KR101962872B1 - Smart recognition system and method using hand motion tracking - Google Patents

Abstract

The present invention relates to a smart recognition system and method using hand tracking. More particularly, the present invention relates to a method and apparatus for recognizing a hand using an input image (A) for hand recognition, defining a palm center point in the input image (B) for hand tracking using the hand recognition information, The finger end point in the image B is calculated using the hand recognition information, the palm center point and the finger end point are connected by a line, the length value of the line is compared with the reference distance value of the image A, . Through this, it is possible to maintain a higher detection rate by simple calculation using the input image for hand recognition in the tracking of the hand movement. This can be applied to various security systems using the tracking of the hand gestures, and further applied to augmented reality technology.

Description

Technical Field [0001] The present invention relates to a smart recognition system and a method using hand motion tracking,

The present invention relates to a smart recognition system and method using a tracking of a hand gesture, and is a technology for tracking a hand gesture through hand recognition, palm center definition, and finger end point calculation. In particular, the present invention has a feature of tracking a hand motion in a continuous image using an input image for hand recognition.

Recently, Samsung and Apple have been talking about technologies that recognize and identify human body with smartphone security system. Various researches have been conducted on such topic backgrounds, such as recognizing and discriminating objects, and continuously tracking continuous images. However, the color-based hand recognition technology has a low detection rate due to the color of the skin or surrounding objects. Specifically, there is a limitation in that the detection accuracy is deteriorated due to the existence of objects having different skin colors and skin colors according to various races. In order to overcome these limitations, the hand recognition technology has been applied not only to the color base, but also to techniques such as filtering after collecting the samples in advance, or estimating the skeleton of the hand. However, the above-described technique is complicated compared to the conventional color-based hand recognition technology, and thus has a problem in that it takes much recognition time.

United States Patent Application Publication No. 9524028 Korean Patent Registration No. 1541421

Accordingly, the present invention has been made in view of the above-mentioned matters, and aims to maintain a higher detection rate by simple calculation using an input image for hand recognition in tracking of the hand movement.

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

According to an aspect of the present invention, there is provided a smart recognition system using hand tracking, which includes a hand recognition unit for recognizing a hand using an input image A for hand recognition, A center tracking unit for defining a center point of the palm in the input image B for tracking the hand using the information, a hand centering point defined by the center tracking unit and hand recognition information recognized by the hand recognition unit, ), A center point and an end point connection line connecting the calculated palm center point and the calculated finger end point with each other, a reference distance value of the image (A) is compared with the length value of the line, And a comparison and output unit for outputting a result.

The hand recognition unit may include a first candidate detection unit for detecting a hand region candidate group by binarizing and labeling the image A, a region having the largest area among the detected hand region candidate group as a hand region, A first contour extracting unit for extracting a contour of a hand from the extracted hand region, a corner extracting unit extracting a corner point of the hand contour from the extracted hand contour by using a Convex Hull, A first center-point defining portion defining a center point of the defined rhombus as a central point of the palm after defining a rhombus,

A finger region and a reference distance value defining unit for defining an area of the finger from an angle of a line connecting the vertex of the extracted hand contour and the center point of the defined palm and defining a reference distance value from the length of the line, And a recognition unit recognizing the hand using the defined area of the finger and the reference distance value using the Convexity Defect of the finger.

The center tracking unit may include a second candidate group detection unit for binarizing and labeling the image B to detect a hand region candidate group, a region having a largest area among the detected hand region candidate groups as a hand region, A second contour extracting unit for extracting a contour of a hand from an area of the image B by moving the contour of the image A in at least one of left, right, up, down, and diagonal directions so as to include rhombus of the image A, And a second center-point defining unit for defining a center point of the moved and adjusted rhombus as a palm center point in the image (B).

The end point calculation unit includes a line connection unit for connecting a center point of the palm within the defined image B with a finger corner point of the image A by a line, a contact extraction unit for extracting a contact point between the connected line and a hand contour line in the image B, (B) a point having a largest distance value between the extracted contact point and the palm center point in the defined image (B) while moving the extracted contact point along the hand contour line in the image (B) And a fingertip point calculating unit for calculating a fingertip end point within the finger. When the calculated finger point is further moved by a predetermined distance to extract a point having a distance value with respect to the central point of the palm within the defined image B that is larger than the calculated finger end point, To a fingertip end point within the finger.

