JPH11213167A - Device, method for detecting position of target object and form analyzing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method capable of exactly and automatically detecting a position of a target object with small amount of memory capacity and in short processing time by adapting two stages of retrieval algorithm. SOLUTION: Digital image information in a prescribed retrieval area including a predicted position where the target object exist is stored in a memory 16 for detection. The retrieval area is divided into plural blocks, blocks are selected one by one in a prescribed order by a first block selecting part 41 and image information of the selected block is analyzed by an image information analyzing part 42. An intermediate target block in which the image information of the target object is included is decided from an analysis result of the image information analyzing part by an intermediate target deciding part 43 (rough mode retrieval). The block is selected as shifting segmenting positions of the blocks by every pixel by centering around the intermediate target block by a second block selecting part 44 and the image information of the selected block is analyzed by the image information analyzing part 42. A final target block of the target object is decided from the analysis result of the image information analyzing part by a final target deciding part 45 (fine mode retrieval).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic position detecting device and method capable of detecting the position of a target such as a golf ball at the time of golf practice with a small memory capacity and a short processing time, and a form analysis using the same. Related to the device.

[0002]

2. Description of the Related Art A form analysis system that captures and reproduces and displays a golfer's swing can perform detailed analysis focusing on the moment of impact by using current digital image processing technology. In order to accurately capture the moment of impact in such a system, it is necessary to accurately detect the position where the golf ball is located in advance. In the case of a driving range, the position where a golf ball is placed is determined by the position of the tee when teeing up. If you use iron,
Generally, a golf ball is placed on an artificial turf of a predetermined size and swings, so that the position of the golf ball is limited within the range of the artificial turf.

The reason that the position of a golf ball cannot be uniquely determined when teeing up or using artificial turf is that even if the position of the ball is predicted to some extent using an automatic ball launching machine, the height of the tee is high. This is because the ball position may change each time depending on the golfer's setting of the ball position on the artificial turf. Since the above-mentioned system is equipped with a monitor device capable of confirming a form on the spot, a golfer or an instructor using the system can artificially indicate the position of the ball to the system each time. However,
It is troublesome to repeat such manual operation many times.

One method of realizing a system for automatically detecting the position of a golf ball is to take an image of the ball as a pattern and perform pattern matching (template matching). In this type of method, (1) a difference between luminance values is obtained for all the pixels in the pattern area, and a position where the sum of the absolute values is smallest is obtained.
The ball position is automatically detected by a method such as (2) obtaining a difference between luminance values and finding a position where the sum of the squares is the smallest, and (3) obtaining a correlation between a pattern area and a search block.

[0005]

In the above-described conventional automatic position detection method, there is a problem that dedicated processing hardware is required to perform pattern matching in real time, and calculation time and calculation cost are required. .
Also, in the case of automatic detection, since the judgment is often made by one pattern matching process, the recognition rate of the actual ball position remains uneasy. Further, when the surrounding brightness changes, the accuracy of pattern matching decreases. In addition, if a memory for one field is used to perform high-speed processing, the cost increases.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and provides a target position detecting apparatus and method and a form analyzing apparatus which can automatically detect a target position accurately with a small memory capacity and a short processing time. It is intended to provide.

[0007]

In order to achieve the above-mentioned object, a target position detecting apparatus according to the present invention provides a target position detecting device which includes at least a target area in image information including a target to be detected. Image storage means for storing image information in a search area corresponding to the target area; and dividing the image information in the search area into a plurality of blocks each consisting of M × N pixels corresponding to the size of the target object. First block selecting means for selecting one block at a time in order, and first position for roughly determining a block closest to the position of the target based on image information of the block selected by the first block selecting means Determining means and, for the block roughly determined by the first position determining means, selecting a block sequentially shifted by one pixel around the block. Block selecting means, and second position determining means for determining in detail a block closest to the position of the target based on image information of the block selected by the second block selecting means. And

In the method for detecting the position of a target according to the present invention, a search area is set at least in a range in which the target is predicted to be located in image information including the target to be detected, and at least a search area within the search area is set. Storing image information, dividing the stored image information of the search area into a plurality of blocks each including M × N pixels corresponding to the size of the target, and selecting one block at a time in a predetermined order; After roughly determining the block closest to the position of the target based on the image information of the selected block,
The block closest to the position of the target is finally determined based on the image information of the obtained block while shifting the block one pixel at a time around the block.

