JP2002135721A - Monitoring device for recording video signals - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a monitoring device for recording video signals in which image data are compressed in a temporary storage with less memory capacity, and the varied region of the data, after highly efficiently compressed, is stored in a recording medium. SOLUTION: The image data from two or more shooting means, CA1 to CAn, is stored in a buffer 3 which deals with at least one image data capacity as a unit. After the image data is compared with those coming before or after to detect the varied region between them, it is compressed by a highly efficient encoding method to be stored in a recording medium 14.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video signal recording monitoring apparatus for recording video signals captured for a long time by an image capturing means such as a monitoring camera such as a closed circuit television (CCTV) system.

[0002]

2. Description of the Related Art In recent years, violent criminal cases have occurred frequently, and a video signal recording monitoring device capable of recording for a long time with high image quality has been desired for monitoring use. For such a device, a long-time recording video tape recorder (VT)
R), time-lapse VTR, and the like. However, since magnetic tape is used for the recording medium, and the magnetic transducer, which is the heart of the recording medium, is always in contact with the tape and the head, recording is necessary due to wear of the tape and adhesion of dust to the head. There was a problem from the aspect of operation that it was not recorded at that time and the reliability. In many places, the use environment is poor, and the reliability often deteriorates. For this reason, a video signal recording device using a hard disk has recently been proposed. However, since the recording time is shorter than that of a tape, a recording time that is satisfactory for a long-time recording cannot be obtained. .

A hard disk has a smaller capacity than a tape and is not suitable for long-time recording. To cope with this, video signal recording apparatuses using a hard disk have responded to long-time recording by extending the recording interval time of intermittent recording or reducing the data amount by reducing the image quality. However, if the recording interval time is increased in the intermittent recording, an important scene is missed, and the image quality is deteriorated because the data amount is reduced to be recorded, so that only a video which is hard to be distinguished for monitoring use can be recorded.

[0004]

In an apparatus for monitoring video signal recording for a long time, a method for recording data for a long time more efficiently by using a digital signal compression means which is a highly efficient encoding method is proposed. JP using intra-frame correlation
EG (Joint Photographic Expe
In the compression of the rts Group (rts Group) method, VHS (Vid
eo HomeSystem) average image quality is about 15 compression ratio
The limit is about one in one. If recording is performed within a range that is practically possible in terms of image quality, a video signal recording monitoring device using a conventional hard disk is about 1 hour for continuous recording, which is far less than a tape in terms of time and cost.

As a general monitoring method in the security field, a plurality of imaging signals are temporally switched for each frame, and intermittent recording is performed by a video signal recording monitoring device using a time-lapse VTR or a hard disk. Is going. Further, in many cases, the camera of the imaging means performs auto-panning by a camera platform or the like, or a zoom camera or the like is used. In a time-lapse recording mode using a video signal recording monitoring device, a recording frame is used. The intermittent time interval is long, and an image input for each frame often has no correlation between the previous frame and the next frame. This is particularly noticeable during panning.

Therefore, MPEG (Moving Pic)
It is practically impossible to increase the compression efficiency by using an inter-frame correlation such as a method of “Two Experts Group”. Because, during intermittent recording, the correlation between frames is very low, and it is difficult to obtain correlation. If this is to be performed further, it is necessary to compare the entire screen, determine the amount of movement and the direction of movement, and calculate the movement vector, which requires an enormous amount of time and circuitry. In addition, a function can be realized by frequently using an expensive dedicated integrated circuit and a large-capacity semiconductor memory, but it is easily imagined that the cost is not appropriate.

On the other hand, to enable long-time recording, it is necessary to increase the compression efficiency. That is, it is necessary to increase the efficiency of compression for image quality. The compression efficiency is affected by the S / N ratio, which is the signal-to-noise ratio, and the frequency characteristics. Of course, the pattern on the screen is fine,
If it is sharp, the compression efficiency will be poor. It is known that, even in the S / N ratio, the compression efficiency decreases when the high-frequency noise component is large. Also, time-lapse VT using tape
In a video signal recording monitoring device using a conventional hard disk or a conventional hard disk, the search takes a long time in proportion to the long recording time, and the search time is greatly reduced by increasing the capacity of a recording medium such as a hard disk. The problem was increasing.

The present invention solves the above-mentioned problems, and
The problem to be solved by the invention is to be able to reliably record important moments in the image signal captured by the imaging means,
An object of the present invention is to provide a video signal recording monitoring device which enables high-quality recording for a long time and which is maintenance-free and highly reliable. More specifically, the present invention utilizes a conventional high-efficiency encoding method to more efficiently record data for a long time and to search for required image data at high speed.

[0009]

According to a first aspect of the present invention, each of imaging signals obtained by obtaining a plurality of subject images via a plurality of imaging means is digitized, and the digitized image data is temporarily stored. A video signal recording monitoring device adapted to be recorded on a recording medium after being stored in a means, wherein a capacity of at least one image data or more is defined as one unit, and the total capacity is equal to the number of inputs of a plurality of imaging means plus one or more. A temporary storage unit having a capacity, a comparison unit that reads out image data before and after the image pickup signal of the same imaging unit written in the temporary storage unit from the temporary storage unit and compares the image data, and an image data from the comparison output of the comparison unit Detecting means for detecting the change area of the image data, and compression means for compressing the change area of the image data by a high-efficiency encoding method. The image data compressed by the compression means and the detection means Is obtained by a video signal recording monitoring apparatus characterized by comprising recording the detected information issued to the recording medium.

