KR100826878B1 - Hand shafe recognition method and apparatus for thereof - Google Patents

Abstract

The present invention relates to a method for recognizing a hand shape and a device therefor, and there is no particular lighting or background limitation, and the hand shape is real-time regardless of the user's age, hairstyle, makeup, or various poses of the user. In order to be recognized, the user detects a face region from the captured image, collects skin color information of the face from the detected face region, detects a hand region based on the skin color information, and then detects the hand. Analyze hand shape based on information about the area.

Hand, Recognition, Awareness, Color, Rock Paper Scissors

Description

HAND SHAFE RECOGNITION METHOD AND APPARATUS FOR THEREOF}

1 is a flowchart showing a schematic flow of a hand recognition method according to a first embodiment of the present invention.

FIG. 2 is a diagram schematically illustrating a relationship between a method for recognizing a hand shape and a hand of a user who needs to read a hand shape according to a second preferred embodiment of the present invention.

3 is a flowchart illustrating a rock-paper-scissors game using a hand recognition method according to a third embodiment of the present invention.

Figure 4 is a flow chart for explaining the game Mukkippa using the hand recognition method according to a fourth embodiment of the present invention.

5 is a structural diagram showing a schematic structure for explaining a hand recognition apparatus according to a sixth preferred embodiment of the present invention.

6 is a structural diagram illustrating a schematic structure of a game device using a hand including a hand recognition apparatus according to a seventh exemplary embodiment of the present invention.

The present invention relates to a method for recognizing a hand shape and an apparatus therefor.

BACKGROUND ART An application field of an image processing system that receives image data photographed using various image input devices such as an optical camera, a CMOS or a CCD optical sensor, and recognizes a movement of an object has been expanded to a TV, a home robot, a game machine, and the like. Accordingly, there is an increasing need for a method for effectively recognizing a person's motion or gesture, particularly a hand shape according to a human's hand gesture.

In order to meet such demands, a method of recognizing gestures by continuously obtaining differential images of successive images, a method of detecting and analyzing edges of major parts from the movement of a target person, and an optical flow A method using flow and a method of extracting and analyzing human contours have been proposed.

However, in order for these technologies to be applied in a real environment, some conditions must be met. First, there should be no special lighting restrictions. Second, there should be no specific background restrictions. Third, regardless of the user's age, hairstyle, makeup, etc. should be. Fourth, it should be independent of the user's various poses. Fifth, real time processing should be possible.

However, the recognition methods proposed above only satisfy some of the above mentioned conditions, but do not satisfy the entire condition. For example, the method of recognizing a gesture using a differential image has a limitation in that it is difficult to apply when the background becomes complicated or when there is a moving person, and the edge of the main part is detected from the movement of the target person. Next, in the case of analyzing this, there is a problem that it is difficult to apply when the color of the user's clothes and the background match. In addition, the method using the light flow and the method of extracting the human body outline require a lot of computation time, and thus there is a problem in that it is not suitable for real time processing.

The present invention is not limited to a special lighting or background, in order to solve the above problems, regardless of the user's age, hairstyle, makeup, various poses of the user, and the like and a method for recognizing the hand shape in real time It is an object to provide a device for.

Hand recognition method according to an embodiment of the present invention for realizing the above object,

Receiving an image including a face and a hand of a target user;

Detecting a face region from the image;

Collecting skin color information of a face from the detected face area;

Classifying areas determined as skin color based on the collected skin color information of the face as an isolated area, and detecting a hand area by determining the largest area within the classified area as a palm area; And,

Analyzing the shape of the hand based on the detected hand region information.

In addition, the scissors bar weaving game method using a hand recognition method according to another embodiment of the present invention,

Receiving an image including a face and a hand of a game participant;

The computer selecting one of the scissors, the rock and the beam;

Analyzing the input image to determine whether a user's hand is a pair of scissors, a rock, or a beam; And,

A rock-paper-scissors game method comprising the step of determining whether or not the user's hand determined and the computer-selected hand shape are determined.

