JP2000197003A - Dynamic image recorder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve image quality and to simplify arithmetic processings of a motion vector by extracting the feature part of frame image signal and changing the encoding form of respective frames, based on the result of detecting the motion of an object in correlation between adjacent frames. SOLUTION: At detection of striking motion of an object in a motion detection circuit 30, the default values for an N parameters and an M parameters are ignored, a GOP is forcibly closed and the practical number of the I pictures of satisfactory image quality is increased. An LPF 31, an HPF 32, a gate circuit 33 and a peak-holding circuit 34, etc., for constituting the motion detection circuit 30 are often loaded on the AF part of an electronic still camera and the circuit can be shared. Since whether or not the object frame of B picture encoding is provided with correlation with only the front or back can be detected by motion detection signals MV, motion compensation inter-frame prediction is applied only with respect to a reference frame with the correlation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture recording apparatus, and more particularly, to an MPEG (Moving Picture Experts Group).
p) The present invention relates to a moving image recording device such as an electronic still camera or a video camera that compresses and records a moving image using a standard encoding method.

[0002]

2. Description of the Related Art <Outline of MPEG Coding Method> MPEG
Is CCITT H.264 for the purpose of applying moving images to storage media systems and communication media systems. H.261 (encoding for videophone and videoconference). MP
EG is MC (Motion Compensation) and D
Basically, H.CT has an encoding tool such as a discrete cosine transform (CT). 261
However, they are different in that they have a special structure for realizing trick modes such as fast forward, rewind, midway reproduction, and reverse reproduction, that is, a GOP (Group Of Pictures) structure.

FIG. 7 shows an MPEG syntax (syntax; order and contents to be satisfied by a bit stream). This syntax is composed of a sequence layer having several GOPs between a sequence header and a sequence end, and a GOP layer below the sequence layer.
It has a structure having picture frames (encoded screen data).

[0004] One GOP is one unit of random access, and this unit enables the above-mentioned trick mode. The types (picture types) of the n picture frames are I picture (abbreviation: I), P picture (abbreviation:
P) or B picture (abbreviation: B), and the content of each picture type is as follows. (1) I picture Abbreviation of intra-coded picture.
This is an image in which all of the screen is intra-coded. It differs from other picture types in that it has independence within the GOP (does not require a reference picture). (2) P picture Predictive-Coded P picture between frames
icture). This is an image predicted in the forward direction from the previous I picture or P picture. (3) B-picture frame-interpolated bidirectionally coded image (Bidirectionally coded image)
Predictive-Coded Picture). This is an image that is bidirectionally predicted from preceding and succeeding I or P pictures.

FIG. 8 is a diagram showing an example of the GOP structure. The number of pictures (N parameter) of the GOP is set to “15”, and the cycle of the I or P picture and the P picture (M parameter) is set to “3”. This is an example of the case where That is, one GOP is composed of 15 frames,
Further, an example is shown in which the number of frames from an I picture (or P picture) to the next P picture is three.

In FIG. 8, an I picture is an intra-coded picture that does not require a reference picture, whereas a P picture and a B picture are forward-coded and bi-directional predicted coded pictures, respectively, as shown in the figure. , And P pictures are images that have been forward- and bi-directionally predicted using the already coded I or P picture as a reference image, and the B pictures using the preceding and succeeding I or P pictures as reference images.

In the case of an I picture, D
The CT operation, quantization, and variable length coding are performed. In the P picture and the B picture, DCT operation, quantization, and variable length coding are performed only on an error signal (prediction error signal) of motion compensation prediction between a reference image and a reference image. In particular, the compression efficiency is excellent when the prediction error signal is small, that is, when the moving image has a gentle motion.

FIG. 9 is a diagram showing a frame arrangement of a GOP. In this figure, the screen order of the original image (B0, B1, I
2, B3, B4, P5,...) Are partially replaced at the stage of the encoding process, and returned to the original arrangement order at the stage of the reproduced image. Insertion of a B picture in the encoding process is performed by inserting an I picture (or a P picture) before and after the B picture.
And after the encoding of the P picture. For example, when focusing on B3 and B4 of the original image, after encoding I2 and P5, B3 and B4 are encoded using these I2 and P5 as reference images, and inserted as a result after I2 and P5. In the order of “I2, B3, B4, P5” at the stage of
2,..., P5, B3, B4 ".

Here, the GOP generation rule only needs to satisfy that the first of the GOP on the medium is an I picture and that the last of the GOP is a P picture in the arrangement order of the original image and the reproduced image. , N parameters and M
Although there is no restriction on the use of the parameters, in practice, a value considered to be optimal from the image quality and the motion of the moving image, for example, N is a value such that the random access unit is about 0.4 to 1 second ( M is often set to a value (about "3") corresponding to an image having relatively gentle movement.