According to the first aspect of the present invention, it is possible to perform software processing of the high-efficiency encoding method with the high performance of the microcomputer, and to increase the capacity of the semiconductor memory, Since it is available at a low price, an image memory is prepared, for example, in units of one field or one frame in which the capacity of at least one image data is one unit. On the other hand, the image pickup signal is switched in synchronization with the vertical synchronizing signal, and image data picked up by the same image pick-up means is output in units of one frame or one field with a fixed cycle, so that image data having a strong correlation can be obtained. . Therefore, the number of inputs of a plurality of image pickup means plus one or more image memories (an image memory in which the capacity of one image data or more is one unit)
By preparing the numbers, it is possible to easily process the inter-relationship between the multiple frames of the image pickup signal by the same image pickup means, and obtain a video signal recording / monitoring device that can achieve high speed.

According to a second aspect of the present invention, an image pickup signal obtained through at least one image pickup means is digitized, and the digitized image data is stored in a temporary storage means and then recorded on a recording medium. A monitoring device for recording a video signal, comprising: a storage unit having at least one image data or more as one unit, and having a total capacity of at least one or more imaging means inputs plus one or more capacities. Means for reading out the image data before and after written in the temporary storage means of one unit from the temporary storage means for comparison, and moving amount extraction for extracting the moving amount of the moving object of the image data from the comparison output of the comparing means Means, and threshold comparing means for comparing the magnitude of the motion vector amount calculated by the moving amount extracting means with a preset threshold value, wherein the output level of the threshold comparing means is It is obtained by a video signal recording monitoring apparatus characterized by starting the recording on the recording means when exceeding a constant level.

According to the second aspect of the present invention, at least one of the image pickup signals of the image pickup means is switched in the time axis direction, for example, frame by frame, and intermittent recording is performed in the time lapse recording mode. In such a case, for example, there is no correlation between frames of image data taken for each of a plurality of imaging units, for example, for each frame, but recording is performed on a recording medium only when there is a large change in an image in image data of the same imaging unit. Since the process is started, a useful video signal recording monitoring device can be obtained even with image data of an imaging signal having no correlation between a plurality of imaging units.

According to the third aspect, the present invention provides at least one aspect.
An image pickup signal obtained through one or more image pickup means is digitized, a video signal recording monitoring device adapted to record the digitized image data on a recording medium after storing the digitized image data in a temporary storage means, at least The capacity of one image data or more is defined as one unit, and as a total capacity, the number of inputs of at least one or more imaging means plus a temporary storage means having a capacity of one or more, and a temporary storage unit before and after being written to the temporary storage means of one unit Comparing means for reading out the image data from the temporary storage means for comparison; moving amount extracting means for extracting the moving amount of the moving object of the image data from the comparison output of the comparing means; detecting means for detecting a change area of the image data; A video signal recording monitor characterized in that each output from a moving amount extracting means for extracting a moving amount of an object is flagged and recorded on a recording medium. It is obtained by the location.

According to the third aspect of the present invention, a flag is attached to the position of a change in motion in the image data of the image pickup signal of the image pickup means, and the flag is recorded on the recording medium. It is possible to obtain a video signal recording monitoring device capable of easily searching a recording medium for a change in image movement such as "ta", "object lost", "someone has entered", and the like.

According to a fourth aspect of the present invention, there is provided a video signal recording apparatus according to the first or second aspect, further comprising noise reduction means for reducing high frequency component noise of an image pickup signal of the image pickup means. Monitoring device.

According to the fourth aspect of the present invention, by providing a nonlinear processing circuit and a high-pass filter in the signal processing unit, it is possible to balance the compression ratio without lowering the frequency characteristics and the resolution. It is possible to obtain a video signal recording / monitoring device capable of changing noise reduction and frequency characteristics and having high compression efficiency.

According to a fifth aspect of the present invention, the compression means of the present invention has a Hadamard transform means as a high-frequency component reducing means of the wavelet transform method. It is a monitoring device.

According to the fifth aspect of the present invention, noise reduction in the signal processing circuit is unnecessary, and it is possible to substitute for noise reduction by using a part of the tree filter of the Hadamard transform means as the compression means. A video signal recording monitoring device can be obtained.

According to a sixth aspect of the present invention, there is provided a temporary storage unit for digitizing an image pickup signal obtained by imaging an object by an image pickup unit by an analog-to-digital conversion unit and recording the digitized image data, a JPEG conversion unit, It comprises a wavelet transform means and a compression rate variable means for varying the compression rate of both transform means, and performs JPEG conversion according to the band condition of the transmission path when transmitting image data or the compression rate when compressing image data. A video signal recording monitoring apparatus characterized in that the method is switched to a method or a wavelet transform method and is recorded on a recording medium to increase the compression efficiency.

The present invention according to claim 7 further comprises means for varying the sampling clock of the vertical and horizontal video sections when converting digital data by the analog-to-digital conversion means. , JPEG conversion method depending on the compression ratio, or
7. A video signal recording monitoring apparatus according to claim 6, wherein the compression efficiency is increased by switching the wavelet transform method.