The comparison and output unit determines that the finger in the image B is folded if the length value of the line is smaller than the reference distance value and if the length value of the line is greater than or equal to the reference distance value, .

In order to achieve the above object, a smart recognition method using a hand movement tracking according to the technical idea of the present invention includes a hand recognition step of recognizing a hand using an input image (A) A center tracking step of defining a center point of the palm in the image (B) input for hand tracking using the hand recognition information of the hand recognition step in the center part, a central tracking step of calculating a palm center point defined in the center tracking step, An end point calculation step of calculating a finger end point in the image B using the hand recognition information recognized in the step of connecting the center point and the end point connection line connecting the defined palm center point and the calculated finger end point in a line at the center point and end point connection, , Comparing the length value of the line and the reference distance value of the image (A) at the comparison and output section, Characterized in that it comprises a step of comparing and the output force.

In this case, the hand recognition step may include: detecting a hand region candidate group by binarizing and labeling the image (A) in the first candidate group detection unit, detecting a hand region candidate group by a first contour extracting unit, Extracting a hand outline from the hand region defined above, extracting a vertex of a hand outline from the extracted hand outline using a convex hull in a vertex extracting unit, Defining a largest rhombus in the extracted hand contour in a center point defining part and then defining a center point of the defined rhombus as a center point of the palm; From the angle of the line connecting the center of the palm and the defined center of the palm, A step of defining a quasi-distance value, and a step of recognizing the hand using the defined area of the finger and the reference distance value using a Convexity Defect based on the convex hole in the recognition unit. do.

The center tracking step may include detecting a hand region candidate group by binarizing and labeling the image B in a second candidate group detection unit, detecting a region having the largest area among the detected hand region candidate groups in the second contour extracting unit, (A) in the image (B) in a moving and scaling unit, the moving and scaling unit may include a step of dividing the rhombus into left and right, upper and lower, and diagonal lines so that the rhombus of the image (A) And defining a center point of the moved and adjusted rhombus as a palm center point in the image B in the second center point defining unit.

The endpoint calculation step may include connecting a center point of the palm within the image B defined by the line connection unit and the finger vertex of the image A with a line, connecting the connected line and the hand outline in the image B, (B), the distance between the extracted contact point and the central point of the palm of the defined image (B) is the largest distance between the extracted contact point and the palm center point of the defined image And calculating a point having a value as a finger end point in the image (B). When the calculated finger point is further moved by a predetermined distance to extract a point having a distance value with respect to the central point of the palm within the defined image B that is larger than the calculated finger end point, To a fingertip end point within the finger.

The comparing and outputting step determines that the finger in the image B is collapsed when the length value of the line is smaller than the reference distance value in the comparing and outputting unit and when the length value of the line is greater than or equal to the reference distance value, ) Of the finger is found to be unfolded.

According to the smart recognition system and method using the tracking of the hand movement as described above,

First, in the tracking of the hand movement, a higher detection rate can be maintained through a simple calculation using an input image for hand recognition.

Second, it can be applied to various security systems using hand tracking.

