KR20070103895A - System of hand-gesture recognition and method thereof - Google Patents

Abstract

A system and a method for recognizing hand gesture are provided to extract and recognize a hand-gesture image exactly in real time so that a gesture recognition system is commonly used. An image acquiring module(110) acquires a hand-gesture image. A shape expression module(120) transforms the hand-gesture image into a primitive image obtained by dividing a specific contour area of the hand-gesture image into primitive elements by using morphological operation. A center calculating module(130) classifies the primitive elements according to sizes and finds centers of the primitive elements. A feature vector setting module(140) sets a feature vector of the primitive image by using an angle of a connection line among the center of a main primitive element and the centers of sub-primitive elements with a horizontal line. The main primitive element is a primitive element having the largest size. The sub-primitive elements are primitive elements except for the main primitive element.

Description

System and hand gesture recognition system and method

1 is a block diagram showing a hand gesture recognition system according to an embodiment of the present invention,

2 is a diagram illustrating a process of inputting an image of a hand gesture;

3 is a diagram illustrating an input gesture image;

4 is a diagram illustrating a primitive shape image that is an image obtained by dividing a gesture image into primitive shape elements;

5 illustrates a line connecting center points of primitive elements;

6 is a diagram illustrating a feature vector recognition region;

7 is a flowchart illustrating a hand gesture recognition method according to an exemplary embodiment of the present invention.

The present invention relates to a hand gesture recognition system and method, and more particularly, to a hand gesture recognition system and method capable of recognizing hand gestures quickly and efficiently, enabling fast interface design, and real-time processing.

In general, digital image processing means processing images with a computer in a narrow sense, but digital image processing in a broader sense means all fields related to images that generate, process, interpret, and recognize images using a computer. Means.

Digital image processing by the computer system has been put to practical use in many applications such as photograph analysis, design, and pattern recognition by rapid advances in memory devices.

In the field of image processing, the hand gesture recognition field is focused as a means of natural information transmission between humans and computers, and recently, a study to use hand gestures as a means for intuitionally conveying human will to a machine. Is widely underway.

For the natural means of communicating information between humans and computers, the design of the interface that understands the hand gestures that are used naturally in everyday conversations is widely used as a natural input means that responds to subtle expressions such as communication and emotions and emotions. It is also widely used as a system for mediation and video content retrieval.

The conventional hand gesture recognition method for the interface design uses a device mounting method for analyzing data obtained by attaching various types of sensors such as gloves to a human body.

However, such a device mounting method requires expensive equipment, and it is necessary to attach various types of sensors connected to the control system, and when the new technology is added to the control system including the sensor, the whole system must be reimplemented. There is a problem in generality and extensibility in that it is.

In order to solve the above problems, recently, a visual-based image processing method for processing an image acquired from a video camera has been widely used.

The hand gesture recognition method of the visual-based image processing method has the advantage of not needing to attach a separate sensor, but the resolution of the video data is low and in the Appearance-based approach which is the main research object of the visual-based image processing. Hidden Markov Model (HMM) and Neural Network Model (HMM), which are mainly applied, have a problem that hardware implementation and real-time processing are impossible due to complex recognition algorithm.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and an object thereof is to provide a hand gesture recognition system and method capable of recognizing hand gestures quickly and efficiently and enabling fast interface design to enable real-time processing. .

Another object of the present invention is to provide a computer-readable recording medium having recorded thereon a program for executing the above hand gesture recognition method on a computer.

Other objects and advantages of the invention will be described below and will be appreciated by the embodiments of the invention. Furthermore, the objects and advantages of the present invention can be realized by means and combinations indicated in the claims.

In order to achieve the above object, the present invention provides a hand gesture recognition system for implementing an interface between a user and an electronic terminal, comprising: an image acquisition unit for acquiring a gesture image that is an image of the user's hand gesture; A shape expression unit for converting a characteristic outer region part of the morphological shape of the gesture image into a primitive shape image which is an image obtained by dividing a predetermined number of simplified transform areas into primitive shape elements by using a morphological operation; A center point calculator for classifying the primitive elements by size and calculating a center point of the primitive elements classified by size; Feature vector setting that sets the feature vector of the primitive image using the angle formed by the horizontal line between the center point of the principal primitive element, which is the largest primitive element, and the center point of each of the primitive primitives, the other primitive element part; And a direction determination unit that determines the direction of the gesture image by using the set feature vector.