According to the inventions according to the sixth and seventh aspects, when recording for a long time or when the bandwidth of the transmission path is narrow, an image with a low compression ratio (high image quality) is recorded by using the wavelet transform method. DCT when the transmission band is wide
By using (discrete cosine transform), efficient compression can be performed.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing one embodiment of a video signal recording monitoring apparatus according to the present invention.
FIG. 3 is a block diagram showing the video switcher signal processing unit of FIG. 1, FIG. 3 is a block diagram of the synchronization signal generation, genlock circuit and system clock generation circuit of FIG. 1, FIG. 4 is a flowchart for explaining a motion vector, and FIG. 6A to 6C are explanatory diagrams of the operation in FIG. 2, FIG. 7 is a block diagram showing the processing of the JPEG encoder unit, FIG. 8 is a block diagram showing the processing of the JPEG decoder unit, and FIG. 10 is a block diagram showing the processing of the wavelet encoder, FIG. 10 is a block diagram showing the processing of the wavelet decoder, FIG. 11 is a block diagram showing the details of the JPEG or wavelet encoder of FIG. 1, and FIG. FIG. 13 is a block diagram showing details of the JPEG or wavelet decoder section of FIG. Block diagram illustrating a process for,
14 is a block diagram showing a subband coding filter configuration, FIG. 15 is a screen diagram for explaining Hadamard transform, FIG. 16 is a block diagram showing an operation of the frame memory of FIG. 2, and FIG. 17 is a noise reduction of FIG. FIG. 18 is an explanatory diagram of the operation of the non-linear amplification processing circuit of FIG. 17, FIG. 19 is a block diagram when the non-linear amplification processing circuit of FIG. 17 is digitized, and FIGS. 2
FIG. 1 is an explanatory diagram for shortening the search time.

FIG. 1 is an overall block diagram of a video signal recording monitoring apparatus according to the present invention. In the present invention, a carrier recording apparatus and a carrier capable of recording or reproducing at least one video signal for a long time. The description will proceed as a video signal recording monitoring device including a reproducing device and a carrier recording / reproducing device.

In FIG. 1, the input of the video switcher / signal processing unit 1 is composed of at least one of input terminals T ₁ , T ₂ , T ₃ ‥‥ T _n , for example, n series, and one or more input terminals are provided. For example, a composite video signal, which is a video input 21a to 21n, is input from a video camera serving as an imaging unit.

There are cases where the above-mentioned plurality of video cameras capture different subject images and cases where a plurality of video cameras capture the same subject image, and the output from the video camera is a composite image such as a moving image or a still image. It is output as a video signal. As shown in FIG. 2, a video switcher / signal processing unit (hereinafter, referred to as a signal processing unit) 1 includes a video switcher 1a, a noise reduction circuit 1b, a Y / C separation circuit (a luminance signal and a chrominance signal separation circuit) 1c, and an NTSC ( International Television Commission), PAL (European TV broadcasting method),
SECAM (Sequential Couleura)
Memorie color television
system) and a color decoder 1d of the method. The output from the color decoder 1d of the signal processing unit 1 is output to an analog-to-digital converter (hereinafter referred to as A / D) 2.

A / D2 has three A / D2s as shown in FIG.
a, 2b and 2c. Video switcher 1
In a, _one of the video inputs which are the imaging outputs of the video cameras CA ₁ , CA ₂ and ‥‥ CA _n is selected, and A / D 2 a,
2b and 2c.

Video input 21 of composite video signal
When a is selected by the video switcher 1a, the Y / C separation circuit 1c separates the signal into a Y signal (luminance signal) and a C signal (color signal). Decoded into a BY signal,
The signal, the RY signal, and the BY signal are A / D 2a, 2
b, 2c, and is converted into digital data by a sampling clock synchronized with the input signal.

This sampling clock is supplied from a synchronizing signal generating genlock circuit and a system clock generating circuit 19 described later.

Y signal from A / Ds 2a, 2b, 2c, R signal
The -Y signal and the BY signal are supplied to, for example, a frame or field memory 3 capable of storing data for one image.

The memory 3 has a video camera C as shown in FIG.
A ₁ , CA ₂ , ΔCA _n , or twice the number of inputs of the video switcher 1 a, or, as will be described later, one image data larger by one than the number of cameras or the number of inputs of the switcher, for example, 1 The memories 23a to 23n for frames are prepared, and the Y signal, the BY signal, and the RY signal are respectively processed by the signal processing circuit 22.
(Only the Y signal is shown in FIG. 2).
3a to 23n.

[0031] The memory 24a~24n of one frame or one field before is input for example from the same video camera CA ₁ to CA _n, the video input 21a~21n captured i.e. when one image data component before incorporation Each of them is prepared for storage, and becomes a memory for one image which is twice as large as one camera.

[0032] In the above configuration, the read output of the memory 24a of the previous frame of the image data of the image signal is stored from the same video camera CA _1, the video camera C
A read output of the memory 23a in which the image data of the current frame from A ₁ is stored is compared by the comparator 6c.

The comparator 6c also compares the image data of the current frame with the image data of one image before one frame in the memories 24b to 24n and the memories 23b to 23n corresponding to the plurality of video cameras CA _{1 to} CA _n. . This comparison is 8x
It is compared every 8 or 16 × 16 matrix image.

The comparator 6c includes buses 10a, 1 of a microcomputer (hereinafter referred to as CPU) 6a for controlling the system.
0b to perform comparison control.

The comparison data of the comparator 6c is compressed by the JPEG or wavelet encoder 4 for highly efficient encoding.

A bus 10a of the CPU 6a has a ROM (Read Only Memory) 7 for a normal system and a DRAM (Dynamic Random) for a work area.
mA Access Memory 8 or DMA (Dir
ct Memory Access) Controller 6b
And a time generation time code unit 9 for generating time information.