Third, it has wide effect such as application to augmented reality technology which shows virtual objects superimposed on the real world.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a smart recognition system using tracking of a hand gesture according to an embodiment of the present invention; FIG.
FIG. 2A is a block diagram for explaining a hand recognition unit in FIG. 1; FIG.
FIG. 2B is a view for explaining the center tracking unit in FIG. 1. FIG.
FIG. 2C is a schematic diagram for explaining the end point calculating unit in FIG. 1; FIG.
3 is a flowchart illustrating a smart recognition method using a hand movement tracking according to an embodiment of the present invention.
FIG. 4A is a flowchart for explaining the hand recognition step in FIG. 3; FIG.
FIG. 4B is a flowchart for explaining the center tracking step in FIG. 3; FIG.
4C is a flowchart for explaining an end point calculation step in FIG. 3; FIG.
5 is a view illustrating a hand movement tracking process according to an embodiment of the present invention.
Figure 6 illustrates binarization in accordance with one embodiment of the present invention.
7 illustrates movement and sizing of a diamond in accordance with an embodiment of the present invention.
8 is a diagram illustrating a process for calculating a finger end point according to an embodiment of the present invention.
9 is a diagram showing a result of applying the present invention to an augmented reality as an embodiment of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A smart recognition system and method using a hand movement tracking according to embodiments of the present invention will be described in detail with reference to the accompanying drawings. The present invention is capable of various modifications and various forms, and specific embodiments are illustrated in the drawings and described in detail in the text. It is to be understood, however, that the invention is not intended to be limited to the particular forms disclosed, but on the contrary, is intended to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. Like reference numerals are used for like elements in describing each drawing. In the accompanying drawings, the dimensions of the structures are enlarged to illustrate the present invention, and are actually shown in a smaller scale than the actual dimensions in order to understand the schematic structure. Also, the terms first and second, etc. may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component. On the other hand, unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the contextual meaning of the related art and are to be interpreted as either ideal or overly formal in the sense of the present application Do not.

FIG. 1 is a block diagram of a smart recognition system using tracking of a hand gesture according to an embodiment of the present invention.

As shown in FIG. 1, a smart recognition system using hand movement tracking includes a hand recognition unit 100 for recognizing a hand using an input image A for hand recognition, a hand recognition unit 100 for recognizing a hand, A central tracking unit 200 for defining a center point of the palm in the input image B for tracking the hand using the information, a center point of the palm defined in the center tracking unit 200, An end point calculation unit 300 for calculating a fingertip point in the image B using the hand recognition information, a center point and an end point connection unit 400 connecting the defined palm center point and the calculated finger end point with a line, And a comparison and output unit 500 that compares the reference distance value of the image A with the reference distance value of the image A and outputs a hand tracking result. At this time, the comparison and output unit 500 determines that the finger in the image B is folded if the length value of the line is smaller than the reference distance value. On the other hand, if the length value of the line is greater than or equal to the reference distance value, it is determined that the finger in the image B is expanded.

FIG. 2A is a block diagram illustrating a hand recognition unit in FIG. 1. FIG.

2A, the hand recognition unit 100 includes a first candidate group detection unit 110 for detecting a hand region candidate group by binarizing and labeling the image A, a second candidate group detection unit 110 for detecting an area of the hand region candidate group A first contour extracting unit 120 for extracting a hand contour from the defined hand region after defining the largest area as a hand region, a first contour extracting unit 120 for extracting a hand contour from the extracted hand contour using a Convex Hull, A first center point defining unit 140 defining a largest rhombus in the extracted hand contour and then defining a center point of the defined rhombus as a center point of the palm, A finger region defining a region of a finger from an angle of a line connecting a vertex of a contour and a center point of the defined palm and defining a reference distance value from the length of the line, It is constructed using the given distance value defining unit 150, a convex-hole based konbek City defects (Convexity Defect) in the recognition section 160 to recognize a hand through the area and a reference distance value of the defined fingers.

FIG. 2B is a view for explaining the center tracking unit in FIG.

As shown in FIG. 2B, the center tracking unit 200 includes a second candidate group detection unit 210 for binarizing and labeling an image B to detect a hand region candidate group, and a second candidate group detection unit 210 for detecting an area of the hand region candidate group A second contour extracting unit 220 for extracting a hand contour from the defined hand region after defining a largest region as a hand region and a second contour extracting unit 220 for extracting a hand contour of the image A from the hand region, A movement and size adjustment unit 230 for adjusting the size of the rhombus while moving the rhombus in the direction of at least one of left, right, up, down, and diagonal lines, a second center point And a definition unit 240.

FIG. 2C is a block diagram for illustrating the endpoint calculator 300 in FIG.

