KR0126443B1 - Swing motion analysis system of golf - Google Patents

Abstract

The golf T-shot swing motion analyzing system comprises a shot analyzer 2, a weight distribution measuring device 3, and an image analyzer 4. The shot form analyzer 2 finds out the velocity and direction of the golf club and the golf ball by analyzing the image detected by a photo-detector. The weight distribution measuring device 3 finds out the change of the body weight distribution when swing the T-shot. The image analyzer 4 takes the swing image and finds out the necessary information to correct the swing form.

Description

Golf Tee Shot Swing Motion Analysis System

1 is a block diagram of a tee shot swing motion analysis system.

2 is an interface block diagram of an indicator.

3 is a block diagram of a hardware system for tee shot swing motion analysis.

4 is a block diagram of an interface board for batting analysis.

5 is a layout view of the optical sensor for hit analysis.

6 shows how to distinguish club faces.

7 is a view showing a method of dividing the hitting direction into nine types.

8 is an input waveform diagram for weight shift analysis.

9 is a graph showing left and right weight shift waveforms and 9 specific position diagrams of a standard swing.

10 is a graph showing the location and definition of features for setting neural network input vector values.

11 is a graph showing a function definition for setting neural network input vector values.

12 is a diagram showing a neural network structure implemented for weight shift waveform analysis.

13 is a graph showing the position ratio between nine feature points in a standard swing waveform.

14 is a flowchart for extracting nine feature points.

15 is a flowchart for extracting nine feature frames from a motion image.

16 is a pyramid structure diagram of an image for motion image analysis.

FIG. 17 is a diagram showing a brightness search region for extracting five feature frames such as back swing 1, back swing 2, top, down swing 1, and follow through.

18 is a time interval diagram between feature frames in standard swing motion.

The present invention relates to the analysis of the tee shot swing motion during a golf exercise, a golf ball and a ball analyzer for analyzing the speed and direction of the golf ball, a weight movement analyzer for analyzing the movement of the weight distribution during the tee shot swing and tee shot swing motion It is characterized by consisting of three analysis equipment of the image analyzer that can analyze the image of the.

Conventional tee shot swing motion analyzer has not comprehensively performed batting analysis, weight shift analysis and image analysis necessary for comprehensive analysis of motion.

Also, these analyzes were more interesting systems than scientific physical exercise analysis.

For example, the batting analysis system is generally a system that shows the changed scene of the golf course using a large screen rather than the technology of the batting analysis itself.

In addition, the analysis technique did not satisfy the problems to be solved as a technique of artificial intelligence by using a simple mathematical analysis method, and the operation of the system was possible only by the user's intervention.

Since this approach is developed and operated without accurate scientific support, it has distrusted golfers who want to accurately analyze their golf swing motion and correct their incorrect posture, and it has a bad influence on sports science by computer analysis. Therefore, in order to be reliable, the existing analysis approach requires the latest advanced scientific approach, and the analysis should be not only one method but also a comprehensive analysis using three analyzers at the same time. It must be a system.

The first object of the present invention is to integrate the three analysis methods required for the tee shot swing analysis to provide a comprehensive analysis results.

The second object of the present invention is the scientific approach of each analysis method.

Therefore, batting analysis was performed using table matching by optical sensor and weight shift analyzer was developed by using neural network. In addition, the image analyzer automatically extracts nine feature frames from the continuous frame by pattern recognition and time interpolation, and performs a comparative analysis with a standard model.

A third object of the present invention is to derive an analysis result that is most suitable for various physical conditions of a golfer by selecting and analyzing a model most similar to each golfer to select a standard model for analysis.

The fourth object of the present invention is to automate the entire process of analysis so that no user intervention is required.

The fifth object of the present invention is to derive a real-time analysis result that can be obtained immediately after completion of the tee shot swing by minimizing the analysis time.

A sixth object of the present invention is to visually show three analysis results, so that even a beginner can easily understand the analysis results.

In the present invention, the tee shot of golf, which is the subject of physical exercise analysis, is a high speed at which the movement speed of the club head is about 200 km / h at impact.

It is very difficult to analyze such a high speed motion using only image information input from a CCD camera.