In this case, the morphological calculation is performed by performing erosion operation on a predetermined component until the gesture image is reduced to a point or a line, and performing extended operation on the eroded gesture image by the number of times the erosion operation is performed. It is desirable to obtain primitive elements.

In addition, the feature vector is preferably set using the following equation.

Where A is the feature vector, N is the number of primitive elements, (x, y) is the center point of the principal primitive element, and (x _k , y _k ) is the center point of the kth subprimary element.

The feature vector setting unit may calculate a distance between the center point of the main primitive element and the center point of the subprimary element, and the remaining center points and the main circle except for a predetermined number of center points corresponding to a predetermined distance or less among the calculated distances. It is preferable to set the feature vector of the primitive image by using an angle formed by a line connecting the center point of the visual element to the horizontal line.

Further, the electronic device may further include a controller configured to control the electronic terminal in response to a criterion for controlling the electronic terminal determined in advance of the determined gesture image.

On the other hand, a hand gesture recognition method for implementing an interface between a user and the electronic terminal, comprising: an image acquisition step of acquiring a gesture image which is an image of the user's hand gesture; A shape expression step of converting a characteristic outer region part of the shape shape of the gesture image into a primitive shape image which is an image obtained by dividing a predetermined number of simplified transform areas into primitive shape elements by using a morphological operation; A center point calculation step of classifying the primitive elements by size and calculating a center point of the primitive elements classified by size; Feature vector setting that sets the feature vector of the primitive image using the angle formed by the horizontal line between the center point of the principal primitive element, which is the largest primitive element, and the center point of each of the primitive primitives, the other primitive element step; And a direction determination step of determining the direction of the gesture image using the set feature vector.

Furthermore, it is desirable to include a computer readable recording medium having recorded thereon a program for executing the hand gesture recognition method on a computer.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, terms or words used in the specification and claims should not be construed as having a conventional or dictionary meaning, and the inventors should properly explain the concept of terms in order to best explain their own invention. Based on the principle that can be defined, it should be interpreted as meaning and concept corresponding to the technical idea of the present invention.

Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not represent all of the technical idea of the present invention, various modifications that can be replaced at the time of the present application It should be understood that there may be equivalents and variations.

1 is a block diagram showing a hand gesture recognition system according to an embodiment of the present invention, Figure 2 is a view showing a process of inputting the image of the hand gesture, Figure 3 illustrates an input gesture image, Figure 4 FIG. 5 is a diagram illustrating a primitive image, which is an image obtained by dividing a gesture image into primitive elements, FIG. 5 is a diagram illustrating a line connecting center points of primitive elements, and FIG. 6 is a diagram illustrating a feature vector recognition region.

First, as shown in FIG. 1, the hand gesture recognition system 100 according to an embodiment of the present invention includes an image acquisition unit 110, a shape expression unit 120, a center point calculation unit 130, and a feature vector. The setting unit 140, the directional determination unit 150 and the control unit 160 are included.

The image acquisition unit 110 acquires a gesture image which is an image of a user's hand gesture through an electronic terminal.

Preferably, the electronic terminal is a digital camera, a camcorder, a CCD camera, a video camera, or the like. An image of a user is input through a lens unit of the electronic terminal, and only a hand region is extracted from the input image to obtain a gesture image.

As shown in FIG. 2, when the user makes a hand gesture in front of the electronic terminal in various directions, an image of the hand gesture is input to the lens unit of the electronic terminal.

In most cases, the input image of the hand gesture is input as an RGB image. In this case, the input RGB image is converted into a YC _b C _r image, and the C _b image. Component or determine the position of the color using the C _r component to extract the hand area, and obtains a gray image composed of Y component with respect to the skin location.

That is, the input RGB image is converted into a YC _b C _r image which is easily converted to each other according to a conversion formula. YC _b C _r The image is another color space that separates the luminance from the color information, where the luminance is symbolized by Y and the color information is each C _b. And C _r .