2C, the end point calculation unit 300 includes a line connection unit 310 for connecting the center of the palm of the defined image B with the finger vertex of the image A by a line, (B), a contact point extracting unit (320) for extracting a contact point to be contacted with a hand contour line in the image (B) And a finger end point calculation unit 330 for calculating a point having the largest distance value from the finger B in the image B as a finger end point. When the calculated finger point is further moved by a predetermined distance to extract a point having a distance value with respect to the central point of the palm within the defined image B that is larger than the calculated finger end point, To the fingertip point in the finger.

FIG. 3 is a flowchart illustrating a smart recognition method using a hand movement tracking according to an embodiment of the present invention.

As shown in FIG. 3, in the smart recognition method using the hand movement tracking, the hand recognition unit 100 recognizes a hand using the input image A for hand recognition (S100). The central tracking unit 200 defines a palm center point in the input image B for hand tracking using the hand recognition information of the hand recognition step S100 at step S200. The end point calculating unit 300 calculates a finger end point in the image B using the palm center point defined in the center tracking step S200 and the hand recognition information recognized in the hand recognition step S100 S300. The calculated center point of the palm and the calculated finger end point are connected by a line at the central point and end point connection unit 400 (S400). The comparison and output unit 500 compares the length value of the line with the reference distance value of the image A (S500). At this time, if the length value of the line is greater than or equal to the reference distance value in the comparison and output unit 500, it is determined that the finger in the image B is expanded and outputted (S610). On the other hand, if the length of the line is smaller than the reference distance value, it is determined that the finger in the image B is folded and outputted (S630).

4A is a flowchart for explaining the hand recognition step in FIG.

As shown in FIG. 4A, in the hand recognition step S100, the first candidate group detection unit 110 binarizes and labels the image A to detect a hand region candidate group (S110). In the first contour extracting unit 120, a region having the largest area among the detected hand region candidates is defined as a hand region, and a hand contour is extracted from the defined hand region (S120). The corner extractor 130 extracts a corner point of the hand contour from the extracted hand contour using Convex Hull (S130). After defining the largest rhombus in the extracted hand contour at the first center point defining unit 140, the center point of the defined rhombus is defined as the center point of the palm (S140). The finger region and the reference distance value defining unit 150 define the area of the finger from the angle of the line connecting the vertex of the extracted hand contour and the center point of the defined palm and define the reference distance value from the length of the line S150). The recognition unit 160 recognizes the hand using the defined area of the finger and the reference distance value using the Convexity Defect based on the Convex hole (S160).

FIG. 4B is a flowchart for explaining the center tracking step in FIG.

As shown in FIG. 4B, in the center tracking step S200, the second candidate group detection unit 210 binarizes and labels the image B to detect a hand region candidate group (S210). In the second contour extracting unit 220, a region having the largest area among the detected hand region candidates is defined as a hand region, and then a hand contour is extracted from the defined hand region (S220). The moving and sizing unit 230 adjusts the size of the image by moving the diamond in at least one of left and right, up and down and diagonal directions so as to include the diamond of the image A in the hand contour line in the image B (S230) . The center point of the moved and adjusted rhombus is defined as the palm center point in the image B at the second center point defining unit 240 at step S240.

4C is a flowchart for explaining the end point calculation step in FIG.

As shown in FIG. 4C, the end point calculation step S300 connects the center point of the palm within the defined image B with the finger's vertex of the image A at the line connection unit 310 at step S310. In step S320, the contact extraction unit 320 extracts a contact point between the connected line and a hand contour line in the image B (step S320). The finger end point calculation unit 330 moves the extracted contact point along the hand contour line in the image B and calculates a point having the largest distance value between the extracted contact point and the palm center point in the defined image B Is calculated as a finger end point in the image B (S330).

FIG. 5 is a flowchart illustrating a hand movement tracking process according to an embodiment of the present invention. FIG. 6 is a diagram illustrating a binarization process according to an embodiment of the present invention.

As shown in FIGS. 5 to 6, in the present invention, an image received using a webcam is processed. In the present invention, the first candidate group detection unit 110 performs binarization according to the skin color based on the YCbCr color space, and labels the image (A) input for hand recognition to detect hand region candidate groups. In this case, the binarization is a preprocessing process for recognizing a hand, which transforms an image into a YCbCr space and performs binarization of surrounding background and objects with different colors. The labeling is performed by attaching the same number to adjacent pixels and assigning different numbers to the other unconnected components. (In order to find out the shape of a binary image, the same pixel values are grouped and numbered will be.)