The determining of the shape of the user's hand may include detecting a face region from an image including a hand shape of a game participant, collecting skin color information of the face from the detected face region, Classifying areas determined as skin color based on skin color information as isolated areas, detecting the largest area within the classified area as a palm area, and detecting a hand area based on the detected hand area Rock-paper-scissors game method comprising analyzing the shape of the hand.

In addition, mujippa game method using a hand recognition method according to another embodiment of the present invention,

Determining a predecessor announcing the start of the game;

Outputting a game start signal in video or audio if the predecessor is a computer, and if the predecessor is a user who is a game participant, requesting a game start signal from the user in video or audio and then receiving a game start signal;

Receiving an image including a user's face and hands;

The computer selecting one of the scissors, the rock, and the beam, and outputting it as an image or an audio;

Analyzing the received image to determine which of the user's hands is a mutt chipper;

If the user's hand shape and the computer's hand shape are determined as described above, the result of the game is outputted. If the hand shape does not match, the predecessor is determined according to the game rules. As a game of how to play mutton,

The determining of the shape of the user's hand may include detecting a face region from an image including a hand shape of a game participant, collecting skin color information of the face from the detected face region, Classifying areas determined as skin color based on skin color information as isolated areas, detecting the largest area within the classified area as a palm area, and detecting a hand area based on the detected hand area Analyzing the shape of the hand.

In addition, a computer-readable storage medium having a computer program implemented to recognize a hand shape according to another embodiment of the present invention,

Receiving an image including a face and a hand of a target user;

Detecting a face region from the image;

Collecting skin color information of a face from the detected face area;

Classifying areas determined as skin color based on the collected skin color information of the face as an isolated area, and detecting a hand area by determining the largest area within the classified area as a palm area; And,

A computer readable storage medium having a computer program implemented to recognize a position of a mobile robot by performing a step of analyzing a hand shape based on the information about the detected hand area.

In addition, the hand recognition device according to another embodiment of the present invention,

An image input module configured to receive an image including a face and a hand of a target user;

A face region extraction module for extracting a face region from the input image;

A face region color information extraction module for collecting color information of a face from the extracted face region;

A hand region extraction module for extracting a portion corresponding to a hand from the captured image of the user based on the color information of the collected face region; And,

And a hand recognition module that determines what hand shape the user has made based on the extracted hand region.

In addition, a game device using a hand including a hand recognition device according to another embodiment of the present invention,

An image input apparatus for photographing a game participant and generating an image including a face and a hand of the game participant;

An image input module for receiving the image, a face region extraction module for extracting a face region from the input image, a face region color information extraction module for collecting color information about a face from the face region, and a color of the collected face region A hand shape recognition device including a hand region extraction module for extracting a portion corresponding to a hand based on the information, and a hand recognition module for determining which hand shape the user has made based on the hand region;

A storage device for storing information about the user and the user's hand;

A controller for determining a corresponding operation of the game device using the hand based on the hand shape determined by the hand recognition device, and controlling the game device using the hand; And,

It includes a display unit and a voice output unit for delivering the information determined by the control unit to the user.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In addition, the same reference numerals are used for parts having similar functions and functions throughout the drawings.

A hand recognition method according to a first preferred embodiment of the present invention will be described in detail with reference to FIG. 1. 1 is a flowchart showing a schematic flow of a hand recognition method according to a first embodiment of the present invention.

Referring to FIG. 1, first, a face region is detected from an image including a face and a hand of a user to be captured using an image input apparatus (S101). The method for detecting a face region may be performed by using ada boosting algorithms known in the art, linearly combining features that are the criteria for face determination, and then hierarchically constructing or detecting the features, or by using the AdaBoosting algorithm. Detection can be made using a face detector using the axis of symmetry described in Publication 2005-60799.