The memory 3, the comparator 6c, the JPEG or wavelet encoder 4, and the JPEG or a wavelet decoder 5, which will be described later, improve the performance of the CPU 6a, extend the bus width, fetch peripheral circuits, and the like.
The processing can be performed within the JP. However, in the case of performing other arithmetic processing or reproducing other images at the same time as recording, the JP
The EG or wavelet encoder 4 and the JPEG or wavelet decoder 5 may be connected on the bus line. Further, for example, a frame memory 3 for one image
Since the data transfer can be performed at high speed by the DMA controller 6b, it may be constituted by the DRAM 8. In this case, it can be directly connected to the A / D converter 2 by the bus line 10b.

The image data compressed by the JPEG or wavelet encoder 4 is transferred to the bus 10a under the control of the CPU 6a, and then transferred to the interface (I /
F) Recording is performed on the recording medium 14 via 13a. As the recording medium 14, a magnetic disk device (HDD) or a magneto-optical disk device capable of recording / reproducing with a specific contact head is used. The memory 3 for temporarily storing the compressed image data to be recorded may use the recording medium 14 or the DRAM 8 together. The bus 10a also has an external control I / F.
An operation unit (panel) 11b, a display unit 11a, and the like are connected to the external control I / F 12.

The external connection I / F 12 has input / output terminals Ta,
A plurality of inputs / outputs 25a, 25b,
25c, ‥‥ 25n are input, I / F13a is connected to an external data interface 27 via the external data interface terminal T _IF1, LAN (Local A
rea Network) I / F13b is connected to an external LAN interface 26 via the external LAN interface terminal T _IF2.

Next, a reproduction system for reading and reproducing image data recorded on the recording medium 14 will be described. The image data stored in the storage area of the recording medium 14 is transmitted to the CPU
6a, the compressed data corresponding to the input channel of each camera is read. Recording medium 14
Is transferred to the JPEG or wavelet decoder 5 via the bus 10a via the I / F 13a. Basically, the texture image of the specified channel is expanded, and the image of the object is expanded on the expanded image, and the expanded image is reproduced. An instruction to specify a channel is sent to the CPU 6a manually by the operation unit 11b.

JPEG or wavelet decoder 5
The extended image data temporarily decoded in the memory 3 is digital data reproduced by switching the movable contact piece a of the recording / reproduction switching switch 15 from the recording-side fixed terminal b to the reproduction-side fixed terminal c. (Y, RY, BY)
Is a digital-analog converter (hereinafter referred to as D / A) 1
6 is transferred.

The signal is converted into an analog signal by the D / A 16, and NTSC, PAL, SECAM is generated by the video signal processing / output unit 17, the synchronous signal generating genlock circuit and the system clock generator 19 connected to the bus 10 a.
Is converted to a signal or the like, and as a video output to the outside, Y
/ C signal mixed composite video signal 28
a and 28b are output via output terminals T _O1 and T _O2 .
Also, the black burst (C.SYNC) output 29 for external synchronization signal in synchronization with the video output signal is output via the black burst output terminal T _F1 from the video synchronization output control unit 20.

The D / A 16 in the reproduction system shown in FIG.
Although three D / As are provided for the Y signal and for the RY / BY signal, the details are omitted.

Next, details of the synchronization signal generating genlock circuit and the system clock generating circuit 19 will be described with reference to the block diagram of FIG.

The signal processing unit 1 supplies a periodic signal generating genlock circuit and a synchronizing separation circuit 19a in the system clock generating circuit 19 via a switch 18, and a 60 Hz vertical synchronizing signal from the synchronizing separation circuit 19a is supplied to a phase comparator 19b. The count value of the counter 19d is supplied to the phase comparator 19b and compared with the vertical synchronizing signal. A voltage controlled oscillator (VCO) 19c oscillates in synchronization with the comparison output of the phase comparator 19b, and the frequency is divided by a factor of 1 / N (N is an integer) by a counter 19d, and a count output synchronized with the vertical synchronization signal is obtained. Output from the counter 19d. That is, 1 / 450,000 for 27 MHz and 1 for scare pixels.
/ 4099091 is output from the counter 19d and fed back to the phase comparator 19b for comparison. Therefore, the switch 19e which can obtain an arbitrary frequency by changing the counter 19d uses a sampling clock of 1 or 1/2. Data loader 19g is C
The frequency to the A / D 2 (2a, 2b, 2c) is varied by obtaining the frequency division ratio from the PU 6a and loading it into the counter. 19f is a synchronization signal generation unit.

The operation of the monitoring apparatus for recording a video signal having the above configuration will be described with reference to FIGS. First, an image processing operation for recording an image signal captured by _one video camera CA1 of the image capturing means on the recording medium 14 will be described.

In FIG. 1, video camera inputs 21a to 21a
21n is supplied to the input terminal T ₁ through T _n. The input terminal T ₁ multiple composite video signal the signal processing unit 1 is in through T _n image pickup signal of the video camera 21a~21n
Supplied to Since the inside of the signal processing unit 1 is configured as shown in FIG. 2, one of a plurality of composite video signals is selected by the video switcher 1a.