At this time, the C _b Ingredients for skin color information 77 <C _b C _r having a value in the range of <127 Ingredients for skin color information 133 <C _r C _b component and C _r having a value in the range of <173; Find the position of the skin color in a given image through the horizontal projection (number of pixels of the binary image in the horizontal direction) and the vertical projection (number of pixels of the binary image in the vertical direction) for a range of components, and the Y component for the given skin color position. Acquire a gray image.

Next, the gray image is represented as a binary image, which is an image of two colors, black and white, using a threshold. The binary image is an image widely used in a simple and practical system for data processing. The image corresponding to the actual position of the hand is expressed in white, and all other values are represented by replacing black with the background. . When using the binary image, compared with the 24-bit image of the RGB color image, the main variable value of recognition, that is, the feature is faster and simpler to calculate the feature, so the monochrome image is used.

In addition, if there is noise in the binary image through a preprocessing process of removing the noise of the image by smoothing before processing the binary image, it is preferable to remove the irregular noise using a low pass filter.

FIG. 3 is a screen illustrating a gesture image of various hand gestures in which gray image processing is performed on the image data acquired through various image input means, a gray image is processed as a binary image, and irregular noise is removed. have.

In FIG. 3, (A) shows when the user pushes the thumb upward, (B) shows when the user pushes the thumb to the left, and (C) shows when the user pushes the thumb to the right. D) is a binary image of the gesture image when the user clenched the fist without lifting the thumb.

Of course, in addition to this, it is obvious that there can be a gesture image in various directions, and for the detailed description of the present invention to be realized in the present invention, it will be described below based on the four hand gesture shapes shown in FIG. This will be explained in detail.

After acquiring a gesture image as a binary image as described above, the shape expression unit 120 simplifies a predetermined number of portions of the outline region of the shape of the gesture image by using a morphological operation. The image is converted into a primitive image, which is an image divided into primitive shape elements, which are transformation areas.

Here, the morphological operation is to extract only a desired part in the image by reflecting information of a target object that knows a geometric shape in advance, and is a calculation that is frequently used to recognize particles or determine the shape of an object. .

The primitive shape element refers to a simplified transformation region of an external shape that appears as a result of performing the morphological operation that can only represent the external shape of an object among the acquired images, and views an image composed of such primitive shape elements. In the present invention, it is defined as a raw image.

In the present invention, the morphological operation is adopted to divide the characteristic outer region part of the gesture image into a primitive shape element.

For example, the decomposition into primitive elements is intended to decompose shape X into several primitive elements X = {X ₁ , X ₂ , X ₃ ... X _n }. Equation 1 is as follows.

Where G (Z ² ) is the 2-D Open Set defined in 2-D Euclidean space Z. When the primitive image for generating X _i is referred to as Y _i , and a component corresponding to Y _i is represented by B, the morphological operation is expressed as Equation 2 below.

Where B is a structuring element, n _i is the size of the constituent, and ○, BIGOPLUS and BIGOMINUS are the hot, dilation and erosion operators, respectively. And B ^s is the reflection of B relative to the origin.

The erosion operation expands the background of the object and reduces the size of the object. The erosion operation removes about one pixel of the outermost line of the object of interest and replaces the background color with the background color. Because it serves to expand the outermost pixels of the object, the size of the object is expanded and the background is reduced. In addition, the thermal operation has the property of a filter that softens the sharp edges existing on the outside of the object and deletes objects smaller than the component.

That is, using the above equation, the gesture image processed by the binary image is eroded by a predetermined component until it is reduced to a point or line, and the eroded gesture image is expanded by the number of times the erosion operation is performed. By acquiring a primitive shape element, the gesture image is expressed as a primitive shape image obtained by extracting a characteristic outer region part of a shape.

When morphological operations are applied to the gesture image shown in FIG. 3 and divided into primitive shape elements, the image shown in FIG. 4 appears as a primitive shape image.

That is, Fig. 3A is Fig. 4A, Fig. 3B is Fig. 4B, Fig. 3C is Fig. 4C, and Fig. 3D. Are respectively converted to (D) of FIG.

The predetermined number refers to a number arbitrarily determined according to memory and processing capacity of a central processing unit, a RAM, a ROM, a data storage device, and the like of the system for processing the raw image.