According to the present invention, the position of a target can be automatically detected accurately in a short time in accordance with a two-step target detection algorithm. That is, in the first stage, the block closest to the position of the target is roughly determined by searching for each block in the search area by the first block selecting means and the first position determining means. The second stage is
The detailed position of the target is determined by a search performed by the second block selecting means and the second position determining means on a pixel-by-pixel basis centering on the roughly determined block.
According to the present invention, only the image information for the search area needs to be selected and processed from the image information for one field, and if the image information for one field is not stored, the size is smaller than the field memory. Processing can be performed by storing image information in the detection memory for the search area of the capacity. Therefore, the capacity of the detection memory is basically unnecessary or small.

[0010] The position of the ball during golf swing, etc.
Since the position of the target is generally almost the same, the predetermined order of selecting the blocks by the first block selecting means is, for example, starting from the block including the previous position as the starting point and spiraling the surrounding blocks. If the transition is made, or if the transition is made from the block with the highest probability that the target is located among the blocks detected up to the previous time to the block with the lower probability, the block in which the target is located can be searched in a short time. The determination as to whether the target is located in the block is made based on, for example, the average value of the luminance values of all the pixels in the block, or based on the hue and / or saturation of all the pixels in the block. . When the target is a golf ball at a golf driving range, the size of the block is set substantially equal to the image information of one golf ball, and since the golf ball is often white, the brightness value is the highest. The block is the block where the golf ball exists.

Further, a form analyzing apparatus according to the present invention includes a moving image capturing means for capturing a series of forms of a subject, an image recording / reproducing means for recording and reproducing moving image information captured by the moving image capturing means, In a form analysis apparatus comprising: a display unit for displaying moving image information recorded in a recording / reproducing unit, the image recording / reproducing unit sets a search area for a target to be detected in a display screen of the display unit. Area setting means, image storage means for storing moving image information of the search area set by the search area setting means, and moving image information of the search area stored in the image storage means each having the size of the target. It is divided into a plurality of blocks of corresponding M × N pixels, and each block is selected one by one in a predetermined order, and based on image information of the selected block. After roughly determining the block closest to the position of the target, for the determined block, a block sequentially shifted by one pixel around the block is selected, and the image information of the selected block is added to the selected block. Target position detecting means for determining in detail a block closest to the position of the target based on the target, and monitoring the target detected by the target position detecting means to record and reproduce the moving image information. And control means for controlling.

As described above, by applying the present invention to a form analyzing apparatus, recording such as recording a series of moving image information by capturing the timing of loss of a target object, for example, detecting the impact of a golf ball. Playback control becomes possible, which is extremely effective for form analysis and the like.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of the present invention applied to a golf foam analyzing apparatus will be described below with reference to the drawings. First, a form analysis to which the present invention is applied will be described with reference to FIGS.

FIG. 1 is a perspective view showing the appearance of a form analyzing apparatus. This device includes a CCD camera 2 for imaging a golf swing of a subject (golfer) 1, an image recording / reproducing device 3 for recording and reproducing moving image information obtained by capturing an image of the subject 1 with the CCD camera 2, ,
It comprises a monitor device 4 for displaying a moving image or a still image reproduced by the image recording / reproducing device 3.

FIG. 2 shows a subject 1 captured by a CCD camera 2.
FIG. 7 is a diagram showing an example of a display screen 10 on which an image including “” is displayed on the monitor device 4. The CCD camera 2 includes, in addition to the subject 1,
Club head 5, golf ball 6, artificial turf 7, tee 8
Are set to positions and postures that are included below the display screen 10. In the display screen 10, an area where the golf ball 6 is likely to exist, that is, an area including the artificial turf 7 and its surroundings is set as the golf ball search area 9.