The composite video signal selected by the video switcher 1a is subjected to noise suppression by a noise reduction 1b, and is separated into a Y signal and a C signal by a Y / C separation circuit 1c. The C signal separated by the Y / C separation circuit 1c is decoded by a color decoder such as PAL, SECAM, or NTSC into an RY color difference signal and a BY color difference signal (where R is a red signal and B is a blue signal). You. Y signal and R
-Y color difference signal and BY color difference signal are three A / Ds 2a, 2
The Y signal and each color difference signal are converted into digital data by a sampling clock from a synchronization signal generating genlock circuit and a system clock generating circuit (hereinafter referred to as a signal generating circuit) 19 in these A / Ds. Is done.

The Y signal and R-signal for one image from one field or one frame from three A / Ds 2a, 2b, 2c are used.
The digital data consisting of the Y color difference signal and the BY color difference signal are converted into respective input signal processing circuits 22 (only 22 is shown in FIG. 2).
In treated and stored, for example, an image signal captured by the video camera CA ₁ in the frame memory 24a constituting the temporary storage means as one image data. In this case, for a plurality of video cameras CA _{2 to} CA _n other than the video camera CA ₁ , for example, a frame memory capable of storing digital data for one image of an image signal captured by each of the video cameras CA _{2 to} CA _n 24b to 24n are connected to video cameras CA _{2 to} C
And it provided in the number corresponding to A _n for storing digital image data of the one image a (hereinafter referred to as image data) is switched by the video switcher 1a in each memory 24B～24n.

Image data for one image of each of the video cameras CA _{1 to} CA _n input at the next timing is the same as the number of video cameras CA _{1 to} CA _n or the video switcher 1.
A number of memories 23a to 23n corresponding to the number of inputs of a are prepared. In this case, for example, the capacity of one memory 23a allows selection of a memory capable of storing digital data for at least one image of one frame or one field.

As described above, each of the video cameras CA _{1 to} CA _n
, For example, each memory 24 stored one frame before
The image data of a to 24n and the image data of each of the memories 23a to 23n stored in the current frame are compared by the comparator 6c.

For example, the image data stored in the memory 24a and the image data stored in the memory 23a are compared every 8 × 8 or 16 × 16 matrix image.

The image data compared by the comparator 6c is detected by the CPU 6a via the buses 10b and 10a by the CPU 6a. The area where the detected image has a change is stored in the JPEG or wavelet encoder 4.
Then, the storage address information of the image data is recorded on the recording medium 14 (HD) together with flags such as frame number, time information and movement vector information.
In this case, the flag is recorded in the search information recording units 52a to 52n of one group image for each frame described later.

A method for comparing and detecting areas where image data changes in the above-described CPU will be described with reference to the flowchart of FIG.

For example, image data of one image of the current frame captured from the video camera CA ₁ is stored in the memory 23.
a of the same camera and one image (for example, one frame before) of the same camera already stored in the memory 24a. Image data is 8x8 pixels or 16x
Fine block N of about 16 pixels (matrix)
And the first block of one image to the i-th block
The N-th block, for example, the i-th block is set in the first step ST
I for one image before to after the form with i = 0 to ₁ of like (before image)
The block and the i-th block of the current image (new image)
Comparing at step ST _2.

The CPU 6a executes the third step ST_ThreeWith the previous image
And whether the i-th block of the new image is the same image is determined.
If the same image data is YES, the ninth step
ST ₉Then, the i + 1-th is established, and i = i + 1.
You.

[0057] CPU6a the i it is determined whether or not a predetermined number of blocks N in the tenth step ST ₁₀ in the following, i =
If YES is N, the moving object detection process ends, but NO
If it is, the process returns to the second step ST ₂ and the next block i
A comparison is made between the +1 previous image and the new image.

[0058] If NO is not the same image is determined in the third step ST ₃ is advanced to the fourth step ST _4. 4th
Step registered as ST is ₄ the image change of the region are denoted by the flag F1 is stored in the search information storage unit 52a~52n of recording medium 14 (see FIG. 20).

In the next fifth step ST ₅ , a motion vector scan of a region having a change is performed.
Determine whether or not the motion vector was detected by ₆ ;
If NO proceed to a ninth step ST ₉ is set to the next i + 1 th block, if the motion vector is detected is YES advances to the seventh step ST _7, the seventh step ST ₇ in its motion vector information In order to save the moving direction, the moving direction and the moving amount are calculated and used for searching and calculating the moving direction and the moving amount of the moving object.

[0060] The recording medium 14 with the flag F ₂ in the eighth step ST ₈ (HD). 8th step ST
After ₈ terminates is advanced to a ninth step ST _9.

The information (motion vector) of the changing area in the image data stored in the memory described above is used for determining the compression range of the image in performing the image compression processing.

[0062] The motion vector detection method of the sixth step ST ₆ in FIG. 4 will be described with reference to the flowchart of FIG. 15.

In the first step _ST6a , 1 to M are set at the upper, lower, left and right positions around the new image block with respect to the previous image block.
A comparison is made to see if it matches the pixel shifted,
As in the second step ST6b, the CPU _6a determines whether or not the comparison images match, and if NO, the motion vector scan ends. If YES, the seventh comparison in FIG. as moving object goes to step ST _7, the moving direction, calculates the moving amount, the seventh step ST
Through ₇ , the motion vector scan ends.

When the image matches, the moving direction and the moving amount for the block are recorded on the recording medium 14 via the bus 10a and the I / F 13a, thereby facilitating the search.