Therefore, the tee shot receives the information about the club head's trajectory and ball movement, and calculates the ball's movement speed and the direction of the ball, and analyzes the results. By performing weight movement analysis to analyze distribution and motion analysis by image, it is possible to comprehensively analyze physical movement during tee shot swing.

The present invention, which is a system for correcting a golfer's posture during swing, is composed of three analyzers, such as a batter analyzer, a weight shift analyzer, and a motion image analyzer, for more scientific and accurate analysis.

Figure 1 shows a block diagram of these three analyzers and indicators.

In the indicator 1, the operation button is composed of six buttons 101a to 101f of 'START', 'MODE', 'SELECT', 'LEFT', 'RIGHT' and 'PRINT'. 'START' button 101a instructs the system to start operation. The 'MODE' button 101b is used to change the mode of the display. The SELECT button 101c is used to select clubs or models to compare.

The LEFT and RIGHT buttons 101d and 101e control the horizontal movement of the screen. Finally, the 'PRINT' button 101f is used to print the analysis result using the printer 9.

The block diagram of the interface board 102 for the operation button 101 interface in the indicator 1 is shown in FIG. Scan data input from the operation button 101 is input after being decoded through a latch 102a on the interface board 102. At this time, the operation of the interface board 102 is controlled by a central processing unit (CPU) 5 on the PC.

As shown in FIG. 3, when the golfer climbs on the electronic weight scale 301 and 302 and presses the START button 101a to the club, light sensor information, weight movement information, and motion image information are started. These three pieces of information are entered through the respective interface boards 209, 303, 402.

The swing end signal automatically detected by using the club track information obtained from the optical sensor information input from the batting analyzer 2 terminates the input of the optical sensor information and is transmitted to the weight analyzer 3 and the image analyzer 4. The input of the weight shift and the input of the motion image are terminated.

Using the optical sensor information input during this time, the batting analyzer 2 analyzes the speed of the club head, the club face, the trajectory of the backswing and downswing, the impact point, and the direction of the ball. The weight shift analyzer 3 detects the swing region from the input weight shift data and determines the characteristics of the swing in the swing region, Ad, back swing 1 (B ₁ ), and back swing 2 (B ₂ ). , 9 operating points of Top Tp, Down Swing 1 (D ₁ ), Down Swing 2 (D ₂ ), Impact (Im), Follow Through (Ft) and Finish (Fn) are automatically extracted.

In addition, the image analyzer 4 automatically extracts, in real time, a frame corresponding to the preceding nine operating points from the input motion image.

In the batting analyzer 2, the information inputted through the optical sensors 201 to 208 is transferred to the computer-first CPU 5 by the Z80 processor on the interface board 209 as necessary information are shown in the block diagram of FIG. From there, it analyzes the club head's direction, angle, and the ball's position and calculates the swing tempo, club head speed, club face, swing path, and ball direction. These information are displayed on the first display 6.

In addition, by using the information input from the photosensors 201 ~ 208 during the swing automatically detects the end of the swing and transfers it to the weight shift analyzer (4) to terminate the input of these information.

The block diagram of the interface board 209 for the operation of the hit analyzer 2 is shown in FIG. The signal input every 168 nS (nano second) from the high speed optical sensors 201 to 208 is amplified by the amplifier 209a, and the data inputted as necessary through the latch 209b is read and stored in the memory 209c. . After extracting only necessary data from the data stored in the memory 209c, the format is arranged through a converter 209d.

The batting data which passed through the above process is transmitted to the PC-first CPU 5 through the communication port 209e.

A total of 64 optical sensor diodes for batting analysis are arranged in eight ports of eight, and the detailed structure thereof is shown in FIG.

Data input from the optical sensor is transmitted to the computer by the Z80 processor 209F in units of information input from one port.

As shown in FIG. 5, the functions of each port and the optical sensor are as follows.

P1 ~ P8: port number

④ to ⑦ of P8 is the start sensor

P8 ① to ③ are end sensors

Determine the club face based on P3 sensing data

The path for back swing and down swing is determined from the combined data (B1) of P1 and P2.

The impact point is determined from the combined data (B2) of P4 and P5.

The shot direction is determined from the combined data (B3) of P6 and P7.

From the information input by the designed batting analyzer 2, six pieces of information such as swing tempo, club head speed, swing path, impact point, club face and ball spin are derived.