<Example of Moving Image Recording Apparatus> An electronic still camera converts an image of a subject into an electric signal by a two-dimensional image sensor, generates a frame image of a predetermined cycle from the electric signal, and depresses a shutter key. It captures a frame image at the time and records the captured image on a storage medium such as a flash memory, and records a still image of the subject like a silver halide camera,
2. Description of the Related Art Recently, there has been known an electronic still camera capable of continuously recording the movement of a subject by utilizing the above-described periodic generation operation of a frame image, and the above-described MPEG is used for compression recording of such a moving image. .

[0011]

However, in the above-mentioned conventional electronic still camera, the M parameter is fixedly set in advance, and the set value is adjusted to an image having a relatively gentle movement, for example, Since it is "3", M-1 = 2 B pictures are always inserted between the I picture (or P picture) and the P picture in the frame arrangement on the media, for the following reason. In addition, there is a problem that the image quality is deteriorated when the movement of the subject is severe.

As described above, the B picture is a frame-interpolated bidirectional predictive coded image, and this image includes an I picture and a bidirectionally predicted error signal (prediction error signal) from the preceding and succeeding I or P pictures. The same amount of information is compressed by applying the same DCT operation, quantization, and variable-length coding. When the movement of the subject is gentle, an appropriate compression effect can be obtained, which is effective. If it is severe, the prediction error signal will increase and the amount of information will increase. Therefore, in order to secure the required transmission rate, a "tight quantization process" is used. Is replaced with a representative value at an appropriate “interval”, and the number of the representative value is encoded. “Tight quantization processing” refers to increasing the interval. ………
This is a cause of image quality deterioration.

If the M parameter is set to "1", B
Although picture insertion is eliminated and image quality degradation can be avoided, the advantage of a B picture, that is, the effect of reducing the amount of information when the movement of the subject is gentle, is not preferable.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a moving image recording apparatus capable of achieving both reduction in the amount of information when the subject moves slowly and improvement in image quality when the subject moves rapidly.

[0015]

According to the first aspect of the present invention,
Input means for inputting a frame image signal of a predetermined period, encoding means for encoding the frame image signal by an MPEG standard encoding method, recording means for recording the encoded frame image signal on a storage medium, Extracting means for extracting a characteristic part of the frame image signal, detecting means for correlating the characteristic part between adjacent frames to detect a movement of a subject, and each of the encoding means based on a detection result of the detecting means. Changing means for changing a frame encoding method. According to a second aspect of the present invention, in the moving image recording apparatus according to the first aspect, the changing unit sets the encoding method of the frame in the encoding unit to I when the remarkable movement of the subject is detected by the detection unit. It is characterized by changing to a picture. According to a third aspect of the present invention, in the moving image recording apparatus according to the first aspect,
The changing means operates the encoding method of the frame in the encoding means when the detecting means detects a remarkable movement of the object, and closes the current GOP and a new GOP.
Is started. According to a fourth aspect of the present invention, there is provided a conversion unit for converting an image of a subject into a frame image signal having a predetermined period, an encoding unit for encoding the frame image signal according to an MPEG standard encoding method, and a frame after encoding. Recording means for recording an image signal on a storage medium, and extraction means for extracting a characteristic portion of the frame image signal,
Detecting means for detecting the movement of the subject by correlating the characteristic portion between adjacent frames, and changing means for changing an encoding method of each frame in the encoding means based on a detection result of the detecting means. The extraction means is used in combination with an automatic focusing circuit of an electronic camera. The invention according to claim 5 is the invention according to claim 1 or claim 4.
In the moving picture recording apparatus described above, the detection result of the detection means is applied to an MPEG motion compensation process, and motion compensation inter-frame prediction is performed only on reference frames having a correlation.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to the drawings, taking an electronic still camera having a moving image (movie image) recording mode as an example.

FIG. 1 is an external view of an electronic still camera. Although the illustrated electronic still camera 1 is not particularly limited,
The camera unit 3 is divided into a main body 2 and a camera unit 3 rotatably attached to the main body 2, and a photographic lens (not shown) is mounted on a front surface (a back surface side in the drawing) of the camera unit 3.
A CCD (image sensor), also not shown, is attached to the back of the photographic lens, and converts an image of a subject taken from the photographic lens into a video signal in a shooting mode, which will be described later. A frame image having a resolution can be generated.

On the other hand, the main body 2 confirms images (through images for composition adjustment, m captured still images of captured images or reproduced images of movie images, or m images of each screen constituting a movie image). For example, a flat panel display device, for example, a liquid crystal display 4 is mounted, and various operation keys such as a shutter key 5 are mounted at appropriate positions. The types and names of operation keys vary depending on the manufacturer and model and cannot be uniquely specified. For example, a plus key 6, a minus key 7, a menu key 8, a power switch 9, a display key 10, a shooting mode key 11, A timer key 12, a strobe mode key 13, a REC / PLAY key 14, etc. The functions (roles) of these keys are as follows.