An area having the largest area among the hand area candidates detected by the first contour extracting unit 120 is defined as a hand area and a hand contour 10 of the image A is extracted from the labeling process.

The corner extractor 130 extracts a corner point 13 of the hand contour of the image A from the hand contour 10 of the image A. [ At this time, the extraction of the corner point 13 of the hand contour of the image A is performed by connecting the outermost points of the hand contour 10 of the image A in a straight line and using a Convex Hull algorithm .

The first center point defining unit 140 defines the largest rhombus 15 within the hand contour 10 of the image A. [ At this time, the center point of the lozenge 15 defined above is defined as the palm center point 20 in the image A. [

The finger region and the reference distance value defining section 150 define the finger distance and the reference distance value from the angle of the first connecting line 19 connecting the vertex 13 of the hand contour of the image A and the palm center point 20 of the image A Defines the area 17 of the finger. A reference distance is defined for each of five finger regions 17 from the length of the first connecting line 19 connecting the vertex 13 of the hand contour of the image A and the palm center point 20 of the image A do.

The recognition unit 160 uses information indicating that the area between the defined five-finger region 17 and the finger is dented. The algorithm used at this time is a Convexity Defect, which finds four points whose distance from the central point of the palm is shorter than the defined reference distance, Define your hand by recognizing it.

In this way, a local search is performed on the palm center point and the finger end point in the input image B for hand tracking using the input image A for hand recognition. This process is shown in FIGS. 7 to 8. FIG.

FIG. 7 is a diagram illustrating movement and size adjustment of rhombus according to an embodiment of the present invention, and FIG. 8 is a diagram illustrating a process for calculating a finger end point according to an embodiment of the present invention.

The second candidate group detection unit 210 and the second contour extraction unit 220 of the present invention are the same as those of the first candidate group detection unit 110 and the first contour extraction unit 120 described above. As shown in FIG. 7, the present invention is characterized in that the movement and size adjuster 230 adjusts the rhombus 15 so that the rhombus 15 of the image A is contained in the hand contour line of the image B, , And the diagonal line, and adjusts its size. 7, when the inside of the diamond 15 is not completely filled with the hand area, the diamond 15 is moved in the direction opposite to the unfilled area. If the inside of the moved rhombus 15 is filled with the hand region so that the rhombus 15 can move in any direction, the size of the rhombus 15 is adjusted. At this time, the size of the diamond 15 is adjusted to the maximum size in which the inside of the diamond 15 is filled with the hand area. The center point of the rhombus 15 in the second center point defining portion 240 is defined as the palm center point 40 in the image B after finding the largest rhombus 15 by modifying the rhombus 15 in this way define.

As shown in FIG. 8, according to the present invention, the image A can be obtained without using Convex Hull and Convexity Defect algorithms as in the hand recognition unit 100 described above. To calculate the end point of the finger.

The line connecting unit 310 connects the palm center point 40 in the image B defined by the second center point defining unit 240 and the vertex 13 of the hand contour line of the image A, 38).

The contact point extraction unit 320 extracts a contact point 31 between the second connection line 38 and the hand contour line 30 of the image B. [

50pixel)로서, 이는 사용자 설정에 따라 용이하게 변경 가능한 것을 특징으로 한다.The distance between the contact point 31 and the palm center point 40 in the defined image B while moving the contact point 31 along the hand contour line 30 of the image B in the finger end point calculation unit 330, And the point having the largest value is calculated as the finger end point 33 of the image B. [ At this time, the calculated finger end point 33 is further moved by a predetermined distance to extract a point having a distance value with respect to the palm center point 40 in the defined image B that is larger than the finger end point 33 And changes it to the fingertip point 33 of the image (B). The predetermined distance in Fig. 8 is 50 pixels or less

50 pixels), which is easily changeable according to user setting.

 The central point 40 and the fingertip point 33 of the palm 40 in the image B are connected to each other at the center point and end point connection unit 400 and are referred to as a third connection line 39.