Referring to (A) to (D) shown in FIG. 4, the primitive shape image is decomposed into five primitive shape elements, but is not limited thereto, and the central processing apparatus of the system for processing the primitive shape image The memory may be divided into various numbers of primitive elements according to memory and processing power of a RAM, a ROM, a data storage device, and the like.

When divided into a small number of primitive elements, the operation processing speed may be increased in the process of recognizing the direction of the hand gesture, but on the other hand, the accuracy of the recognition of the hand gesture may be lowered.

It is preferable to determine the number of primitive elements by the median of such internal relations.

The center point calculator 130 classifies the primitive elements by size and calculates a center point of the primitive elements classified by size.

First, classify primitive elements by size, determine the largest primitive element as the main primitive element, and classify other primitive elements as the sub primitive element, and then classify the primitive elements classified by the size. Calculate the center point.

That is, if the total number of decomposed primitive elements is n, the primitive element having the first size becomes the main primitive element, and X _i (i = 2,3 ..., n extracted from the following decomposition) -1) are classified into first-order shape elements in the order of first, second, third, ..., n-1, and the like.

The center points of the primitive elements classified as described above are determined by performing erosion operations. First, the erosion operation using the Rohmbus component is repeated to the 3 * 3 square component until the shape of the primitive element is smaller than the component, and the result is again 90 °, An erosion operation is performed with the Streak components in 180 °, 45 °, and 135 ° directions to calculate the center points of the primary primitive elements and the non-removable subprimary elements.

In addition, since the center point of the primitive shape element corresponds to the distance on the x, y axis on the two-dimensional coordinate plane relative to the main primitive shape element, the relative coordinates can be obtained by analyzing the shape of the image data. It will be apparent that can be determined in various ways by those skilled in the art.

4A shown in FIG. 4A shows the center point of the main primitive elements, and 4b, 4c, 4d and 4e show the center points of the respective primitive elements.

The feature vector setting unit 140 uses the angle between the center point of the main primitive element that is the largest primitive element and the center point of each of the primitive primitive elements that are the remaining primitive elements with the horizontal line to form a primitive image. Sets the feature vector of.

Typically, the primary primitive element is an image of the palm of a person's hand, and the subprimary primitive element is a segment of the thumb and the finger where the outer features of the hand appear.

The present invention is processed by recognizing the orientation of the thumb around the palm, the direction of the thumb using the angle between the palm and thumb to focus on the primitive shape element of the thumb will usually be the largest negative primitive element Adopting a way to recognize.

In order to implement this recognition systemically, first, the center point calculation unit 130 calculates the center point, and then connects the center points of the sub-element shape elements with the center points of the main primitive elements in a straight line.

As described above, a straight line connecting the center points is shown in FIG. 5, and when the center points of the primitive elements of (A) shown in FIG. 4 are connected, they appear as shown in FIG. 5A and FIG. 4B. When connecting the center points of the primitive elements of (C) and (D), they appear as (B), (C) and (D) of FIG. 5, respectively.

The feature vector is set using an angle formed by a line connecting the center point with a horizontal line.

For example, the _primary primitive elements are referred to as X, the subprimary primitive elements are referred to as X ₁ , X ₂ ... X _n-1 , and the center point of each primitive primitive element and each sub primitive element If the angle formed by the line connecting the center point of the horizontal line with X = {θ ₁ , θ ₂ , θ ₃ , ..., θ _n-1 }, the feature vector is set using Equation 3 below. It is desirable to be.

Where A is the feature vector, N is the number of primitive elements, (x, y) is the center point of the principal primitive element, and (x _k , y _k ) is the center point of the kth subprimary element.

Referring to FIG. 5A, an angle between a horizontal line and a line connecting 5a, which is the center point of one main primitive element, and 5b, 5c, 5d, and 5e, which is the center point of the remaining primitive elements, is θ ₁ , θ ₂ , θ _3, and θ ₄ . By adding all the angles of θ ₁ , θ ₂ , θ _3, and θ ₄ , and dividing by 4, a feature vector can be obtained.

In other words, the feature vector calculated by this process becomes an index in which the sub-element shape element forms an angle with respect to the main primitive element, and this feature vector eventually represents the direction of the thumb of a person.