(A) Shutter key 5: In the shooting mode, as the name implies, it functions as a "shutter key" (fixing exposure and focus by half-pressing, capturing an image or starting movie shooting by full-pressing). However, when the menu key 8 is pressed in the shooting mode or the reproduction mode, the multi-function key also functions as a YES key for understanding various selection items displayed on the liquid crystal display 4. (B) Plus key 6: a key used to select a reproduced image and various system settings. “Plus” means the selection direction, which is the direction of the latest image in the case of image selection, and the scanning direction of the liquid crystal display 4 in the case of system setting selection. (C) Minus key 7: Same function as the plus key except that the direction is reversed.

(D) Menu key 8: a key for performing various system settings. In the reproduction mode, various items including the delete mode (image erasing mode) are displayed on the liquid crystal display 4, and in the photographing mode, for example, the definition of the recorded image, such as the definition of the recorded image and the auto focus, In addition to ON / OFF, in the present embodiment, the liquid crystal display 4 displays a selection item for designating a shooting time of a movie image. (E) Power switch 9: a switch for turning on / off the power of the camera. (F) Display key 10: a key for overlappingly displaying various information on the image displayed on the liquid crystal display 4. For example, in the shooting mode, information such as the remaining number of images that can be shot and the shooting mode (normal shooting, panorama shooting or movie shooting) are displayed in an overlapped manner, and in the playback mode, attribute information (page number, definition, etc.) of the playback image is displayed. Display overlap.

(G) Shooting mode key 11: A key which can be used only in the shooting mode. In addition to selecting normal shooting or panoramic shooting, in this embodiment, in particular, a movie shooting mode is selected. (H) Self-timer key 12: A key for turning on and off a self-timer function. (I) Strobe mode key 13: Various keys for strobe light, for example, a key for forcibly firing light, prohibiting light emission, and preventing red-eye. (J) REC / PLAY key 14 This is a key for switching between a shooting mode and a reproduction mode. In this example, the switch is a slide switch, and slide up to set a shooting mode, and slide down to set a playback mode.

FIG. 2 is a block diagram of the electronic still camera according to the present embodiment. The electronic still camera according to the present embodiment has an automatic exposure (AE) function and an automatic focus (A).
F) functions are provided, and elements specific to these functions (for example, a light amount measuring sensor, a distance measuring sensor, an auto-focus driving mechanism, and a control mechanism thereof) are provided. Not shown in the block diagram.

In FIG. 2, reference numeral 15 denotes a photographic lens, 16 denotes a CCD (image sensor; corresponding to input means (conversion means) described in the gist of the invention), 17 denotes a horizontal / vertical driver, 1
8 is a timing generator (abbreviation is TG), 19 is a sample and hold circuit (abbreviation is S / H; corresponding to input means (conversion means) described in the gist of the invention), 20 is an analog / digital converter (abbreviation is A / H). D: input means (conversion means) described in the gist of the invention), 21 is a color process circuit (corresponding to input means (conversion means) in the gist of the invention), 22
Is a video transfer circuit, 23 is a buffer memory,
Reference numeral 24 denotes a compression / expansion circuit (corresponding to the encoding means and changing means described in the gist of the invention), 25 denotes a flash memory (corresponding to a storage medium described in the gist of the invention), and 26 denotes a CPU (corresponding to the gist of the invention). 27 corresponds to a key input unit,
Reference numeral 28 denotes a digital video encoder, reference numeral 29 denotes a bus, and reference numeral 30 denotes a motion detection circuit (corresponding to the extraction means and the detection means described in the gist of the present invention) which is one of the unique elements of the present embodiment.

The functions of these components are as follows. (A) Photo lens 15: This is for forming an image of a subject on the light receiving surface of the CCD 16, and has a focusing mechanism for an automatic focusing function. Note that a zoom function may be provided, or a retractable type may be provided. (B) CCD 16: a solid-state imaging device that transfers electric charges in an array, converts a subject image captured via the photographic lens 15 into an electric signal, and outputs a frame image signal of about several tens of frames per second. is there.

In general, a CCD has a photoelectric conversion section in which a large number of photoelectric conversion elements are arranged in an array, a charge accumulation section for accumulating output charges of the photoelectric conversion elements, and a charge readout for reading out the charges of the charge accumulation section in a predetermined manner. And each of the photoelectric conversion elements becomes a pixel. For example, in a CCD having one million effective pixels, at least one million cells in the array are arranged. Hereinafter, for convenience of explanation, the number of effective pixels of the illustrated CCD 16 is 1280 × 96.
Set to 0. That is, 1280 in the row direction (lateral direction),
12 pixels composed of 960 pixels in the column direction (vertical direction)
Assume that it has an array structure of 80 columns × 960 rows.