The comparison and output unit 500 compares the length value of the third connection line 39 with the length value of the first connection line 19 and outputs a hand tracking result. If the length of the third connecting line 39 is smaller than the length of the first connecting line 19, it is determined that the finger of the image B is folded. On the other hand, if the length value of the third connecting line 39 is greater than or equal to the length value of the first connecting line 19, it is determined that the finger in the image B is unfolded.

9 is a diagram showing a result of applying the present invention to an augmented reality according to an embodiment of the present invention.

As shown in FIG. 9, when the present invention is applied to an augmented reality, special effects can be applied not only to overlapping images but also to real images. As a result, the image (B) is output to the center of the hand of the real image, and the special effect is output to the face area detected by the face detection technology using the robust real-time face detection algorithm.

When the user takes a hand gesture, the special effect according to the gesture is applied to the face every frame of the image. (The user's face is mosaiced in Fig. 9). Hand tracking is not stopped, hand detection is performed in each hand frame, and special effects are performed in the face.

 The special effect is characterized in that it can be easily changed according to user's setting such as image output, upside-down, left-right reversal, monochrome, mosaic, color reversal and the like.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. It is clear that the present invention can be suitably modified and applied in the same manner. Therefore, the above description does not limit the scope of the present invention, which is defined by the limitations of the following claims.

10: Hand contour of input image (A) for hand recognition
13: Corner of the hand contour of the image (A)
14: Corner point between fingers
15: Rhombus
17: area of five fingers
19: First connecting line (a line connecting the corner point 20 of the palm with the corner point 13 of the hand contour)
20: palm center point in image (A)
30: Hand contour of image (B) input for hand tracking
31: Contact point
33: Finger end point
38: second connecting line (a line connecting the vertex 13 of the hand contour and the palm center point 40)
39: third connecting line (line connecting the contact point 31 and the palm center point 40)
40: center of palm in image (B)
100: hand recognition unit 110: first candidate group detection unit
120: first contour extracting unit 130:
140: First center point defining section
150: finger region and reference distance value defining unit
160:
200: center tracking unit 210: second candidate detection unit
220: second contour extracting unit 230: moving and sizing unit
240: second center point defining section 300: end point calculating section
310: line connecting part 320: contact extracting part
330: finger end point calculation unit 400: center point and end point connection unit
500: comparison and output unit

Claims (12)