Swing tempo is the time taken from the address to the impact, and the club head speed is the head speed at the point of impact.

The swing path represents the club's trajectory during the tee shot, and the impact point means the part where the club head hits the ball, and is classified into three types: heel, center, and toe.

The club face is divided into nine types, which are divided into an open state and a closed state with respect to 90 degrees. The hit direction is divided into nine categories based on three pieces of information: impact point, club face, and downswing pass.

The criterion for obtaining the above-mentioned six pieces of information from the data input from 64 light sensors is based on the formula or table matching in 1) to 6) below.

1) Swing tempo (unit: second)

Total scans / 8184.91

Here, the total number of scans means the number of times 64 optical sensor data is read from an address to an impact, and 8184.91 means the number of times 64 data is read per second.

2) Speed of club head (unit: km / h)

3600 * 0.00002 / (number of scans between B1 and B2)

Here, 3600 represents one hour in seconds, and 0.00002 represents the distance between B1 and B2 in km.

3) Swing Pass

Based on the median of the combined data (B1) of P1 and P2, it is divided into nine as follows.

Back swing pass

(1): Inside 4 (INSIDE 4) (2): Inside 3

(3): Inside 2 (4): Inside 1 (INSIDE 1)

(5): STRAIGHT (6): Outside 1 (OUTSIDE 1)

(7): Outside 2 (8): Outside 3 (OUTSIDE 3)

(9): Outside 4 (OUTSIDE 4)

Downswing pass

(1): Inside-out 4

(2): INSIDE-OUT 3

(3): Inside-out 2

(4): INSIDE-OUT 1

(5): STRAIGHT

(6): OUTSIDE-IN 1

(7): OUTSIDE-IN 2

(8): OUTSIDE-IN 3

(9): OUTSIDE-IN 4

4) Impact Point

Based on the median of the combined data (B2) of P4 and P5 is divided into three as follows.

5) Club Face

From the data of P5, as shown in FIG. 6 based on 90 degrees, it is divided into the open state and the close state, and they are divided into four types and divided into nine types.

SQUARE: 90 degrees

Open-1: 1 ~ 7 degree open (OPEN)

Open-2: 8 ~ 15 degree open (OPEN)

Open-3: 15-30 degrees open (OPEN)

Open-4: Open more than 30 degrees (OPEN)

CLOSE-1: 1 to 7 degrees close

CLOSE-2: 8 to 15 degrees close

CLOSE-3: 15 to 30 degrees close

CLOSE-4: CLOSE above 30 degrees

6) Ball spin

The batting direction is divided into nine types as shown in FIG. 7 based on the impact point, club face, and down pass information.

[Table 1] When the impact point is CENTER

[Table 2] If the impact point is a heel

[Table 3] Impact point is Toe

As shown in FIG. 1, the weight distribution movement analyzer 3 displays the continuous change of the weight on the left foot and the right foot during the tee shot swing of the golfer by means of two electronic scales 301 and 302. At 101, 12 times per second are detected from the moment when the START signal is input, and the information is received.

As mentioned above, the interface board 303 receives the end signal of the swing detected by the batter analyzer 2 and ends the input of the weight distribution. The weight distribution moving waveform input in this way has various forms depending on the swinging person. It is very difficult to classify these various types of waveforms by certain classification criteria. That is, it is difficult to find a rule that satisfies the swing condition for all types of waveforms, and there are many ambiguities.

In order to overcome this drawback, the present invention applies a method for classifying weight distribution moving waveforms using an error backpropagation neural network model having good performance for pattern classification. In order to use neural networks, it is necessary to select and extract input vectors that can efficiently distinguish patterns, and to improve recognition rates by adjusting the number and learning frequency of each node of the neural network. The result of analyzing the weight distribution shift waveform can be used as an important information for posture correction of the golf swing together with image analysis.

As shown in FIG. 3, when the golfer swings three to four times continuously on the electronic balances 301 and 302, an example of a waveform in which the body weight is extracted 12 times each second is shown in FIG. 8.

Nine motions of the golf swing in the weight distribution waveform are mapped to one point each.