FIG. 3 is a block diagram showing the internal configuration of the image recording / reproducing apparatus 3 of FIG. The image input signal of the NTSC system supplied from the CCD camera 2 of FIG.
It is decoded and A / D converted by the NTSC decoder 11,
It is converted into time-series digital color image data based on horizontal / vertical scanning. This image data is M / JPEG
(Motion Joint Photograph Coding Experts Group) Compressed by the compression / decompression unit 12. The M / JPEG compression / decompression unit 12 converts the image data into compressed image data (JPEG bit stream) for each field. This compressed image data is stored in a memory 14 using a DRAM via a system bus 13. The memory 14 has a large capacity capable of recording, for example, 450 frames of compressed image data corresponding to a moving image of 15 seconds in the image storage area. The image storage area of the memory 14 has a ring buffer configuration, and the oldest data is always overwritten by continuously writing image data. The compressed image data read from the memory 14 is supplied to the M / JPEG compression / decompression unit 12 via the system bus 13, where it is decompressed. The decompressed data is sequentially read, and the D / A / NTSC encoder 15
And D / A converted and encoded, and output as a monitor signal of the NTSC system.

The detection memory 16 shown in FIG. 3 is a small-capacity memory or a field memory used for detecting the position of the golf ball 6. The memory controller 17 writes data in the search area 9 and data described in block units described later. Is controlled, and the image data in the search area 9 is stored for each field, for example. The detection of the position of the golf ball is processed by the CPU 18 and necessary software. When the CPU 18 recognizes that the instantaneous image of the impact has been recorded, the CPU 18 also executes processing such as driving the sound source 19, amplifying its output by the amplifier 20, and informing the speaker 21 of the loudspeaker. The operation unit 22 connected to the CPU 18 includes a switch, a volume, a light emitting diode, a joystick, a fader, and the like, and starts and ends recording, starts and ends reproduction, a recording mode, a reproduction mode, and a recording / reproduction speed of a moving image and a still image. In addition, setting of the search area 9 and the like can be performed.

FIG. 4 is a detailed view of the search area 9 of FIG. 2 which is the basis of the processing method for detecting the position of a golf ball. The search area 9 is divided into a plurality of blocks 31 each including M × N pixels 30. The size of one block 31 (the size of M × N) is preferably set to be substantially the same as or smaller than the image 6A of the golf ball 6. These can be set using a joystick or the like of the operation unit 22.

FIG. 5 shows a detection memory 16 and a memory controller 17 and a CPU
FIG. 9 is a functional block diagram of a portion that executes a golf ball position detection process including a golf ball 18 and necessary software.
The first block selection unit 41 includes digital image information of a predetermined search area 9 including a predicted position at which the target (the golf ball 6 described above) stored in the detection memory 16 is arranged, including M × N pixels. It is divided into a plurality of blocks 31 and selected block by block in a predetermined order. The image information analysis unit 42 analyzes the image information of one block 31 selected by the first block selection unit 41. The intermediate target deciding unit 43 analyzes the image information of the block 31 selected by the first block selecting unit 41 and, based on the analysis result of the image information analyzing unit 42, the intermediate target block including the most target image information 6A. To determine.

The search method is preferably as shown in FIG.
As shown in (A), the position where the ball was detected last time or the average position of the positions detected in the past is used as the start block 3.
As 1S, a method of searching spirally from the block, or as shown in FIG. 2B, a position where a ball has been detected in the past is recorded several times, and the detection frequency of each detected block is recorded. , And searching in order of frequency from the block with the highest frequency is conceivable. However, in the simplest case, the search area 9 is sequentially moved from the upper left to the right along the scan line as shown in FIG. A method may be used in which the search is performed, and when the right end is reached, the search is performed from the left end of the next scan line.

The second block selecting section 44 is configured as shown in FIG.
As shown in the figure, around the intermediate target block 31M,
While shifting the cutout position of the block by one pixel,
The target blocks 31 are selected one by one. The minimum target determination unit 45 determines the image information analysis unit 42 for each target block 31 selected by the second block selection unit 44.
As shown in FIG.
A block containing the most image information 6A of the target, for example, a block having the highest luminance value, is determined as the final target block 31. The mode switching unit 46 includes a rough mode using the first block selection unit 41 and the intermediate target determination unit 43, and a second block selection unit 44 and the final target determination unit 4
5 is switched to the fine mode. In the above configuration, a part of the image information analysis unit 42 and the intermediate target determination unit 43 constitute a first position determination unit, and another part of the image information analysis unit 42 and the final target determination unit 45 2 position determination means.