The detection method at the time of detecting the motion vector will be described with reference to FIGS. 6A to 6D.
Is a captured image in which the camera is monitoring the building, and shows a state in which no moving object exists in the screen. In FIG. 6B, a moving object 26a is approaching the building, and the moving object 26b is detected as an area where the image changes. FIG.
FIG. 6C shows a case where a is further approaching the building and 27b is detected as a region where the image changes.
8 indicates that the moving object stops at the position of the area 27b and is the same image as the image of 27.

The image data 25 in which the moving object 26a is not detected is stored at a certain arbitrarily determined interval (for example, 1 per second).
Times), the entire screen is compressed and recorded on the recording medium 14. When the moving object 26b is detected by the comparator 6c as an area where the image changes, only the area 26b where the image changes is compressed by the compressor 4, which will be described later, and is recorded on the recording medium 14. Next, a captured image 26 and a captured image 27
And the moving direction and the moving amount (distance) of the moving object 26a.
When the position of the moving object 26a is determined by the vector operation and the region 27b where the image changes is detected, only the region 27b where the image changes is compressed by the compressor 4 and recorded on the recording medium 14. At this time, information such as the detected moving direction of the moving object 26a and the position of the moving object 26a determined by the moving amount (distance) vector calculation is recorded on the recording medium 14 as a flag.

As described above, when the comparison processing is performed by the comparator 6c and a region having a change in the image is detected, the detected image data is sent to a JPEG or wavelet encoder (compressor) 4 as a compressor. Sent.

It should be noted that A / D2a to A / D2a shown in FIG.
By switching the contact piece a of the EE system switch 15 for directly outputting from the 2c (see FIG. 2) to the D / A 16 to the fixed contact b, for example, one image equivalent to one image at the time of expansion in the reproduction path.
Since the frame memory 3 stores the Y data, RY data, and BY data for a frame via a JPEG or wavelet decoder 5 as an expander, the frame memory 3 outputs the data to the D / A 16 via the switch 15. and it outputs an analog video signal 28a, and 28b to the output terminal T _01, T ₀₂ via the video signal processing output unit 17.

The outline of the image compression / decompression of the JPEG conversion method using the JPEG or wavelet encoder 4 and the JPEG or wavelet decoder 5 will be described with reference to FIGS. In the JPEG conversion method, image data is divided into blocks of 8 × 8 pixels and input to the DCT calculator 35. DCT calculator 35
, A two-dimensional DCT operation is performed, and the image data is converted into DCT coefficients, that is, spatial frequency components.

The obtained DCT coefficient is quantized by the quantizer 3
In step 7, linear quantization is performed using the quantization coefficient 38. JPE
In the G conversion method, the image quality and the compression ratio can be controlled by changing the quantization coefficient 38.

The quantized DCT coefficients are encoded by a run-length encoder 39 and an entropy encoder 40 having a table value given from a Huffman table 41, and output as compressed data. When stretching,
As shown in FIG. 8, the Huffman table 41, the entropy decompressor 47, the run-length decompressor 46, the decompression coefficient 45, the inverse quantizer 44, and the inverse DCT operator 42 perform the above steps in reverse to expand the image data. Is done.

Next, a wavelet (wavelet)
The conversion method will be described with reference to FIGS.

FIG. 9 is a system diagram of an encoder based on the wavelet method, and FIG. 10 is a system diagram of a similar decoder. The same parts as those in the JPEG conversion method shown in FIG. 7 and FIG.
Wavelet encoder uses DC of JPEG conversion method
The T-transform unit 35 is a wavelet separation filter 36, and the inverse DCT transform unit 42 of the JPEG conversion method is a wavelet combining filter 43 in the wavelet decoder. That is, in the transform by the wavelet,
By using subband coding, an input signal is divided into several frequency bands and processed as a set of data by subbands and filters without dividing an image into blocks. As a result, JPEG and MPEG
Specific block distortion and luminance distortion can be eliminated. In the wavelet transform, the fact that the low frequency component among the subband encoded frequency components contains a lot of information,
Similar to CT, quantization is performed to give a large number of bits to a component having a large amount of information, thereby achieving efficient compression. In other words, compression /
In the decompression, as shown in FIGS. 10 and 11, the DCT operator 35 and the inverse DCT operator 42 in JPEG are replaced with subband encoders / decoders (wavelet separation filters and wavelet combining filters) 36 and 42, respectively. With common hardware, it is possible to switch between JPEG and wavelet.

FIG. 11 corresponds to FIGS. 7 and 9, and FIG. 12 corresponds to FIGS. 8 and 10. 11 and SW ₂ and SW ₃ in the system path of the encoder and decoder of FIG. 12 JP
A switch for controlling switching between the EG conversion method and the wavelet conversion method, and is controlled by the CPU 6a (not shown).

The above-described JPEG conversion method and wavelet conversion method will be further considered.

In JPEG and MPEG, a compression method using DCT is used. In the orthogonal transform encoding to which the DCT belongs, a plurality of pixel data are collectively transformed by an orthogonal transform matrix, quantized, and replaced with encoded data. In other words, whether the information amount of the image signal can be efficiently compressed depends on whether
It depends on how you choose this transformation matrix. Currently D
CT has come to be recognized as the most efficient compression method. However, there are problems with DCT. If the quantization of the transform coefficient is coarse, block distortion occurs, and if the DCT coefficient lacks high-frequency components, mosquito noise often occurs. Block distortion relates to quantization errors of DC components and low-frequency components. On the other hand, the transform base of DCT does not converge to a value of zero at both ends of each block. Here also, a quantization error occurs.