Therefore, in order to analyze the weight distribution during the swing, one swing pattern as shown in FIG. 9 must be extracted from the continuous waveform as shown in FIG. By analyzing 9 feature points in the extracted swing pattern, we can know the characteristics of weight distribution shift during swing.

In the present invention, a total of eight input vectors used as input vectors of a neural network are defined to classify a weight distribution waveform during a swing. The method of extracting these input vectors is as follows.

First, find the intersection point in the continuous input waveform as shown in FIG. At the next intersection, the weight of the left foot is decreased and the weight of the right foot is increased. If applicable, the result is separated into a waveform having two intersection points as shown in FIG. In this separated waveform, the input vector values are found as in the example of FIG. 11, and the real values between 0 and 1 are normalized to input these values into the neural network.

That is, the function V normalizes the extracted values of the eight input vectors to a value between 0 and 1, and Max (WL), Min (WL), Max (WR), and Min (WR) correspond to the maximum of left and right, respectively, in the following equation. The minimum weight value is indicated. These input vectors analyze the input waveforms from several people, find common features, and apply these rules. Based on these rules, eight input vectors were determined to be most suitable for classification.

1) V (PrevRun): Vector value by time from the intersection of the first left and right body weight to the intersection of the second weight distribution

2) V (FirstRun): Vector value by time from the intersection of the second left and right weight distribution to the intersection of the third left and right weight distribution

3) V (SecondRun): Vector value by time from the third left and right weight distribution intersection to the fourth weight distribution intersection

4) V (PeakRun): The ratio of the width and height of the first vertex that exists within the interval between the third and fourth weight distribution intersections.

5) V (PrevRatio): The ratio between the minimum value of the right weight and the maximum value of the left weight within the PrevRun section.

6) V (FirstRatio): The ratio between the minimum value of left weight and the maximum value of right weight within the FirstRun interval.

7) V (SecondRatio): The ratio between the minimum value of the right weight and the maximum value of the left weight in the SecondRun section.

8) V (PeakLoc): Vector value by relative position with respect to the whole section of time when the first peak is present

The error backpropagation model implemented for weight distribution shift analysis consists of three layers and has eight input nodes and two output nodes. We used a learning rate of 0.1 and momentum of 0.9 as the neural network variables. The initial weight that connects each node at the time of learning has a value between 0 and 1 using a random function.

As the training data, 178 swings and 422 nonswings were extracted from 60 golfers and 600 data were used. Based on this, in order to construct the optimal neural network, the variation of recognition rate was examined by adjusting the hidden node and the number of training.

In the present invention, the best case is 10 hidden nodes, 5,000, 6,000, 7,000 learning counts, showing a high recognition rate of 97.75% or more.

Based on the experimental results thus obtained, we can construct an optimal neural network model when we train 5,000 times with 10 hidden nodes.

Figure 12 shows the structure of the implemented neural network.

In order to analyze the tee shot motion, it is necessary to find 9 feature points corresponding to Ad, B1, B2, Tp, D1, D2, Im, Ft and Fn, which are the characteristics of the golf swing, from the extracted weight distribution waveform. It occurs at the same time point as nine feature frames in the analysis. Address (Ad) using the location of the intersection and peak in the weight distribution waveform. Impact (Im) and follow-through (Ft) are easy to find.

Therefore, the present invention uses a method of first finding three positions that can be easily found and finding the remaining positions by time interpolation based on the three positions.

The most invariable and distinct feature of the feature points is the impact (Im) portion at the intersection.

Since the starting position of the waveform extracted using the neural network is near the address (Ad) of the swing waveform, in order to extract the impact (Im) part, it moves over a certain distance to the right from this position to find the intersection point and then the impact (Im) position. Save as.

Next, the position of the peak is found within a certain interval on the right side based on the impact position Im and stored as the position of the follow through Ft.

Next, find the intersection by moving to the left over a certain distance based on the position of impact (Im) and store it as an address (Ad) position. When the address Ad and the impact Im are found, the position of the saw Tp located about two thirds between the intervals is found.

Since the peak does not always occur in the position of the saw Tp, when there is no peak, the 2/3 point is set as the position of the saw Tp. Once the positions of the three feature points that can be easily found are found, the remaining feature points can be easily obtained by time interpolation.