The CCD 16 according to the present embodiment is a color CCD. Generally, since the pixel information of the CCD itself does not have color information, a color filter (a primary color filter using three primary colors of light or a complementary color filter using three primary colors) is mounted on the front surface of a color CCD.

Further, CCDs can be classified into two types according to a charge reading method. The first type is a "jump-out readout method" (also referred to as an interlaced CCD) in which pixels are skipped one by one when reading a signal. The second is a "full-surface readout method" (in both progressive CCDs) in which all pixels are read out in order. Say) type. Although the second type is often used in electronic still cameras, the first type may be used in recent megapixel electronic still cameras exceeding 1 million pixels. Hereinafter, for convenience of explanation, the CCD 16 of the present embodiment will be described.
Is of a second type (full-surface reading method).

(C) Horizontal / vertical driver 17 and timing generator 18: This is a portion for generating a drive signal necessary for reading the CCD 16, and the CCD 16 of the present embodiment is
Since it is assumed that the entire surface is read out, a drive signal capable of transferring (reading out) pixel information in units of rows while designating each column of the CCD 16 one after another.
It generates horizontal and vertical drive signals for serially reading out pixel information in the direction from the upper left to the lower right of the array structure of 80 columns × 960 rows (this direction is similar to the television scanning direction). . (D) Sample and hold circuit 19: Sampling (for example, correlated double sampling) a time-series signal (which is an analog signal at this stage) read from the CCD 16 at a frequency suitable for the resolution of the CCD 16. . Note that automatic gain adjustment (AGC) may be performed after sampling. (E) Analog-to-digital converter 20: Converts a sampled signal into a digital signal.

This section generates a luminance / color difference multiplex signal (hereinafter, referred to as a YUV signal) from the output of the analog-to-digital converter 20. The reason for generating the YUV signal is as follows. Analog-to-digital converter 20
Output is substantially equal to CCD1 except for the difference between analog and digital, and errors in sampling and digital conversion.
6 corresponding to the three primary colors of light (RGB
Data) itself, but this data is large in size, causing inconvenience in terms of limited memory resource utilization and processing time. Therefore, it is necessary to reduce the amount of data to some extent by some method. YUV signal is generally RG
Each element data of the B data (R data, G data, B data) can be represented by three color difference signals GY, RY, BY for the luminance signal Y, and these three color difference signals If the redundancy of is removed, GY need not be transferred,
This can be said to be a kind of data amount reduction signal based on the principle that it can be reproduced by GY = α (RY) −β (BY). Here, α and β are synthesis coefficients.

The YUV signal is converted to a YCbCr signal (Cb
And Cr may be referred to as BY and RY, respectively, but in this specification, they are unified to YUV signals. The signal format of the YUV signal is composed of three fixed-length blocks called “components” each independently including a luminance signal and two color difference signals, and has a length (number of bits) of each component. The ratio is called the component ratio. Although the component ratio of the YUV signal immediately after the conversion is 1: 1: 1, the two components of the color difference signal are shortened, that is, 1: x: x (where x
<1) can also reduce the data amount. This utilizes the fact that human visual characteristics are less sensitive to color difference signals than luminance signals.

(F) Video transfer circuit 22: The video transfer circuit 22 is a color process circuit 2
1. It controls the flow of data flowing between the buffer memory 23, the digital video encoder 28 and the compression / expansion circuit 24.

The "flow" is a convenient expression conceptually capturing the movement of data flowing between the color process circuit 21, the buffer memory 23, the digital video encoder 28, and the compression / decompression circuit 24. Although the words themselves have no special meaning, generally for digital systems, the fast movement of data directly affects its performance, especially for electronic still cameras that deal with a large amount of pixel information (quick movement of data). Since it is one of the design conditions that must be taken into consideration, all or a part of the above-mentioned flow means a data flow utilizing a high-speed data transfer method. That is, all flows are, for example, DMA (direct memory acces
s) The flow by transfer, and the video transfer circuit 22 includes a control unit (DMA controller) and other peripheral parts (for example, FI for adjusting the transfer speed) necessary for the transfer.
FO memory and interface circuit, etc.), and the operation of each of these units enables "quick data transfer" (for example, DMA transfer) between the color process circuit 21, the buffer memory 23, the digital video encoder 28, and the compression / decompression circuit 24. Is what makes it possible.

(G) Buffer memory 23: It is composed of a DRAM which is a kind of rewritable semiconductor memory. Generally, a DRAM is different from a static RAM (SRAM) in that data is dynamically rewritten (refreshed) in order to retain storage contents, but a DRAM is different from a static RAM (SRAM).
Although the writing and reading speeds are inferior to those of AM, it is particularly suitable for an electronic still camera because the unit cost per bit is low and a large-capacity temporary storage can be configured at low cost. However, the present invention is not limited to a DRAM. Any rewritable semiconductor memory may be used.