[Table 1], using the above Equation (3), the center point and the sub-element of each primary primitive element shown in (A), (B), (C) and (D) of FIG. The angle formed by the straight line connecting the center points of the horizontal line is calculated, and the average is calculated and summarized in a table.

 Gesture  Mean Value  (A)   115.72 °  (B)    177.54 °  (C)   15.24 °  (D)  Origin

As shown in Table 1, the feature vector of (A) shown in Fig. 5 is set to 115.72 °, the feature vector of (B) is set to 177.54 ° and so on.

In addition to the method of setting the feature vector of the primitive image using the average value of each angle as described above, the feature vector may be set by various other methods.

For example, since the sub-prime shape element that is far from the center point of the primitive shape element among the primitive shape elements is the primitive shape element that plays the most important role of recognizing the direction, it is located farthest from the main primitive element. Directional recognition can be more accurately recognized by setting the feature vector by calculating the average value with the largest weight using the formulas such as * 2, * 3 and so on.

In addition, the feature vector setting unit 140 calculates distances between the center points of the main primitive elements and the center points of the sub primitive elements, and excludes a predetermined number of center points corresponding to a predetermined distance or less from the calculated distances. Preferably, the feature vector of the primitive image is set using an angle formed by a line connecting a center point and a center point of the main primitive element to a horizontal line.

Since the sub-elements that fall below the predetermined distance among the calculated distances can be assumed to have no significant influence on the directional recognition, the calculation speed may be calculated after excluding the center point of the sub-elements that fall below the predetermined distance. Can be further improved.

The predetermined distance means an arbitrary threshold set to exclude sub-elements that can be assumed to have a low influence on the direction recognition when the sub-elements are located near the main element. do.

In addition, since the predetermined number is to exclude the sub-elements that are determined to be unnecessary for the hand gesture among the sub-elements, only one sub-element may be excluded or two sub-elements may be excluded. May be excluded. Since this occurs irregularly according to the shape to be recognized, it is preferable that a predetermined number of secondary shape elements are removed according to the algorithm.

 The direction determining unit 150 determines the direction of the gesture image by using the set feature vector.

The set feature vector recognizes the direction of a gesture image in the feature vector recognition region illustrated in FIG. 6.

As shown in FIG. 6, when the average value of the feature vector is in the range of 45 ° to 135 °, Resion 2 is used, in the case of 135 ° to 225 ° area, it is Resion 4, and in the case of the 325 ° to 45 ° area. Resion 3 is set to recognize feature vectors.

In general, when the shape of the hand is a fist, the secondary primitive element is generally not far from the main primitive element. Therefore, the center point and the primitive element of the secondary primitive element are reflected in consideration of these morphological features. If the center point is less than or equal to the predetermined distance, it can be recognized as a shape holding a fist.

Through this algorithm, we can recognize the morphological characteristics of another hand gesture.

That is, the region of Resion 1 shown in FIG. 6 is the region of the visual feature, and may be configured to recognize additional hand gestures.

Therefore, in the case of (A) shown in [Table 1], since the value of the feature vector is 115.72 °, the average value of Resion 2 is included in the range of 45 ° to 134 ° and the directionality of (A) can be recognized in the direction of Resion 2. It becomes possible.

Since the ideal value of the hand gesture's directionality is 90 ° for Resion 2, 180 ° for Resion 4, and Resion 3 is 0 °, the feature vectors calculated from the respective gesture images in Table 1 above. Are 115.72 °, 177.54 °, and 15.24 °, respectively, and even if the direction of the gesture image is slightly different for each user, it is determined to be sufficiently included in the setting range, and it can be seen that accurate recognition is possible.

The controller 160 controls the electronic terminal in response to a criterion for the control of the electronic terminal, which is determined in advance in the direction of the determined gesture image.

The process for determining the directionality and controlling the electronic terminal from the calculation of the feature vector will be described below.

 Gesture  Vector illustration  Region  Recognition  (A)  115.72  2  play  (B)  177.54  4  left  (C)  15.24  3  right  (D)  ORIGIN  One  stop

For example, if the control unit 160 controls Play to Resion 2, Right for Resion 3, Left for Resion 4, Stop for Resion 1, and if the direction of the gesture image is recognized as Resion 2, The electronic terminal drives Play.