Next, the operation will be described. FIG. 8 is a flowchart of the ball position detection process. In the present invention, a two-step process of a rough mode and a fine mode is basically performed for one ball position detection. In the rough mode, a process of determining a threshold value for ball position detection is performed as needed ( S11). If a threshold has been set, the threshold is basically used (S12). In determining the threshold value (S11), the user may directly input the value. However, it is preferable to provide a processing routine for automatically determining the optimum threshold value as shown in FIG. . Then, for example, this threshold value determination processing is performed once at the time of starting the program, or this routine is executed once every predetermined time before the rough mode processing, and furthermore, when the user specifies as necessary. Execute the routine.

First, a flowchart showing the above-described threshold value determination processing (S11) will be described with reference to FIG. In this process, when detecting the position of the golf ball 6 from the luminance information, for example, the artificial turf 7 shown in FIG. 2 focuses on the characteristic of the detection environment in which the luminance is green and the luminance is low, and the golf ball 6 is white and the luminance is high. In this environment, a luminance threshold value for distinguishing the golf ball 6 from the artificial turf around the golf ball 6 is automatically detected. Is to seek. Therefore, this processing needs to be executed in an environment similar to an actual measurement environment and in a stable environment as much as possible. That is, during this process, it is necessary that the process be performed in an environment in which a person inadvertently crosses the screen and other movable elements are removed as much as possible. Specifically, first, the latest image is captured, and 0 is given as the initial value of the maximum value and the average value (S2).
1) In the search area, each pixel is sequentially scanned for each block (all areas can be collectively processed), and the average luminance value in each block and the final average luminance value of all blocks are calculated. It is determined (S22). In this process, the average luminance value of each block is compared with the maximum value, and if there is an average luminance value larger than the maximum value, the average value is set to the maximum value, and the position of the block at which the maximum value is obtained is determined. It is stored (S22). Step 22 is executed for all the blocks (S23), and when the difference between the maximum value of the average luminance value for each block and the average luminance value of all the blocks is within a predetermined value,
When it is determined that the ball is not placed, the process returns to the start (S21), and this is repeated until the ball is placed and an effective threshold value is calculated. If the difference is large, for example, if the maximum value of the average luminance value for each block is about twice the average luminance value of all blocks, it is determined that the ball exists in the search area 9 and an appropriate threshold value is calculated. Determine the threshold based on the formula. As the threshold value, a value between the maximum average luminance value for each block and the average luminance value for all blocks is set. However, the maximum average luminance value for each block is not always the optimum value. A better result can be obtained experimentally by using a slightly lower luminance threshold value in consideration of dirt and other luminance variation factors. In this example, the average of the average luminance value of each block and the average of the average luminance values of all blocks are set as the threshold value (S25). As described above, it is not necessary to set the threshold every time the ball is searched, but if the threshold is set periodically every time a certain time elapses from the process of determining the threshold,
It is possible to automatically cope with a case where the surrounding brightness or the like changes over time.

Once the threshold value has been determined, when the ball position detection processing is instructed by the mode control unit 46, the determined threshold value is used to perform the rough mode (S13 to S16) after (S12). Is executed. First, after the latest image is fetched (S13), for example, as described with reference to FIG. 6, if there is a memory of the position where the ball was detected in the past, it is used, as described with reference to FIG. Block search is performed in a predetermined order from a block position that is highly likely to be performed, and an average luminance in each block is obtained (S1).
4) A search is made for the presence or absence of a block having an average luminance exceeding the threshold (S15). When a block having an average luminance exceeding the threshold is detected, the block position is stored (S1).
5) Then, shift to the next fine mode. If the threshold value detection is optimized, there should be only one detection result in the rough mode. However, depending on the arrangement of the balls, the threshold value may be exceeded in any of the adjacent blocks. However, in this case, it is certain that the detected block is in any case around the correct ball position. As a result, the correct ball position is finally detected and specified by the next fine mode, and there is no problem. On the other hand, if no block having a luminance exceeding the threshold value is detected (S15) and all blocks in the search area 9 have been searched (S16), this is a situation where the ball has not been placed yet. Returning to the top of the rough mode again, the threshold value determination processing is normally bypassed, and the previous threshold value is used (S12), and the loop processing is repeated until the ball is placed in the search area 9. repeat. Depending on the environment arrangement in the search area, or due to an environmental change after the threshold detection mode, the threshold may be accidentally exceeded at a place other than where the ball is actually placed. It may be rare. However, according to the present invention, in order to realize a high-speed search, if there is a detection in the rough mode, the subsequent block search is aborted and the mode is shifted to the next fine mode. become. However, even in such a case,
When the user confirms the recorded image or the like based on the erroneous detection result, or when the search progress itself is confirmed by the monitor device 4 or the like described later, it is easy to confirm that the erroneous recognition is performed at that stage. If an incorrect recognition is found, the threshold is set again by user operation and reset to a threshold that does not cause incorrect recognition. By resetting to be forcibly excluded from the search range in the rough mode, it is possible to avoid subsequent erroneous recognition.