Here, the storage capacity of the buffer memory 23 must satisfy all of the following conditions. The first condition is that the capacity is such that a sufficient work area (work space) necessary for the work can be secured. The size of the work space is determined by the architecture of the CPU 26, the OS (Operating System), and various application programs executed under the control of the OS. I just need. The second condition is that at least the color process circuit 2
1 is a capacity that can secure a buffer area large enough to store the information of the high-definition image generated in Step 1 (image information of 640 × 480 pixels and a YUV signal having a component ratio of 1: 1: 1). The third condition is that the capacity is such that a buffer area large enough to store a moving image (640 × 480 pixels) can be secured.

(H) Compression / expansion circuit 24: A section for performing JPEG compression and expansion for a normally shot image, and MPEG compression and expansion for a movie shot image (movie image). MPEG is a moving picture coding standard as described at the beginning, and JPEG (joint photographer)
The icexperts group is an international coding standard for color still images (full-color or grayscale still images that do not include binary images or moving images). The compression / expansion circuit 24 should be dedicated hardware in terms of processing speed.
It is also possible to perform this in U26 by software.

The default value (predetermined value) of the N parameter of MPEG is set to an appropriate value (described later) in consideration of an appropriate random access unit. Further, the M parameter is set to an appropriate value (described later) corresponding to a relatively gentle image. These default values are applied when a “motion detection signal MV” described later is inactive (L logic for convenience).

That is, if “MV = L” when compressing a movie image, GO
Construct P and I picture (or P picture)
M-1 B pictures are inserted between the P picture and the P picture. When the “motion detection signal MV” is active (H logic for convenience) and the length of the active period extends for two frames. In the case (see case 3 and case 4 in FIG. 6; details will be described later), the default value is ignored and G
The OP is forcibly closed.

(I) Flash memory 25: rewritable read-only memory (PROM: programmable area)
d only memory) that can be electrically rewritten by erasing the contents of all bits (or blocks). Flash EEPROM (flash electrically era
sablePROM). The flash memory 25 in the present embodiment may be a fixed type that cannot be removed from the camera body, or may be a removable type such as a card type or a package type. (J) CPU 26: centrally controls the operation of the camera by executing a predetermined program. The program is CP
The program for the mode is written in the instruction ROM inside the U26. In the recording mode, the program for the mode is read from the instruction ROM in the reproduction mode.
M is loaded and executed.

(K) Key input section 27: a section for generating operation signals for various key switches provided on the camera body. (L) Digital video encoder 28: converts the digital value display image read from the buffer area of the buffer memory 23 via the video transfer circuit 22 into an analog voltage, and at the same time the timing according to the scanning method of the liquid crystal display 4. Are sequentially output. (M) Bus 29: A data (and address) transfer path shared among the above components. Although not shown in the figure, necessary control lines (control lines) are also provided between the units.

(N) Motion detection circuit 30: The output of the A / D 20, that is, the motion of the subject based on the raw frame image signal before being converted into the YUV signal with the output signal of the imaging system converted into digital data. This is a circuit that detects and outputs a motion detection signal MV that becomes active (H logic) when the motion is large (or severe).

FIG. 3 is a block diagram of the motion detecting circuit 30, in which 31 is a low-pass filter (LPF), 32 is a high-pass filter (HPF), 33 is a gate circuit, 34 is a peak hold circuit, and 35 is a high-luminance determining circuit. , 36 are integrating circuits, and 37 is a motion determining unit. LPF31 and HPF3
In step 2, the high frequency component of the frame image signal is extracted, the gate circuit 33 masks a portion other than the specific range in the screen, and the peak hold circuit 34 holds the peak value PK of the high frequency component in the specific range. At the same time, the high-brightness determination circuit 35 determines a high-brightness portion in the screen, instructs the gate circuit 33 to mask the high-brightness portion, and calculates an integrated brightness value BT of the high-brightness portion by the integration circuit 36.

FIG. 4 is a block diagram of the motion judging section 37.
38 is a register for holding the PK for one frame, 39 is a difference calculator for calculating a difference ΔPK between the current PK and the PK of the previous frame, and 40 is comparing ΔPK with a reference value RPK to obtain “ΔPK ≧ RPK”. Is a comparator that outputs H logic at the time of. 41 is a register for holding the BT for one frame, and 42 is a difference ΔB between the current BT and the BT of the previous frame.
A difference calculator 43 for calculating T is a comparator for comparing .DELTA.BT with the reference value RBT and outputting an H logic when ".DELTA.BT.gtoreq.RBT". Further, an OR gate 44 outputs an H logic when at least one of the two comparators 40 and 43 outputs an H logic.
This is a register held during the frame, and the value held in the register 45 is output as the motion detection signal MV. Note that F
Is a frame signal synchronized with the frame period of the CCD 16, and the three registers 38, 41 and 45 update the held values in response to the frame signal F.