Next, skin color information of the face is collected from the face area detected in step S101 (S105). That is, once a face is detected, the skin color of the face is modeled and used to find an area corresponding to the hand. Preferred methods that can be applied to the present invention to model skin color are as follows. First, the minimum and maximum values r _max , r _min , g _max , g _min , b _max , for each of the red, green, and blue colors determined as skin color while scanning the detected face area b _min ). At this time, when each pixel satisfies the condition of r> g and r> b, it is determined as skin color.

When modeling information about the skin color is obtained from the skin color of the face by the above method, the skin color of the hand is finally determined through the following equation (1) or (2). Equation 1 is a result image obtained by primarily determining the skin color in a predetermined region [i = 0 ~ T (width), j = 0 ~ T (height)],

Equation 2 below is a result of applying a rule for removing a region determined as skin color that is too weak red or too dark red in the T (i, j).

Next, the hand region is detected based on the collected skin color information of the face (S110). For this purpose, it is preferable to use a labeling algorithm. The areas binarized by the skin color are classified as isolated areas, and the largest area within the classified areas is determined to determine the palm area. The task is to find the actual palm area, not the noise, on the assumption that there is no area having the same color as the skin color in the image captured by the target user, or even smaller than the actual palm area.

In general, it is easy to find the hand region by using the isolation algorithm in the binary image. However, if a person is wearing short-sleeved clothes or the color of the clothes is similar to the skin color, the palm area cannot be found only by the isolation algorithm. Therefore, it is preferable to perform an algorithm that cuts out the palm area to compensate for this.

Preferred examples of the algorithm for cutting out the preferred palm area applicable to the embodiment of the present invention include, firstly, the target user collecting the captured image in step S105 while scanning the image from the lower right and the far right and up and down of the captured image. Check whether there is a pixel corresponding to the skin color. However, this is based on the assumption that the palm exists in the right area, and if the palm exists on the left side, it is reasonable that the left and right starting directions are reversed.

Next, it is examined whether one of the image lines meets the skin color pixels while scanning from right to left. At this time, when the skin color pixel is met, the size of the skin color pixel is checked from right to left. If the length of the skin color pixel is greater than or equal to a predetermined threshold, the color is set to the hand and then the height of the hand is increased. If it is below, it is set as noise. If the height of the hand portion increases by a certain value or more, it is determined that the current hand portion is tracked. If, during the inspection, the skin color pixels cannot be met and the height of the previously calculated hand part is equal to or less than a predetermined value (a), it is determined as a noise, and the height of the calculated hand part that does not meet the skin color pixels is more than a predetermined value (a). If it is, it is determined by fingertips. At this time, when judging by the fingertips, all parts below the fingertip threshold threshold [Th = (length of the found face size * b)] are allocated to the non-skin color region. In this case, the predetermined value (a) is preferably 1/3 of the height of the entire input image, and b is preferably 1.5, but the present invention is not limited thereto, and may be variable according to a situation selected by those skilled in the art. Can be.

When the above process is completed, it moves back to the above image line, and then proceeds by repeating the scanning from the right to the left to check whether the skin color pixel is met.

If the hand region is detected because all of the lower regions except the palm region having a specific size are not skin color by the algorithm of cutting the palm region of step S110, the hand shape is analyzed based on the information on the detected hand region. (S115).

A preferred method that can be applied to the present invention for analyzing a hand shape is to comprehensively use a run length algorithm and a recognition algorithm using a neural network based on the information on the hand region detected in step S110.

At this time, the run length algorithm is an algorithm that checks how many objects exist among the same background and objects while scanning the recognition target area from left to right. For example, in the following configuration, if the run length algorithm is performed by assuming 0 is a background and 1 is an object, three backgrounds of eight lengths, one background of ten lengths, one object of five lengths, and six lengths You get the result of two objects.

According to this result, the number of fingers based on the height of the entire palm, that is, the number of rows with two or more objects having a length of three or more, the number of rows with two, and the number of rows with one, are calculated according to the number of fingers. You can distinguish the shape of the hand.