13 shows a ratio of positions between feature points for obtaining the remaining feature points by time interpolation. In general, the golfer's total swing time is different, but the ratio between the nine feature point positions in the overall swing time is found to be mostly constant.

14 shows a flowchart of a feature point extraction algorithm. In this way, the weight distribution moving waveform calculated by the first CPU 5 is displayed on the first display 6.

The motion image analyzer 4 captures the golfer's tee shot scene with the CCD camera 401 from the front and analyzes the shot.

As shown in FIG. 1, the motion image is input at 30 frames per second from the moment when the 'START' signal is transmitted from the operation button 101 of the indicator 1. The end of the motion image input is performed at the moment when the swing end signal sensed by the hit analyzer 2 is transmitted to the interface board 402.

The image analyzer 4 automatically extracts nine feature frames necessary to correct the swing posture from the input motion image.

The motion image used for analysis has a size of 256 × 240 pixels in each frame. It takes a lot of processing time to analyze the original image of each frame of 256x240 pixels in the motion image displayed at 30 frames per second.

Therefore, in order to reduce processing time, a pyramid image showing a layered structure is constructed by gradually reducing the size of each frame of the input motion image. Back swing 1 (B1), back swing 2 (B2), top (Tp), and down swing (1) using the brightness information of the feature area where a club may exist at the edge of the pyramid image except the golfer's image area. (D1) and extract five feature frames of follow through (Ft). In the four frames of the address (Ad), downswing 2 (D2), impact (Im) and finish (Fn) except these, the club and the golfer's images overlap or the club moves at high speed. It is very difficult to determine the club's location using the brightness information.

Therefore, these four frames are obtained by temporal interpolation between five feature frames obtained using the brightness characteristic.

Through the above process, nine feature frames necessary for analyzing the swing posture can be extracted. Motion image data is supplied to the second CPU 8 to extract the feature frame.

15 shows a flow for extracting nine feature frames.

In order to process the tee shot motion image input at 30 frames per second in real time, the input image is hierarchized into a pyramid structure.

I _{k = 1} (i, j) = (1/4) × (I _k (2i, 2j) + I _k (2i, 2j + 1) + I _k (2i + 1,2j) + I _k (2i + 1,2j + 1))

The pyramid structure of the image has a hierarchical form as shown in FIG. In the pyramid structure, the size of an image of one layer is 1/4 of the size of an image of the previous layer. In order to ensure the accuracy of the interpretation in the pyramid structure, the problem of determining the depth of the pyramid hierarchy is important.

In the present invention, the height from the lowest layer to the highest layer of the pyramid is composed of three layers.

The size of the image I ₀ from the lowest layer, layer 0 of the pyramid construction in the present invention is the image I ₁ in a same 256 × 240 pixels and the size of the original image, the intermediate layer, the layer 1 The size of the layer 0, the size of 1 / 4 pixels of 128 x 120 pixels.

Image I ₂ on the size of the p-top layer, a mid layer 2 is 1/4 of 64 × 60 pixels of the first layer size. The higher layer image is not considered in the present invention because the degree of preservation of the original image information is significantly reduced.

In the present invention, a black background and a silver white club are used, and it is assumed that the golfer wears white gloves. This assumption is the most suitable environment in the indoor golf driving range to which the present invention is applied. In the input image I, the club position is defined as a high brightness region that appears in the edge region excluding the golfer's image region, and the feature frame is extracted using the club position information. In this case, since only the edge area of the input image is searched except for the golfer area, the clothing of the golfer does not affect the performance of the algorithm.

In order to locate a club, we use a priori knowledge of the area in which a club may exist in each feature frame.

Extraction of five feature frames by brightness characteristics in a specific region is performed in the following manner. First, the input motion image is displayed from the first frame, and each frame is constructed in a pyramid structure having three layers.

Next, the feature frame is extracted using brightness information of a specific area where a club may exist in the top layer of the constructed pyramid.

17 shows a search area for selecting a feature frame based on brightness characteristics in five feature frames.

Experimental results in the environment constructed in the present invention was found that the difference in brightness between the background and the club is at least 50 in the motion image having a resolution of 8 bits each RGP.

In back swing 1 (B1), the club faces the lower left of the screen.