Of course, in the case of Resion 3, it is also possible to play a different control role in the recognized direction, such as fast forward, Rewind in the case of Resion 4, it is preferable to be controlled by different driving according to the various settings of the electronic terminal applied. Do.

An interface between the user and the electronic terminal is implemented by the above-described components, and the interface drives the electronic terminal by recognizing the direction of the user's hand gesture.

The hand gesture recognition system configured as described above may be implemented in the following manner. Hereinafter, the implementation of the hand gesture recognition method according to the present invention will be described in detail with reference to FIG. 7.

7 is a flowchart illustrating a hand gesture recognition method according to an embodiment of the present invention.

First, as shown, the image acquisition unit 110 obtains a gesture image that is an image of a user's hand gesture (S710).

 At this time, a gesture image, which is an image of a user's hand gesture, is acquired through an electronic terminal.

When the user makes a hand gesture in various directions in front of an electronic terminal such as a digital camera, a camcorder, a CCD camera, a video camera, or the like, an image of the hand gesture is input to the lens unit of the electronic terminal, and the hand region is included in the input image. Only the image is extracted to obtain a gesture image.

The description of the gray image processing step and the binary image processing step of the image of the color image has been described above in detail and thus will be omitted.

Next, the surgical region portion characteristic of the morphological shape of the gesture image is converted into a primitive image, which is an image obtained by dividing a predetermined number of simplified transformation regions into primitive shape elements using a morphological operation ( S720).

Using the morphological operation, the gesture image is eroded until it is reduced to a point or a line, and the eroded gesture image is extended to the number of times the erosion operation is performed to obtain a primitive shape element. The gesture image is expressed as a raw shape image obtained by extracting a characteristic outer region part of a shape.

Next, the primitive elements are classified by size and a center point of the primitive elements classified by size is calculated (S730).

The primitive elements are classified by size, the largest primitive element is designated as the primary primitive element, and the other primitive elements are designated as the subprimary primitive element, and then classified. Calculate the center point.

Next, the feature vector of the primitive image is set using an angle formed by a horizontal line between the center point of the main primitive element, which is the largest primitive element, and the center point of each subprimary element, the other primitive element. (S740).

The feature vector is preferably set using Equation 3 described above.

In addition, the feature vector setting step (S740) calculates the distance between the center point of the main primitive shape element and the center point of the sub primitive shape element, and excludes a predetermined number of center points corresponding to a predetermined distance or less from the calculated distances. Preferably, the feature vector of the primitive image is set using an angle formed by a line connecting a center point and a center point of the main primitive element to a horizontal line.

Since the sub-elements corresponding to the predetermined distance or less of the calculated distances are not necessary for the directional recognition, the computational speed is further improved by calculating after subtracting the center point of the sub-elements corresponding to the predetermined distance or less. You can.

Next, the direction of the gesture image is determined using the set feature vector (S750).

The set feature vector recognizes the direction of a gesture image in the feature vector recognition region illustrated in FIG. 6.

Finally, the electronic terminal is controlled in response to a criterion for the control of the electronic terminal, in which the direction of the determined gesture image is predetermined (S760).

For example, in step S760, if the direction of Resion 2 is Play, the direction of Resion 3 is Right, the direction of Resion 4 is Left, and the direction of Resion 1 is controlled to Stop, the direction of the gesture image is Resion 2. When recognized, the electronic terminal can be controlled to recognize the Play command.

Of course, in the case of Resion 3, it is also possible to play a different control role in the recognized direction, such as fast forward, Rewind in the case of Resion 4, it is preferable to be controlled by different driving depending on the setting of the electronic terminal to be applied. .

The present invention can be embodied as computer readable codes on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data is stored as a medium that can be read by a computer system.

Examples of computer-readable recording media include ROM, RAM, CO-ROM, magnetic tape, floppy disks, optical data storage devices, and the like, which are also implemented in the form of carrier waves (for example, transmission over the Internet). Include. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

As mentioned above, although this invention was demonstrated by the limited embodiment and drawing, this invention is not limited by this, The person of ordinary skill in the art to which this invention belongs, Of course, various modifications and variations are possible within the scope of equivalents of the claims to be described.