Next, with respect to the hand region detected in step S110, a mesh feature of the 5x5 region is extracted using a recognition algorithm using a neural network, and then the hand shape is recognized.

Next, the final recognition result is obtained by integrating the result recognized through the run length algorithm and the result recognized using the neural network into a support vector machine (SVM) technique.

According to a second preferred embodiment of the present invention, a hand shape recognition method includes detecting a face region, collecting skin color information of a face from the detected face region, determining a hand to be used, and collecting the face. Detecting a hand region based on skin color information of the detected face, and analyzing a hand shape based on the information on the detected hand region. That is, the method may further include determining a hand to be used in contrast to the hand recognition method of the first embodiment.

FIG. 2 is a diagram schematically illustrating a relationship between a method for recognizing a hand shape and a hand of a user who needs to read a hand shape according to a second preferred embodiment of the present invention.

In general, a person has two hands, but in many cases, one of two hands is used as an object to be analyzed as necessary. At this time, when the used hand is the right hand, the palm region is located on the right side with respect to the face, and when the used hand is the left hand, the palm region is generated on the left side.

Referring to FIG. 2, in the hand recognition method according to the second embodiment, the first embodiment is performed with respect to the face image once found, instead of finding the entire hand image of the user as the hand region. After performing the face recognition by the method according to the example, it is checked whether the hand used by the user is the right hand or the left hand based on the recognized face information. If the used hand is the right hand, only the right region is assigned to the hand search region, and if the left hand is used, only the left region is assigned to the hand search region to perform binarization.

A rock-paper-scissors game method using a hand recognition method according to a third preferred embodiment of the present invention will be described in detail with reference to FIG. 3. 3 is a flowchart illustrating a rock-paper-scissors game using a hand recognition method according to a third embodiment of the present invention.

Referring to FIG. 3, first, a game start signal is output to notify that a rock-paper-scissors game starts (S300). This can be done with video or audio. In general, a method of outputting the rock-paper-scissors in order may be used, and a method of outputting a number of 1 to 3 in the forward or reverse direction may be used.

Next, an image including a face and a hand of a user who is a game participant is input, and the computer selects one of the scissors, the rock, and the beam (S305). At this time, the step of receiving an image including the user's face and hand means that the user who takes a specific hand shape is photographed as a whole to include the face and the hand, and then the image is input. After receiving the image, it is preferable that the computer selects one of the scissors, the rock, or the beam, but the computer first selects one of the scissors, the rock, or the beam, and then the user's hand may be input.

In this case, the step of selecting one of the scissors, the rock, or the beam by the computer is a step of arbitrarily determining a corresponding hand shape for playing the rock-paper-scissors game with the game participant. .

Next, by analyzing the input image, it is determined whether the user's hand shape is scissors, rocks, beams (S310). In this case, the determining of the hand shape of the user may use the hand recognition method of the first or second embodiment.

Next, when the hand shape is determined, the hand shape determined in step S310 is compared with the hand shape selected by the computer in step S305, and the win / loss is determined based on the rules of the rock-paper-scissors game (S315). A typical rock-paper-scissors game is a draw if the hands are the same, but if the hands are different, the rule is that the scissors win the beam, the beam wins the rock, and the rock wins the scissors. However, the rule may be modified differently before starting the game, and if the rule is modified, the winning or losing is decided according to the modified rule.

If the win or loss is determined as described above, to output the video or audio to inform the user (S320), and waits to receive input from the user whether to resume the game (S325). If an input from the user to resume the rock-paper-scissors game is received, the process starts again from the step S300.

As a preferred embodiment applicable to the present invention with respect to the third embodiment, before the step S300, further comprising the step of receiving and storing the user's name, the rock-paper-scissors game determines the win or lose the S315 After the step may further include the step of calculating the user's record and the output along with the received title.