Therefore, as shown in FIG. 17A, an area R _b1 is defined to detect whether pixels having high brightness appear at the bottom left of the screen, so that the image at the moment when pixels having a brightness difference greater than or equal to 50 is displayed in R _b1 . Select (B1). Similarly, in back swing 2 (B2), the club is located at the top left of the screen, so the area R _b2 is defined at the top left of the screen to select the costume as the back swing 2 (B2) when the bright pixels appear in R _b2 . do.

In the top Tp, it is positioned horizontally in the upper right corner of the club screen. The top (Tp) is the moment that changes from the back swing to the down swing, so the club descends from the top right of the screen and then rises again.

Therefore, the area R _tp is defined on the upper right side of the screen, and the top Tp is used to select an image at the moment when high brightness pixels appear at the lowest position of R _tp .

In down swing 1 (D1), the club faces the upper left corner of the screen.

Therefore, as shown in FIG. 17B, the area R _d1 for detecting whether high brightness pixels appear in the upper left of the screen is defined, and the image at the moment when the high brightness pixels appear in R _d1 is selected as the down swing 1 (D1).

Finally, in follow-up (Ft), the club is located on the upper right side of the screen, so define the area R _ft in the upper right-hand corner of the screen and select the image of the moment when the bright pixels appear at R _ft as the follow-up (Ft). .

Back swing 1 (B1), back swing 2 (B2), down swing 1 (D1), and follow through (Ft) of the nine feature frames desired by the club's position information from the motion image input using the above method. And five feature frames.

Four feature frames (Address, Ad, Downswing 2 (D2), Impact (Im) and Finish (Fn), except for 5 feature frames obtained using the brightness information of the area where the club can appear in the feature frame) are obtained. Obtained by time interpolation from two feature frames. The reason for using the temporal interpolation method is that it is difficult to extract using the brightness information of a specific area since the club overlaps the golfer's video area or moves at high speed as described above.

Experimental results on several swing motion images show that the time between down swing 1 (D1) and down swing 2 (D2) is 1 for the normal swing, and the time between down swing 2 (D2) and impact (Im) About 1 and the time between impact Im and follow through Ft is about 2. The time between the address Ad and the back swing 1 B1 is about 4, and the time between the follow through Ft and the finish Fn is about 5.

18 shows time intervals between feature frames used for time interpolation.

Through the above process, nine feature frames that can be used as basic motion frames for comparison and review with the standard tee shot motion operation of the model can be extracted. If the image corresponds to the feature frame, the second CPU 8 controls the motion image analyzer 4 to output the image to the second display 7.

The effect of the present invention is to enable the comprehensive and scientific analysis of physical movement by using the batting analyzer, weight movement analyzer and image analyzer at the same time in the tee shot swing motion analysis, and break away the conventional simple mathematical analysis method by using artificial intelligence technology In this paper, you can find the rules that satisfy the swing conditions for all types of waveforms in the weight shift analysis waveform, which shows various shapes according to the swinging life.You can perform real-time analysis by minimizing the analysis time in motion image analysis. The system's operation is done automatically without user intervention, and it can contribute to sports science by applying the system to motion analysis in various sports fields.

Claims (16)