Next, the fine mode (S17 to S20) will be described. In this fine mode, the average value of the luminance of all pixels is obtained for the block whose position has been stored in step 13 of the previous rough mode (S17), and the cutout position of the block is set to one pixel in accordance with the predetermined order described with reference to FIG. While shifting by the unit (S18), the average value of the luminance of all pixels is obtained again for the shifted block (S1).
7) This is repeated for the K blocks within the predetermined range (S19), and the block 31F having the largest luminance value among the K blocks is set as the ball position (S20). In the above-described search process, if the position to be searched for each field is displayed on the screen of the monitor device 4 with a cursor, the golfer 1 is visually informed that the search is currently being performed and the state of the search. This is extremely convenient for knowing the timing of the start of swing and the like.

The position search may be based not only on the luminance information but also on the hue or saturation of each pixel included in the color image or a combination of both.
In particular, in an environment where the surrounding brightness changes, the brightness fluctuates, but the hue and saturation are stable, which is convenient. It is desirable that the size of the block 31 be set to be substantially equal to or smaller than the image 6A of one golf ball 6.

The search area 9 is not necessarily one, but a plurality of sets are set. If a target object cannot be detected in one search area, the processing shifts to another search area and the same processing is repeated. Good. For example, in the case of irons and woods, the position where the ball is present is usually largely distributed over two areas. For this reason, the method of setting a plurality of sets of search areas in this way, for example, when changing the place where a golfer hits with wood and iron,
It can respond widely. From which search area the next ball search is started is solved by setting the previous search area as the start area. Then, when a ball is detected in this start area, the process is ended there, and only when the ball is not detected, the process moves to the other search area. This change in the search area does not occur frequently in the general form of golf practice.

When practicing in a situation without an automatic ball launching device, for example, a plurality of balls 6 may be placed on or around the artificial turf 7.
It is common practice to place the player in a casual manner and draw one of them with the club head 5 to strike. In such a case,
All the balls 6 in the search area 9 are detected, and one of them is detected.
That is, the ball position on the rightmost side in the screen of FIG. 2 may be determined as the target position (the reverse is true for a left-handed golfer). Or, the ball you are about to hit,
Since one ball is isolated from another ball, all the distances between a plurality of scattered balls can be obtained, and the position of the ball having the largest dispersion of the distance can be determined as the target position.

FIG. 10 is a flowchart of a golf swing recording process using the above-described ball position search process. First, continuous recording to the memory is started (S31), and then the above-described ball position search processing is executed (S32).
In order to prevent erroneous detection of the ball due to a change in luminance or the like, it is determined whether or not the ball has been stationary at the same position for a certain period of time (S33).
Returning to 32, if it is determined that the ball is stationary, the ball at the detected position is monitored (S34). Then, it is determined whether or not the ball has disappeared as a result of the impact by the golf club (S35). If it has not disappeared, monitoring is continued (S3
If it is determined that 4) has disappeared, the recording is stopped after a predetermined time (S36). Thereby, the golf swing including before and after the impact can be recorded. By setting the predetermined time to an appropriate value, it is possible to make the impact time point coincide with the intermediate time point of the recorded moving image. Subsequently, the recorded swing is reproduced twice (S37), and the next recording operation is started. This process is repeated until there is a stop instruction (S38).