In addition, as a preferred embodiment applicable to the present invention with respect to the third embodiment, after the step S305 in which the computer selects any of scissors, rocks or beams and before the step S315 in which the win or loss is determined, In operation S305, the method may further include displaying a hand shape selected by the computer.

The mutt papp game method using the hand recognition method according to the fourth embodiment of the present invention will be described in detail with reference to FIG. 4. Figure 4 is a flow chart for explaining the game Mukkippa using the hand recognition method according to a fourth embodiment of the present invention.

Referring to FIG. 4, first, a predecessor for notifying the start of the game is determined (S400). This is in accordance with the characteristics of the game of Mukjipa that win or lose is determined according to the hand shape determined by the opponent in response to the hand shape of the predecessor, it is preferable to set the computer to the initial value.

Next, by determining who the predecessor is (S405), if the predecessor is a computer outputs the game start signal by video or audio (S410), if the predecessor is a user who is a game participant rather than a computer, start the game to the user by video or audio After requesting a signal (S415), the game start signal is received (S420).

In this case, the game start signal applicable to the present invention is preferably displayed through a display device or through a voice output device. In general, a method of displaying or outputting mujipa in order may be used, and a method of outputting numbers 1 to 3 in a forward or reverse direction may be used.

Next, after receiving an image including a face and a hand of a user who is a game participant (S425), the computer selects one of mute, bob, or bar (S430), and then outputs the determined image or audio (S430). S435). At this time, the step S430 is preferably performed after receiving the user's hand, but the step S430 may be performed before the user's hand is input.

In this case, the step S430 in which the computer selects one of the mute, the bob, or the bar, is a step of arbitrarily determining a corresponding hand shape for playing the game with the user, preferably using a random function.

Next, by analyzing the shape of the user's hand received in step S425, it is determined whether the hand shape is scissors, rocks, beams (S440). In this case, it is preferable to use the hand recognition method of the first or second embodiment to determine the input hand shape of the user.

Next, when the hand shape is determined, the hand shape thus determined is compared with the hand shape selected by the computer in step S430, and it is determined whether or not the hand shape matches to determine a win or loss based on the rules of the game. S445). In a typical mute game, the player wins when the opponent presents the matching hand with respect to the hand presented by the predecessor, but the winner is not determined when the hand does not match.

If the outcome is determined through the step S445, the outcome is output by video or audio (S450), and if the outcome is not determined, the predecessor is changed (S465), and the game is resumed. The predicate is the one who presented the bob, if the bob and mu were presented. However, the rule may be modified differently before starting the game, and if the rule is modified, it is desirable to follow the modified rule.

After the win or loss is output through the step S450, the user waits to receive input from the user whether to resume the game. If the user receives an input to resume the game, the predecessor is determined based on the result of the previous game (S465), and then starts again from the above step S405.

As a preferred embodiment applicable to the present invention with respect to the fourth embodiment, before the step S400, further comprising the step of receiving and storing the title of the game participant, in the step S450 to output the game win or lose The method may further include calculating a record of a game participant and outputting the received title along with the received title.

According to a fifth preferred embodiment of the present invention, a computer-readable storage medium having a computer program implemented to recognize a hand shape includes detecting a face region from an image of a target user, and detecting the face. Collecting skin color information of a face from an area, detecting a hand area based on the collected skin color information of the face, and analyzing a hand shape based on the detected hand area information; A suitably programmed system is a computer readable storage medium having a computer program implemented to recognize the position of a mobile robot.

A hand recognition apparatus according to a sixth exemplary embodiment of the present invention will be described in detail with reference to FIG. 5. 5 is a structural diagram showing a schematic structure for explaining a hand recognition apparatus according to a sixth preferred embodiment of the present invention.

5, the hand recognition apparatus according to the preferred embodiment of the present invention,

Image input module 500, face region extraction module 501, face region color information extraction module 502, left / right hand separation module 503, hand region extraction module 504, and hand recognition module 505 It is configured to include.