FIG. 5 shows the truth values of the motion judging section 37.
When “ΔPK ≧ RPK” or “ΔBT ≧ RBT”, MV =
H, that is, the motion detection signal MV is “active”.

Here, ΔPK is the difference between frames of the peak value PK of the high frequency component in a specific range in the screen, and the peak value PK is the edge component of the subject image (corresponding to the characteristic portion described in the gist of the invention). ), And when ΔPK corresponding to the difference between the frames is large (ΔPK ≧ RP
K) shows that the motion of the subject is large and the correlation between the two frames is sparse.

.DELTA.BT is the difference between frames of the integrated value BT of the luminance component (especially the high luminance component) in the screen, and the integrated value BT is the brightness of the entire screen including the subject image (the feature described in the summary of the invention). When the value of ΔBT corresponding to the difference between frames is large (ΔBT ≧ RBT)
Also, it can be seen that the movement of the subject is large and the correlation between the two frames is sparse. Therefore, in any case, since a remarkable motion of the subject is detected, the motion detection signal MV is held in the active state during the corresponding frame.

Next, the operation will be described. <Recording Mode of Normal Photographing> In this mode, the CCD 16 disposed behind the photographic lens 15 is driven by a signal from the driver 17, and the image collected by the photographic lens 15 is photoelectrically converted at regular intervals to generate one image. Minute video signals are output. Then, the video signal is sampled by the sampling and holding circuit 19 and converted into a digital signal by the analog-to-digital converter 20, and then a YUV signal is generated by the color process circuit 21. This YUV signal is transferred to the buffer area of the buffer memory 23 via the video transfer circuit 22, and after the transfer to the buffer area is completed, is sent to the liquid crystal display 4 via the video transfer circuit 22 and the digital video encoder 28. It is displayed as a through image.

When the direction of the camera is changed in this state, the composition of the through image displayed on the liquid crystal display 4 changes.
When the desired composition is obtained, set the exposure and focus by “half-pressing” the shutter key 5 and then “press fully”.
Then, the YUV signal stored in the buffer area of the buffer memory 23 is fixed at the current YUV signal,
Further, the through image displayed on the liquid crystal display 4 is also fixed at the image at the same time.

Then, the YUV signal stored in the buffer area of the buffer memory 23 at that time is sent to the compression / decompression circuit 24 via the video transfer circuit 22, and the YUV signal is transmitted to each of the Y, Cb and Cr components. After being JPEG-encoded in units called a basic block of × 8 pixels, it is written into the flash memory 25 and recorded as a captured image for one image.

<Normal Shooting Reproduction Mode> In this mode, the path from the CCD 16 to the buffer memory 23 is stopped, and the latest captured image is read from the flash memory 25 and decompressed by the compression / decompression circuit 24. After that, it is sent to the buffer area of the buffer memory 23 via the video transfer circuit 22. Then, the data in the buffer area is sent to the liquid crystal display 4 via the video transfer circuit 22 and the digital video encoder 28, and is displayed as a reproduced image.

<Recording Mode of Movie Shooting> The basic operation is the same as that of the normal shooting, except that the YUV from the color process circuit 21 is pressed for a preset movie shooting time after the shutter key 5 is fully pressed. Compress signal
MPEG compression by a decompression circuit 24 and flash memory 25
In that it is recorded in For example, when the movie shooting time is set to 9 seconds, the through image for 9 seconds is continuously MPEG-compressed and recorded in the flash memory 25 after the shutter key 5 is fully pressed.

<Playback Mode for Movie Shooting> In this playback mode, two modes can be selected. One of the modes is a mode of reproducing a moving image recorded in the flash memory 25 and displaying the moving image on the liquid crystal display 4, which is a so-called moving image reproduction mode. The other is a mode of reproducing a moving image. In this mode, m screens are selected from among them and the multi-screen is displayed on the liquid crystal display 4. The number m of screen selections is arbitrary, but if m = 9 for convenience,
Nine still images can be displayed on the liquid crystal display 4. For example, nine still images every second can be displayed from a nine-second moving image of a golf swing taken to correct the form. .

<MP in Movie Recording Mode
Example of EG Compression> FIG. 6 is a diagram showing a GOP structure in a recording mode of movie shooting. In this figure, F
n is an input image signal (more precisely, a YUV signal taken into the buffer memory 23 from the color process circuit 21 via the video transfer circuit 22), and tn is an index of time corresponding to the frame number of Fn. Note that n is CC
This is the order of frames corresponding to the drive cycle of D16.

The figure shows the GOP structure of three patterns of a, b and c, and these patterns are alternatively selected based on the logical state (case 0 to case 4) of the motion detection signal MV. Selected. That is, in case 0 where the MV is always at the L logic or in cases 1 and 2 where the MV is at the H logic (active state) for one frame, the GO of pattern a
In case 3 or case 4 in which the P structure is selected and the MV goes to the H logic (active state) for two consecutive frames,
The GOP structure of the pattern b or the pattern c is selected.