Input means (instructions) for inputting a system operation command by a user; A batting analyzer for receiving information such as the direction of the club head, the angle, the correct position of the ball, and the like, by detecting the end point of the swing, and processing the recognized information; In order to analyze the weight distribution moving waveform, the change in weight on the left and right feet of the golfer is continuously measured from the moment when the START signal is input from the input means (indicator) to the end of the swing detected by the hit analysis. A weight shift analyzer for processing data; Calculate swing tempo, club head speed, swing path, impact point, club face and batting direction by receiving the processed information from the batting analyzer, and calculate the weight distribution moving waveform by receiving the processed information from the weight moving analyzer. A first CPU for outputting the calculated result; A first display for displaying a swing tempo, a club head speed, a swing path, an impact point, a club face, a hitting direction, and a weight distribution movement waveform calculated in the first CPU; A motion image analyzer for outputting motion image data by capturing a swing image of a golfer from a moment when a START signal is input from the indicator to a swing end point detected by the batter analyzer to extract an image during swing posture correction; A second CPU that receives the motion image data output from the motion image analyzer and analyzes and outputs a swing motion of a golfer; Golf tee shot swing motion analysis system comprising a second display for displaying a swing motion analysis image of the golfer output from the second CPU. The motion image analyzer of claim 1, wherein the motion image analyzer is connected to a second CPU and supplies motion image data to the second CPU to extract a feature frame from an input motion image, and the second CPU corresponds to a feature frame of the motion image data. If it is an image, the golf tee shot swing motion analysis system, characterized in that for controlling the motion image analyzer to output the image to the second display. The apparatus of claim 1, wherein the batter analyzer comprises: a plurality of optical sensors configured to sense information during a swing; An amplifier for amplifying signals input to the plurality of optical sensors; A latch and a memory for storing data amplified by the amplifier; A converter for formatting necessary data among the data stored in the memory; A communication port for transmitting the formatted data to a first CPU; A microprocessor controlling the latch, the memory, the converter and the communication port to store data in the memory, format the converter, and transmit the formatted data to the first CPU. Motion analysis system. The method of claim 1, wherein the input means (instructions) comprises a plurality of operation buttons for inputting an appropriate operation command of the system, and the operation button and the interface board for exchanging data with the ball analyzer and the first CPU. Golf tee shot swing motion analysis system. The optical sensor of claim 3, wherein the optical sensor of the batting analyzer comprises 64 optical sensors including eight ports in total for detailed analysis of the impact shot and three blocks based on information to analyze each port. Golf tee shot swing motion analysis system characterized in that the sensor arrangement. 6. The golf tee shot swing motion analysis system according to claim 5, wherein the swing path is classified into nine types by data input from each port of the optical sensor array. 6. The golf tee shot swing motion analysis system of claim 5, wherein the impact points are distinguished based on a combination of two ports (ports 4 and 5) of the light sensor array. 6. The method according to claim 5, wherein the club face is divided into four types, each of which is divided into an open state and a closed state based on 90 degrees from data of any one port (port 5) of the optical sensor array. Golf tee shot swing motion analysis system, characterized in that divided into nine categories. The method of claim 5, wherein the end of the swing is sensed by using information obtained from four optical sensors (photo sensors 0 to 3) of any one port (port 8) of the optical sensor array, and the detected signal is used. A golf tee shot swing motion analysis system characterized in that it terminates input of information from a batting analyzer, a weight moving analyzer, and a motion image analyzer. The weight shift analyzer of claim 1, wherein the weight shift analyzer uses a neural network to find a swing region from a continuously input weight shift distribution waveform, and detects a shift of an input weight based on a vector value given to each input node of the neural network. Golf tee shot swing motion analysis system characterized in that obtained from the pattern of the waveform. The method of claim 10, wherein the nine feature points necessary for the analysis of the swing posture are extracted from the extracted swing region using the input vector value of the neural network, and the weight movement analysis is performed using the extracted feature points. Golf tee shot swing motion analysis system. The pyramid image of claim 1, wherein the motion image analyzer constructs a pyramid image having a hierarchical structure by gradually reducing the size of each frame of the input motion image to reduce processing time. A golf tee shot swing motion analysis system comprising selecting a specific area where a club may exist at an edge except for an image area of a golfer, and extracting a feature frame of a motion image using brightness information of the specific area. The method of claim 12, wherein five feature frames of back swing 1, back swing 2, top, down swing 1, and follow through are first extracted from the feature frames, and four feature frames of remaining address, down swing 2, impact, and finish. Golf tee shot swing motion analysis system characterized in that the extraction based on the five feature frames in a time interpolation method. The method of claim 12, wherein the feature frame of the motion image analyzer is extracted into nine feature frames of address, back swing 1, back swing 2, top, down swing 1, down swing 2, impact, follow through, and finish. Golf tee shot swing motion analysis system. 12. The golf tee shot swing motion analysis of claim 11, wherein the feature points are extracted into nine feature points: address, back swing 1, back swing 2, top, down swing 1, down swing 2, impact, follow through, and finish. system. The method of claim 16, wherein three feature points that are easy to find among the feature points, an impact, and a follow through are searched based on impact, and the back swing 1, the back swing 2, the top, the down swing 1, and the down swing except for the three feature points are identified. Golf tee shot swing motion analysis system characterized in that six feature points such as 2 and finish are obtained by time interpolation based on three feature points.