On the other hand, in the wavelet transform, as described above, the frequency is converted to a QMF (Quadrature Mirror).
or Filter: orthogonal mirror image filter). This group of QMFs is called a subband filter bank, and a method of transforming each of them using such a group of filters is generally called subband coding (coating). This division is encoded independently for each frequency. In the case of orthogonal transform, information can be compressed using the redundancy between different coefficients in a pixel block.

In the conversion by the wavelet, the image is processed as one unit of data without dividing the image into blocks by using sub-band coding. As a result, block distortion and luminance distortion peculiar to JPEG and MPEG can be eliminated. In the wavelet transform, the low-frequency component among the sub-band coded frequency components utilizes the fact that it contains a large amount of information. Achieving effective compression. Since the wavelet transform uses sub-band coding, when the compression ratio is increased, high-frequency components are lost, resulting in an overall blurred image. On the other hand, DCT
As described above, an image with a high compression ratio due to the above has extremely large block distortion and mosquito noise, and cannot maintain an image quality level required for use in a monitoring device or the like that can identify a person. On the other hand, in an image with a low compression ratio, DCT can obtain a sharp and sharp image.

That is, when the transfer capability to the recording medium 14 is high or when the bandwidth of the external transmission path is sufficiently wide, the D
Recording is performed using the JPEG conversion method using CT, and conversely, when the transfer capacity of the recording medium 14 is low or when the external transmission path band is narrow, recording is performed using compression of the wavelet conversion method. It is preferable to perform data transfer suitable for the transmission path by designating a band that can be considered in advance. This switching of the conversion method can also be performed by a difference in compression ratio. In other words, when recording on the recording medium 14 for a long time, the compression rate must be increased.
If the conversion method is used and the compression ratio is set high, block distortion occurs, which is not suitable for image recording in applications such as monitoring devices. Therefore, when recording is performed for a long time with a high compression ratio, by switching to the wavelet transform method and performing recording, an overall blurred impression is obtained. An image quality level at which identification is possible can be maintained, and recording can be performed for a long time.

Next, the Hadamard transform (H
A method for increasing the compression efficiency will be described with reference to “adamard transform”. In the encoder of FIG. 13, reference numerals 36 to 41 are the same as those of the encoder for wavelet transform of FIG. 9, and the corresponding parts are denoted by the same reference numerals and redundant description is omitted. In FIG. 13, a Hadamard transform processing circuit 48 is inserted between the wavelet separation filter 36 and the quantizer 37. In the Hadamard transform, a transform matrix is composed of +1 and -1, and a rectangular waveform is used as a base vector.

The wavelet separation filter 36 corresponds to FIG.
As shown in FIG. 4, it is composed of a sub-band coating filter such as QMF.

In FIG. 14, a digital image input which is video data is divided into a high frequency component and a low frequency component. In order to secure an area with a lot of noise components, instead of multi-stage division, generally, a high-frequency component obtained by dividing a tree into two in two is used.

In FIG. 14, a first band-pass filter (hereinafter referred to as water, low BPF) 50 for extracting low-frequency components in the horizontal direction
a, a first band-pass filter (hereinafter referred to as water, referred to as high BPF) 50b for extracting high-frequency components in the vertical direction, and sampling by decimation (decimation) 50g 2:
1 through the decimation filter 50e,
It is divided into a second low-frequency component in the vertical direction (hereinafter, referred to as a vertical, low BPF) 50c and a second high-frequency component in the vertical direction (hereinafter, referred to as a vertical, high BPF) 50d.

Water, output of high BPF 50b is also vertical, low BPF
15 through the decimation 50g and the decimation filters 50e and 50f.
High region W1 in the vertical and horizontal directions on the screen shown in FIG.
LH. The partial image data of W1HH49 is obtained. This part collects the high frequency components of the video image. Since noise components are generally concentrated in high-frequency components, a Hadamard transform processing circuit 48 is inserted into the W1HH49 portion and subjected to Hadamard processing, whereby noise components can be removed without impairing the resolution of image quality.

Here, the Hadamard transform processing circuit 48 shown in FIG. 13 comprises a Hadamard transform section 48a, an energy component comparator 48b, a filtering section and a non-linear coating section 48c.

In FIG. 14, the output of the vertical BPF 50d is obtained through the decimation 50g and the decimation filter 50f to obtain the image data in the W1HL area of FIG. 15, and the output of the vertical BPF is divided into two tree-like forms. Figure 1
5 is obtained, and similarly divided into a tree three times to obtain W3LH, W3LH, and W3LH of FIG.
Image data of the W3HH and G areas is obtained.

In FIG. 14, the image data supplied to the Hadamard transform processing circuit 48 is differential data and is suitable for performing Hadamard transform. Each of the high-frequency energy components is compared by the comparator 6c, and a biased high portion is extracted, and only the frequency component is filtered, but the presence or absence of the two-dimensional periodicity of each block is detected. If no such detection is made, the component is noise. However, since it is sometimes difficult to discriminate between noise and image data, that portion is weighted and the signal component is reproduced again by conversion.