The GOP structure of pattern a applies default values to the N and M parameters of the GOP.
Here, the default value is not particularly limited, but N = 9,
M = 3. One GOP of the pattern a is, for example,
F0, F-2, F-1, F3, F1, F2, F6, F4
, F5, and 9 frames (N frames), and the encoding method (encoding picture type) is M =
3, I, B, B, P,
B, B, P, B, and B are arranged. Note that a forward reference frame represents a reference frame in P picture coding and a forward reference frame in B picture coding, and a backward reference frame represents a backward reference frame in B picture coding.

Here, the MV in case 0 is always L logic (inactive state), and the motion detection circuit 30 does not detect a remarkable motion of the subject (in other words, the motion detection circuit 30 Motion is detected), the default N and M parameters are applied to case 0.
This is because the prediction error signal (motion compensation prediction error signal) when the motion of the subject is gentle is small, and the compression efficiency can be improved.

On the other hand, focusing on the MVs of Case 1 and Case 2, it is shown that these are in H logic (active state) for one frame, and that the motion detection circuit 30 detects a remarkable motion of the subject. Assuming that the frame is Fi, the frame to be detected is Fi-
2 and Fi-1 (only F0 and F1 in case 1 because Fi = F2, and F1 and F2 in case 2 because Fi = F3), and the correlation between these two frames is sparse (case 1). Then F0 ≠ F1, and in case 1, F1 ≠
F2), other reference images (F3 in case 1 and F0 in case 2) necessary for encoding the B picture
And F3) and Fi (in case 0, F1 = F
3. In case 1, F0 = F1 and F2 = F3) are recognized, so there is no problem assuming that the motion of the subject is gentle, and the default N and M parameters for case 1 and case 2 are set similarly to case 0. Apply.

On the other hand, the MVs of Case 3 and Case 4 keep the H logic (active state) for two frames.
In each case, the correlation between the two reference pictures required for encoding the B picture is clearly sparse. That is, since F0 ≠ F1 ≠ F2 in case 3 and F1 ≠ F2 ≠ F3 in case 4, the motion of the subject is sharp in any case and the prediction error signal (error signal of motion compensation prediction) is large. Since the insertion of a B-picture rather impairs the compression efficiency, in such cases 3 and 4, when the coding of the corresponding frame (F-1 in case 3 and F1 in case 4) is completed, N And forcibly close the GOP ignoring the default values of M and M, and start generating a new GOP.

According to this, the head of a new GOP always becomes an I picture, and the picture quality of the I picture is equivalent to, for example, JPEG of the still picture coding standard. It is possible to improve the quality of recorded images of movie shooting.

When a new GOP is generated, the M parameter may be set to “1”. That is, a GOP structure including only an I picture and a P picture that does not include a B picture may be used, or the M parameter may be returned to a default value after the motion of the subject becomes gentle.

As described above, according to the present embodiment, when the motion detection circuit 30 detects a remarkable motion of the subject, N
The GOP is forcibly closed by ignoring the default values of the parameters and the M parameter, so that the IOP with good image quality is
A special effect is obtained in that the number of pictures can be increased to improve the quality of a recorded image of a movie.

The LPF constituting the motion detection circuit 30
31, an HPF 32, a gate circuit 33, a peak hold circuit 34, a high luminance determination circuit 35, an integration circuit 36, etc.
Since it is often mounted in the AF section of an electronic still camera and can share a circuit, there is also an advantage that the above-described effect can be obtained at low cost without requiring a major modification or the like.

Further, by using the output signal (motion detection signal MV) of the motion detection circuit 30, the amount of calculation for MPEG motion compensation can be reduced. That is, MPEG
In this example, a part of an image of a subject is cut out, shifted by the amount of motion, and predicted as a next frame image (motion-compensated inter-frame prediction). Since it is possible to detect whether or not has a correlation only before or after,
In the motion compensation processing of PEG, by applying motion compensation inter-frame prediction only to reference frames having a correlation, a large amount of computation processing such as computation of a motion vector can be simplified, and power saving can be achieved. The advantage is obtained.