According to the present invention, since the direction of the hand gesture is recognized using the feature vector, the gesture can be analyzed more easily and the information can be quickly provided to an electronic device requiring a real-time processing speed.

In addition, by applying morphological operations that decompose complex images into simple primitive elements, hand gestures can be recognized quickly and efficiently, enabling fast interface design and saving resources of the entire system.

In addition, it is possible to commercialize the gesture recognition system by enabling accurate extraction and recognition of hand gestures in real time, and has the advantage that it can be widely applied to the interface design related to the retrieval of video data and other electronic systems.

Claims (11)

In a hand gesture recognition system for implementing an interface between a user and an electronic terminal, An image acquisition unit which acquires a gesture image which is an image of the user's hand gesture; A shape expression unit for converting a characteristic outer region part of the morphological shape of the gesture image into a primitive shape image which is an image obtained by dividing a predetermined number of simplified transform areas into primitive shape elements by using a morphological operation; A center point calculator for classifying the primitive elements by size and calculating a center point of the primitive elements classified by size; Feature vector setting that sets the feature vector of the primitive image using the angle formed by the horizontal line between the center point of the principal primitive element, which is the largest primitive element, and the center point of each of the primitive primitives, the other primitive element part; And And a direction determination unit for determining the direction of the gesture image by using the set feature vector. The method of claim 1, wherein the morphological operation is Performing an erosion operation to a predetermined component until the gesture image is reduced to a point or a line, and performing the extended operation to the number of times the erosion operation is performed to perform the erosion operation to obtain the primitive shape element. Hand gesture recognition system characterized by. The method of claim 2, wherein the feature vector, Hand gesture recognition system, characterized in that set using the following equation.

Where A is the feature vector, N is the number of primitive elements, (x, y) is the center point of the principal primitive element, and (x _k , y _k ) is the center point of the kth subprimary element. The method of claim 3, wherein the feature vector setting unit, Computing the distance between the center point of the main primitive element and the center point of the sub-primary element, and connecting the remaining center point except the predetermined number of center points corresponding to a predetermined distance or less of the calculated distance and the center point of the main primitive element Hand gesture recognition system, characterized in that for setting the feature vector of the raw image using the angle formed by the line to the horizontal line. The method of claim 1, And a controller configured to control the electronic terminal in response to a criterion for the control of the electronic terminal determined in advance of the determined direction of the gesture image. In a hand gesture recognition method for implementing an interface between a user and an electronic terminal, An image acquisition step of acquiring a gesture image, which is an image of the user's hand gesture; A shape expression step of converting a characteristic outer region part of the shape shape of the gesture image into a primitive shape image which is an image obtained by dividing a predetermined number of simplified transform areas into primitive shape elements by using a morphological operation; A center point calculation step of classifying the primitive elements by size and calculating a center point of the primitive elements classified by size; Feature vector setting that sets the feature vector of the primitive image using the angle formed by the horizontal line between the center point of the principal primitive element, which is the largest primitive element, and the center point of each of the primitive primitives, the other primitive element step; And And a direction determination step of determining the direction of the gesture image by using the set feature vector. The method of claim 6, wherein the morphological operation is Performing an erosion operation to a predetermined component until the gesture image is reduced to a point or a line, and performing the extended operation to the number of times the erosion operation is performed to perform the erosion operation to obtain the primitive shape element. Hand gesture recognition method characterized by. The method of claim 7, wherein the feature vector, Hand gesture recognition method characterized in that set using the following equation.

Where A is the feature vector, N is the number of primitive elements, (x, y) is the center point of the principal primitive element, and (x _k , y _k ) is the center point of the kth subprimary element. The method of claim 8, wherein the setting of the feature vector comprises: Computing the distance between the center point of the main primitive element and the center point of the sub-primary element, and connecting the remaining center point except the predetermined number of center points corresponding to a predetermined distance or less of the calculated distance and the center point of the main primitive element And a feature vector of the primitive image is set using an angle formed by a line to a horizontal line.  The method of claim 6, And a control step of controlling the electronic terminal in response to a criterion for the control of the electronic terminal determined in advance of the determined directionality of the gesture image. The method according to any one of claims 6 to 10, A computer-readable recording medium having recorded thereon a program for executing the hand gesture recognition method on a computer.

Images