A hand recognition unit for recognizing a hand using an input image (A) for hand recognition;
A center tracking unit for defining a palm center point in the input image B for hand tracking using the hand recognition information of the hand recognition unit;
An end point calculation unit for calculating a finger end point in the image (B) using the palm center point defined by the center tracking unit and the hand recognition information recognized by the hand recognition unit;
A center point and an end point connection unit connecting the defined palm center point and the calculated finger end point by a line; And
And a comparison and output unit for comparing the length value of the line with the reference distance value of the image A to output a hand tracking result,
Wherein the endpoint calculator comprises:
A line connecting portion connecting lines of the center point of the palm within the defined image (B) and the finger vertex of the image (A);
A contact extraction unit for extracting a contact point between the connected line and a hand contour line in the image (B); And
A point having the largest distance value between the extracted contact point and the palm center point in the defined image B while moving the extracted contact point along the hand contour line in the image B is referred to as a finger And a fingertip point calculation unit for calculating a fingertip end point based on the calculated endpoints. 2. The apparatus of claim 1,
If the length value of the line is smaller than the reference distance value, it is determined that the finger in the image B is folded,
If the length value of the line is greater than or equal to the reference distance value, it is determined that the finger in the image (B) is unfolded. The handwriting recognition device according to claim 1,
A first candidate group detection unit for binarizing and labeling the image A to detect a hand region candidate group;
A first contour extractor for extracting a hand contour from the defined hand region after defining a region having the largest area among the detected hand region candidate regions as a hand region;
A vertex extracting unit for extracting a vertex of a hand contour line from the extracted hand contour line using a Convex Hull;
Defining a largest rhombus within the extracted hand contour, and defining a center point of the defined rhombus as a center point of the palm;
A finger region and a reference distance value defining unit defining an area of the finger from an angle of a line connecting the vertex of the extracted hand contour and the center point of the defined palm and defining a reference distance value from the length of the line; And
And a recognition unit for recognizing the hand using the defined area of the finger and the reference distance value using the Convexity Defect based on the Convex hole. The apparatus according to claim 1,
A second candidate group detection unit for binarizing and labeling the image B to detect a hand region candidate group;
A second contour extractor for extracting a hand contour from the defined hand region after defining a region having the largest area among the detected hand region candidate regions as a hand region;
A moving and scaling unit for scaling the size of the image by moving the rhombus in at least one of left, right, up and down and diagonal directions so that the image A includes rhombus in the hand contour of the image B; And
And a second center-point defining unit that defines a center point of the moved and adjusted rhombus as a palm center point in the image (B). delete 2. The method according to claim 1, further comprising:
And extracts a point having a value larger than the calculated fingertip point by a predetermined distance so that a distance value between the defined center of the image B and the center of the palm is greater than the calculated fingertip point, A Smart Recognition System Using Tracking of Hand Motion. A hand recognizing step of recognizing the hand using the input image (A) for recognizing the hand in the hand recognizing part;
A center tracking step of defining a palm center point in the image (B) input for hand tracking using the hand recognition information of the hand recognition step in the center tracking unit;
An end point calculation step of calculating a fingertip point in the image (B) using the palm center point defined in the center tracking step and the hand recognition information recognized in the hand recognition step in the end point calculation unit;
Connecting a center point and an end point connecting the center of the palm defined above and the calculated finger end point by a line at the center point and end point connection portion; And
And comparing and outputting the hand tracking result by comparing the length value of the line and the reference distance value of the image (A) at the comparing and outputting unit,
The endpoint computing step may include:
Connecting a center point of a palm in the image (B) defined by the line connecting part and a finger vertex of the image (A) with a line;
Extracting a contact point between the connected line and a hand contour line in the image (B) at the contact point extracting unit; And
A point having the largest distance value between the extracted contact point and the central point of the palm of the defined image B while moving the extracted contact point along the hand contour line in the image B, Calculating a finger end point in the fingerprint image (B). 8. The method of claim 7,
If the length value of the line is smaller than the reference distance value in the comparing and outputting unit, it is determined that the finger in the image B is folded,
And if the length value of the line is greater than or equal to the reference distance value, it is determined that the finger in the image (B) is unfolded. 8. The method according to claim 7,
Detecting a hand region candidate group by binarizing and labeling the image (A) in the first candidate group detection unit;
Extracting a hand contour from the defined hand region after defining a region having the largest area among the detected hand region candidate group as a hand region in the first contour extracting unit;
Extracting a vertex of a hand contour from the extracted hand contour using a convex hull in a vertex extracting unit;
Defining a largest rhombus in the extracted hand contour at a first center point defining unit and then defining a center point of the defined rhombus as a center point of the palm;
Defining a region of a finger from a line connecting a vertex of the extracted hand contour and a center point of the defined palm in a finger region and a reference distance value defining unit and defining a reference distance value from the line length; And
Recognizing the hand using the defined area of the finger and the reference distance value using the Convexity Defect based on the Convex hole in the recognition unit, Way. 8. The method according to claim 7,
Detecting a hand region candidate group by binarizing and labeling the image (B) in a second candidate group detection unit;
Extracting a hand contour from the defined hand region after defining a region having the largest area among the detected hand region candidate group as a hand region in the second contour extracting unit;
Adjusting the size of the image by moving the rhombus in at least one of left, right, up, down, and diagonal directions so that the image of the image (A) includes rhombus in the hand contour line in the image (B) And
And defining a center point of the moved and adjusted rhombus as a palm center point in the image (B) in the second center point defining unit. delete The method according to claim 7, further comprising:
And extracts a point having a value larger than the calculated fingertip point by a predetermined distance so that a distance value between the defined center of the image B and the center of the palm is greater than the calculated fingertip point, A smart recognition method using a tracking of a hand gesture.

Images