[0063]

According to the first aspect of the present invention, input means for inputting a frame image signal of a predetermined period, encoding means for encoding the frame image signal by an MPEG standard encoding method, Recording means for recording a subsequent frame image signal on a storage medium, extraction means for extracting a characteristic portion of the frame image signal, and detection means for detecting the movement of the subject by correlating the characteristic portion between adjacent frames, Changing means for changing the coding method of each frame in the coding means based on the detection result of the detection means, so that the motion of the subject is detected using a simple motion detection circuit, The coding method of each frame can be optimized according to the degree of movement, and both the reduction of the amount of information when the movement is gentle and the improvement of the image quality when the movement is severe can be achieved.
According to the second aspect of the present invention, in the moving image recording apparatus according to the first aspect, the changing unit is configured to encode the frame by the encoding unit when the detecting unit detects a remarkable movement of the subject. Is changed to an I-picture, so that the substantial number of I-pictures having good image quality can be increased to improve the quality of a recorded image. According to the third aspect of the present invention, in the moving image recording apparatus according to the first aspect, the changing unit is configured to encode the frame by the encoding unit when the detection unit detects a remarkable movement of the subject. Is operated to start the closing of the current GOP and the generation of a new GOP. Therefore, since the beginning of the GOP is always an I picture, the real number of high quality I pictures is increased to improve the quality of a recorded image. be able to. According to the invention as set forth in claim 4, conversion means for converting an image of a subject into a frame image signal of a predetermined cycle, coding means for coding the frame image signal by an MPEG standard coding method,
Recording means for recording an encoded frame image signal on a storage medium; extracting means for extracting a characteristic portion of the frame image signal; and detecting means for detecting the movement of a subject by correlating the characteristic portion between adjacent frames And changing means for changing the coding method of each frame in the coding means based on the detection result of the detection means, wherein the extraction means is used in combination with an automatic focusing circuit of the electronic camera. Therefore, it is possible to reduce the amount of information when the motion is gentle and improve the image quality when the motion is severe by optimizing the encoding method of each frame without increasing the cost. According to the fifth aspect of the present invention, in the moving image recording apparatus according to the first or fourth aspect, the detection result of the detecting means is set to M
Applying to PEG motion compensation processing and performing motion compensation inter-frame prediction only for reference frames having correlation, it is possible to simplify a large amount of computation processing such as motion vector computation and to save power. be able to.

[Brief description of the drawings]

FIG. 1 is an external view of an electronic still camera.

FIG. 2 is a block diagram of an electronic still camera.

FIG. 3 is a configuration diagram of a motion detection circuit.

FIG. 4 is a configuration diagram of a motion determination unit.

FIG. 5 is a truth table of the motion determination unit.

FIG. 6 is a GOP structure diagram in the present embodiment.

FIG. 7 is a syntax diagram of MPEG.

FIG. 8 is a diagram illustrating an example of a GOP structure.

FIG. 9 is a state diagram showing a change in the order of screens of an original image.

[Explanation of symbols]

16 CCD (input means (conversion means)) 19 sample hold circuit (input means (conversion means)) 20 analog-digital converter (input means (conversion means)) 21 color process circuit (input means (conversion means)) 24 compression / Decompression circuit (encoding means, changing means) 25 Flash memory (storage medium) 26 CPU (recording means) 30 Motion detection circuit (extraction means, detection means)

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5C022 AA13 AB02 AB28 AB68 AC42 AC69 5C053 FA11 GA11 GB22 GB26 GB28 GB30 GB32 GB37 KA04 KA21 KA24 LA01 LA06 5C059 KK08 KK19 KK47 LA09 LB18 LC01 MA00 MA05 MA14 MA23 MA47 MB04 MC35 ME01 NN36 NN43 PP01 PP05 PP16 PP26 SS14 SS15 TA03 TA66 TA74 TB03 TC13 TC41 TC43 TD05 TD18 UA02 UA30 UA33

Claims (5)

[Claims] An input unit for inputting a frame image signal having a predetermined period; an encoding unit for encoding the frame image signal according to an MPEG standard encoding method; and recording the encoded frame image signal on a storage medium. Recording means, extracting means for extracting a characteristic portion of the frame image signal, detecting means for correlating the characteristic portion between adjacent frames to detect the movement of the subject, and A moving image recording apparatus, comprising: changing means for changing the coding method of each frame in the coding means. 2. The apparatus according to claim 1, wherein said changing means changes a frame coding method of said coding means to an I picture when said detecting means detects a remarkable movement of the subject. Moving image recording device. 3. The changing means starts closing a current GOP and generating a new GOP by operating a frame coding method in the coding means when a remarkable movement of a subject is detected by the detection means. 2. The moving image recording apparatus according to claim 1, wherein the moving image is recorded. 4. A converting means for converting an image of a subject into a frame image signal having a predetermined period, an encoding means for encoding the frame image signal by an MPEG standard encoding method, and a frame image signal after encoding. Recording means for recording on a storage medium; extracting means for extracting a characteristic portion of the frame image signal; detecting means for detecting a movement of a subject by correlating the characteristic portion between adjacent frames; and a detection result of the detecting means A changing means for changing the coding scheme of each frame in the coding means based on the above, wherein the extracting means is used in combination with an automatic focusing circuit of an electronic camera. apparatus. 5. The method according to claim 1, wherein the detection result of the detection unit is applied to an MPEG motion compensation process, and the motion compensation inter-frame prediction is performed only on a reference frame having a correlation. The moving image recording apparatus according to claim 4.