[0030]

As described above, according to the present invention, since the two-stage search algorithm is employed, the target position detecting device which can automatically and accurately detect the position of the target with a small memory capacity and a short processing time. And methods can be provided. More specifically, by reading information of a specific area of the field memory by the CPU and making a determination, real-time (once per field) automatic position detection becomes possible without a special image processing device. Further, the two-step position detection algorithm enables position detection with a higher recognition rate. Further, by selecting the search block in a spiral shape, the time until the target is detected can be reduced as compared with the simple scanning type. When the luminance value is used for image information analysis, the set value (threshold) of the luminance value for detecting a target is automatically set according to the brightness of the entire screen. There is an advantage that manual adjustment is not required.

[Brief description of the drawings]

FIG. 1 is a perspective view showing the appearance of a form analysis apparatus to which the present invention is applied.

FIG. 2 is a diagram showing an example of a display screen of the device of FIG.

FIG. 3 is a block diagram showing an internal configuration of the image recording / reproducing apparatus of FIG.

FIG. 4 is an explanatory diagram of a search area serving as a basis for a position detection process according to the present invention.

FIG. 5 is a functional block diagram illustrating an embodiment of a position detection device according to the present invention.

FIG. 6 is an explanatory diagram showing a block selection order in a rough mode according to the present invention.

FIG. 7 is an explanatory diagram of a fine mode of the present invention.

FIG. 8 is a flowchart illustrating an embodiment of a position search method according to the present invention.

FIG. 9 is a detailed flowchart illustrating the threshold value determining method of FIG. 8;

FIG. 10 is a flowchart of a recording mode process incorporating the position search method of FIG.

[Explanation of symbols]

REFERENCE SIGNS LIST 1 subject (golfer), 2 CCD camera, 3 image recording / reproducing device, 4 monitor device, 5 club head,
6 ... Target object (golf ball), 6A ... Ball image, 7 ...
Artificial turf, 8 tea, 9 search area, 11 A / D
NTSC decoder, 12 M / JPEG compression / decompression unit,
15: D / A / NTSC encoder, 16: detection memory, 17: memory controller, 18: CPU, 30 ...
Pixel, 31 ... block, 31S ... start block, 31M
... Intermediate target block, 31F ... Final target block, 41
... First block selector, 42... Image information analyzer 43
... intermediate target determination unit, 44 ... second block selection unit, 45
... A final target determination unit, 46... A mode control unit.

Claims (3)

[Claims] An image storage unit that stores at least image information in a search area corresponding to a range in which the target is predicted to be located among image information including a target to be detected; and image information of the search area. Is divided into a plurality of blocks each including M × N pixels corresponding to the size of the target, and a first block selecting unit for selecting one block at a time in a predetermined order; First position determining means for roughly determining a block closest to the position of the target based on the image information of the selected block; and for the block roughly determined by the first position determining means, Second block selecting means for selecting a block sequentially shifted by one pixel around the block; and a block selected by the second block selecting means. And a second position determining means for determining in detail a block closest to the position of the target based on the image information of the target. 2. A search area is set at least in a range in which the target is predicted to be located among image information including a target to be detected, at least image information in the search area is stored, and the stored search is performed. The image information of the area is divided into a plurality of blocks each including M × N pixels corresponding to the size of the target, and one block is selected in a predetermined order, and based on the image information of the selected block. After roughly determining the block closest to the position of the target, the block closest to the position of the target is finally shifted based on the obtained block image information while shifting the block one pixel at a time around the block. A method for detecting the position of a target, characterized in that the position is determined. 3. A moving image capturing means for capturing a series of forms of a subject, an image recording and reproducing means for recording and reproducing moving image information captured by the moving image capturing means, and a moving image information recorded on the moving image recording and reproducing means Display means for displaying a display area, wherein the image recording / reproducing means comprises: a search area setting means for setting a search area for a target to be detected in a display screen of the display means; Image storage means for storing moving image information of the search area set by the means; and a plurality of M × N pixels each of which corresponds to the size of the target object, storing the moving image information of the search area stored in the image storage means. , And one block at a time is selected in a predetermined order, and the block closest to the position of the target is enlarged based on the image information of the selected block. After the block is determined, for the determined block, a block that is sequentially shifted in units of one pixel around the block is selected, and based on image information of the selected block, the block that is closest to the position of the target object is selected. Target position detecting means for determining a close block in detail; and control means for monitoring the target detected by the target position detecting means and controlling recording and reproduction of the moving image information. Form analyzer characterized by the above-mentioned.