The image input module 500 receives an image captured to include a face and a hand of a target user, and provides the image to the face region extraction module 501 and the hand region extraction module 504.

The face region extraction module 501 extracts a region corresponding to the face of the user from the image data provided from the image input module 500, and uses a module known as an AdaBoosting algorithm or face determination. After linearly combining the features that are the criteria of the configuration, it is hierarchically configured to detect, or may be a module using the axis of symmetry described in the present publication No. 2005-60799. The data collected by the face region extraction module 501 is provided to the face region color information extraction module 502 and the left / right hand separation module 503.

The face region color information extracting module 502 collects color information used to extract a user's hand region from the face region extracted by the face region extracting module 501, and the left / right hand separating module 503 may include: The user determines whether the user's hand is a right hand or a left hand based on the face information recognized by the face region extraction module 501.

The hand region extraction module 504 collects the color information collected by the face region color information extraction module 504 for the hand identified by the left / right hand classification module 503 in the image data provided from the image input module 500. Based on the extracted parts corresponding to the hand.

The hand recognition module 505 finally determines what hand shape the user has on the basis of the hand region collected by the hand region extraction module 504.

A game device using a hand including a hand recognition apparatus according to a seventh exemplary embodiment of the present invention will be described in detail with reference to FIG. 6. 6 is a structural diagram illustrating a schematic structure of a game device using a hand including a hand recognition apparatus according to a seventh exemplary embodiment of the present invention.

Referring to FIG. 6, a game device using a hand according to an exemplary embodiment of the present invention may include an image input device 600, a hand recognition device 601, a storage device 602, a controller 603, and an input device ( 604, a display unit 605, and an audio output unit 606.

The image input apparatus 600 is a device for generating an image including a face and a hand of a user by photographing a user who needs to read a hand shape, and detects an optical image and processes the image in a computer. It is a device that converts data.

In this case, as the image input apparatus, various image input apparatuses such as an optical camera, a CMOS or a CCD optical sensor, etc. may be used by those skilled in the art. For example, an image input apparatus capable of capturing at 320x240 resolution at a distance of 0.5 to 2 m may be used. At this time, the image input device is preferably positioned so that the user's hand can be picked up from below for the preferred embodiment of the present invention. This is because when the image captured from the front or upper surface of the hand is binarized using the skin color as described below in order to recognize the shape of the hand, it is difficult to distinguish the pattern of the hand, which increases the probability of error.

The image data generated by the image input apparatus 600 is transmitted to the hand recognition apparatus 601 to read the shape of the user's hand. For this purpose, the hand recognition apparatus 601 preferably has the same configuration as in the sixth embodiment. The read result is stored in the storage device 602 and then transferred to the control unit 603.

The storage device 602 may store information on the user and the user's response input through the image input device 600 and the input device 604 as well as the result read through the hand recognition device 601. .

As a result of the reading by the hand recognition device 601, the controller comprehensively considers the user's response input through the input device 604 and the information stored in the storage device 602 to play a game using the user's hand. If necessary, the display unit 605 or the voice output unit 606 notifies the user of the game progress.

The present invention described above is not limited to the above-described embodiments and the accompanying drawings, and it is common in the art that various substitutions, modifications, and changes can be made without departing from the technical spirit of the present invention. It will be apparent to those skilled in the art.

The conventional hand recognition technology extracts a hand region using only the skin information of the hand region regardless of the skin information of the face when extracting a hand shape. As a result, the detection rate of the hand region is remarkably decreased according to the lighting, the change in the external environment of the background, and the change in the user's clothes style. Also, if you have a camera at a lower height when playing rock-paper-scissors games, the angle that the camera sees and the position of the hand will lie in a straight line when a person reaches out for the rock-paper-scissors game, clearly distinguishing the rock and scissors form. It was hard to do. In addition, if the camera is located at a high height, it is only necessary to play the game while looking down at the floor rather than the face of the person.

