gear, neutral gear, The driver does not generate a warning sound.
Obtaining a current frame of the object;
Matching the current frame and the reference frame outline based on the result of the refinement of the previous frame;
Refinement-matching the current frame to which the coarse matching is applied and the reference frame; And
Tracking the current posture of the object based on the refinement matching result of the current frame and the reference frame to which the rough matching is applied;
Wherein the method comprises the steps of:
17. The method of claim 16,
Prior to obtaining the current frame of the object
Registering an arbitrary frame obtained from an image of an object at an arbitrary time point as a reference frame of the object;
Wherein the object tracking method further comprises:
17. The method of claim 16,
After the step of tracking the current posture of the object
Updating the ref- erence match result of the current frame and the reference frame to which the raster matching is applied to the reference frame;
Wherein the object tracking method further comprises:
18. A computer-readable recording medium on which a program for performing the method of any one of claims 16 to 18 is recorded.

Images