In the present invention to solve this, when a person plays a game of scissors, rock, beam, raise the hand to the shoulder height to devise a game. In this case, since the face and hands of the person are always at the same height, the shape of the hand is first detected by detecting the face, and the hand is detected based on the skin information and lighting conditions extracted from the detected area. Detection is possible. In this case, when the hand region and the hand region are connected, an algorithm for separating them is applied. Finally, the obtained hand region is recognized using run length algorithm and neural network, and finally, it is determined by scissors, rock, beam shape through SVM which combines two pieces of information.

Claims (29)

Receiving an image including a face and a hand of a target user; Detecting a face region from the image; Collecting skin color information of a face from the detected face area; Classifying areas determined as skin color based on the collected skin color information of the face as an isolated area, and detecting a hand area by determining the largest area within the classified area as a palm area; And, And analyzing a hand shape based on the detected hand region information. The method of claim 1, The detecting of the face region may include a feature that linearly combines a feature as a criterion for face determination using an adamant boosting algorithm, and then hierarchically constructs and detects the face region. The method of claim 1, The detecting of the face region may include detecting a face by calculating a local similarity of left and right regions of the face using a symmetry axis. The method of claim 1, The step of collecting skin color information of the face may include scanning the detected face area while checking the minimum, maximum values (r _max , r _min , g _max , g _min , b _max , b for each of the red, green, and blue colors determined as the skin color. _min ), and if each pixel satisfies the condition of r> g, r> b, it is determined as skin color. The method of claim 1, The step of collecting skin color information of the face may include scanning the detected face area while checking the minimum, maximum values (r _max , r _min , g _max , g _min , b _max , b for each of the red, green, and blue colors determined as the skin color. _min ), and then determine the hand color according to the following equation (1). [Equation 1]

Here, the determination section is within [i = 0 ~ T (width), j = 0 ~ T (height)]. The method of claim 1, The step of collecting skin color information of the face may include scanning the detected face area while checking the minimum, maximum values (r _max , r _min , g _max , g _min , b _max , b for each of the red, green, and blue colors determined as the skin color. _min ), and then determine the hand color according to the following equation (2). [Equation 2]

Here, the determination section is within [i = 0-T (width), j = 0-T (height)]. delete The method of claim 1, And only cutting out an area corresponding to a palm from the palm area determined in the detecting of the hand area. The method of claim 8, The cutting of the palm region may include first scanning the bottom right and the right and left sides of the image, and checking whether there is a pixel corresponding to the skin color, and if the skin color pixel exists, inspecting the size from the right to the left. If the length of the skin color pixel is above a certain threshold, set it to the hand and increase the height of the hand. If the length of the skin color pixel is less than the certain threshold, set it to noise. Hand recognition method characterized by tracking the hand part. The method of claim 9, Scanning from the lower right and the far right to the left and up of the image, if the pixel corresponding to the skin color is not met and the height of the previously calculated hand part is 1/3 or less of the height of the entire input image, it is judged as noise and the skin color pixel The hand shape recognition method, characterized in that it is determined by the fingertips if the height of the previously calculated hand portion is not more than 1/3 of the height of the entire input image. The method of claim 10, And if it is determined as a fingertip, all the portions below the fingertip by the length of the found face size (1.5) are allotted to the non-skin color region. The method of claim 1, The analyzing of the hand shape may include a SVM (SVM) based on the information on the hand region detected in the detecting of the hand region, and a result recognized through a run algorithm and a recognition algorithm using a neural network. A hand recognition method characterized by integrating the recognition into a support vector machine) technique. The method of claim 1, And determining whether the hand used is left or right between collecting the skin color information of the face and detecting the hand region. Receiving an image including a face and a hand of a user who is a game participant; The computer selecting one of the scissors, the rock and the beam; Analyzing the input image to determine whether a user's hand is a pair of scissors, a rock, or a beam; And, A rock-paper-scissors game method comprising the step of determining whether or not the user's hand determined and the computer-selected hand shape are determined. The determining of the shape of the user's hand may include detecting a face region from an image including a hand shape of a game participant, collecting skin color information of the face from the detected face region, Classifying areas determined as skin color based on skin color information as isolated areas, detecting the largest area within the classified area as a palm area, and detecting a hand area based on the detected hand area Rock-paper-scissors game method comprising analyzing the shape of the hand. The method of claim 14, And a step of outputting a game start signal to notify the start of the rock-paper-scissors game before the input of the user's image. The method of claim 14, Before receiving the user's image, the step of receiving and storing the user's title, and after the step of determining whether the rock-paper-scissors game or not, calculating the total number of game participants and outputting along with the received title Rock-paper-scissors game method further comprising. The method of claim 14, And wherein the computer selects one of the scissors, the rock, and the beam, is determined using a random function. The method of claim 14, And after the computer selects one of the scissors, the rock, or the beam, and before determining the win or loss, displaying the hand rock selected by the computer. The method of claim 14, And after the determining of the outcome, outputting a result of informing the user of the outcome, and waiting for input from the user whether or not to resume the game. Determining a predecessor announcing the start of the game; Outputting a game start signal in video or audio if the predecessor is a computer, and if the predecessor is a user who is a game participant, requesting a game start signal from the user in video or audio and then receiving a game start signal; Receiving an image including a user's face and hands; The computer selecting one of the scissors, the rock, and the beam, and outputting it as an image or an audio; Analyzing the received image to determine which of the user's hands is a mutt chipper; If the user's hand shape and the computer's hand shape are determined as described above, the result of the game is outputted. If the hand shape does not match, the predecessor is determined according to the game rules. As a game of how to play mutton, The determining of the shape of the hand of the user may include detecting a face region from an image including a hand shape of a game participant, collecting skin color information of the face from the detected face region, Detecting a hand region based on skin color information, and analyzing a hand shape based on the detected hand region information. The method of claim 20, Before the step of determining the predecessor to inform the start of the game, further comprising the step of receiving and storing the title of the user, and in the step of outputting the win or loss of the game, calculating the total of the game participants and outputs along with the received title Mukchippa game method comprising the step of further comprising. delete An image input module configured to receive an image captured to include a face and a hand of a target user; A face region extraction module for extracting a face region from the input image of the user; A face region color information extraction module for collecting skin color information of the face from the extracted face region; Based on the skin color information of the collected face area, the area determined as the skin color in the image of the captured user is classified as an isolated area, and the largest area within the classified area is determined as the palm area to be a portion corresponding to the hand. Hand region extraction module for extracting the; And, And a hand recognition module configured to determine which hand shape the user has made based on the extracted hand region. The method of claim 23, wherein And a left hand / right hand discrimination module configured to check whether the hand used by the user is a right hand or a left hand based on the face information recognized by the face region extraction module. An image input apparatus for photographing a user who is a game participant and generating an image including a face and a hand of the user; An image input module for receiving the image, a face region extraction module for extracting a face region from the input image, a face region color information extraction module for collecting color information about a face from the face region, and a color of the collected face region A hand shape recognition device including a hand region extraction module for extracting a portion corresponding to a hand based on the information, and a hand recognition module for determining which hand shape the user has made based on the hand region; A storage device for storing information about the user and the user's hand; A controller for determining a corresponding operation of the game device using the hand based on the hand shape determined by the hand recognition device, and controlling the game device using the hand; And, Game device using a hand comprising a display unit and a voice output unit for delivering the information determined by the control unit to the user. The method of claim 25,  Hand game device further comprises an input device for receiving data about the user from the user. The method of claim 25, The image input apparatus may be an optical camera, a CMOS or CCD optical sensor. The method of claim 25, The image input device is a hand-held game device, characterized in that the image input device capable of imaging at 320x240 resolution within a distance of 0.5 ~ 2m The method of claim 25, The image input device is a game device using a hand, characterized in that positioned so that the user's hand can be captured from below

Images