As described above, the wavelet separation filter 3
The image data having passed through the area 6 is stored in an area 49 of W1HH in FIG.
High-frequency components are collected in the image. On the other hand, since the noise component generally has a characteristic of being concentrated in the high-frequency component, the Hadamard transform process is performed on the portion of the region 49 of the W1HH through the Hadamard transform processing circuit 48, without deteriorating the resolution of the image quality. Noise components can be removed. In other words, the noise component has a very high randomness, and especially the noise component superimposed on the moving image data is
Very low correlation with general natural images. Focusing on this, in the two-dimensional data of NxN sample blocks used in the JPEG conversion method and the like, the random signal has no correlation by performing two-dimensional Hadamard transform coding for each divided block. The components can be extracted. That is, the presence or absence of the periodicity of each block in two dimensions is detected, and if there is no such component, the component is noise, so that the portion is discarded, so that noise reduction can be realized. In this case, the noise reduction 1b in the signal processing unit 1 in FIG.
Can be substituted.

Next, referring to FIG. 16, the memory 23a shown in FIG.
23n and 24a to 24n will be described.

[0090] for one video camera CA ₁ is an imaging means 2, two one image memory 23a and 2
4a were used as a pair, but in this example, a plurality of video cameras C
The input of the n video cameras A _{1 to} CA _n or the n video switchers 1 is made up of n plus one image memory.

In FIG. 16, parts corresponding to those in FIG. 2 are denoted by the same reference numerals, and redundant description is omitted.
Video switcher against A ₁ to CA _n 1 and A / D2
For the case where the movable contact piece a sequential (n + 1) fixed contacts b, c, d, e ‥‥ n, is CPU6a as sequentially switched to n + 1 are controlled by the switch SW ₄ via, the input image data memory The method of storing data in the storage areas 23a to 23n _{+ 1} will be described below.

First, first, the video cameras CA _{1 to} CA _n
Image data corresponding to the camera inputs 21a to 21n
The images are stored in the memories 23b, 23c, 23d ‥‥ 23n + 1 configured as frames or field memories for the images, respectively.

Next, the switch SW_FourContact piece a is fixed contact b
Back to the video camera CA₁Next image data of one image
In the memory 23a. Here already a video camera
CA ₁23b storing the image data before
And the comparator 6c.

[0094] After the comparison process leaves the image data in the memory 23a as the image data of the camera input 21a of the video camera CA _1, a new camera input 21b of the video camera CA ₂ is stored as image data in the memory 23b, a video camera The image data in the memory 23c in which the image data as the camera input 21b of the CA ₂ is stored is compared through the comparator 6c. Likewise leaving image data in the memory 23b as the video input 21b of the video camera CA _2,
Video input 21c of video camera CA ₃ into memory 23c
Captures image data. By such repetition, the number of inputs of the video switcher 1 or the video cameras CA ₁ -C
Comparison of the temporal change of the image data of the video camera CA ₁ to CA _n number plus one for one image memory is of n if the A _n becomes possible.

Next, the noise reduction 1b in FIG. 2 will be described in detail with reference to FIGS. FIG. 17 shows a noise reduction circuit formed by an analog circuit.

In FIG. 17, image data input from _one of the video circuits CA _{1 to} CA _n is supplied to a high-pass filter (hereinafter, referred to as HPF) 29 and a low-pass filter (low-pass filter: hereinafter, LPF). The process branches to 30. The output of the LPF 30 is supplied to a delay circuit 32 to correct a delay amount of a path of the HPF 29 and the nonlinear amplification processing circuit 31.

The output of the HPF 29 is a non-linear amplification circuit 3
The signal is supplied to one large-amplitude passing circuit 31a and one small-amplitude passing circuit 31b. The output of the small-amplitude passing circuit 31b is supplied to a first adding circuit 31d via an integrating circuit 31a, and the output of the large-amplitude passing circuit 31a is supplied to a second adding circuit 31d. As described above, for example, suppression and averaging are performed as in a characteristic curve 31e utilizing diode characteristics, and the second adder circuit 33 adds the image data of the system path on the LPF 30 side and outputs the result to the Y / C separation circuit 1c.

At the time of reproduction, a compression efficiency can be further improved by providing an expansion circuit in the video signal processing / output unit 17 for expansion processing based on the inverse characteristic of the nonlinear amplification processing circuit 31 and expanding the expanded processing. There is also a method of extending this method, monitoring the amount of compressed data, and adding feedback to make the data transfer rate constant. That is, this nonlinear amplification processing circuit 31 is put in a feedback loop,
The amount of data is monitored, and when the amount of data increases, the amount of amplification of the high frequency component of the nonlinear amplifier circuit 31 is reduced. On the other hand, when the data amount is small, the high-frequency component can be made to be nearly flat. In this way, the transfer rate can be kept constant, and the transfer rate can be set arbitrarily.

Further, as to the method of changing the frequency characteristic, the frequency axis (frequency domain)
However, in this method, the efficiency of compression becomes a problem. Therefore, using the sub-Nyquist theorem, a filter on the frequency axis is provided before A / D2 so that the sampling frequency is variable, and an image without aliasing is obtained. Can also be reduced. Of course, the same effect can be obtained even if the filter placed before A / D2 is digitally performed.

FIG. 19 shows a block configuration of a high-pass system in the case of performing a digital circuit. The output of the HPF 29 is output to a nonlinear amplification processing circuit 31 having a digital configuration. This non-linear amplification processing circuit 31 includes an upper bit delay and through circuit 31a ', a lower bit 8 × 8 pixel addition circuit 31
b ', a divider 31c', and an adder 31d '.

Upper bit delay and through circuit 31a '
In this case, the upper 5 bits and the lower 6 bits give a delay equivalent to the processing time of the lower bits. In the lower bit n × n pixel adding circuit 31b ′, for example, the lower 2 to 3 bits divide the sum (sigma) of 8 × 8 pixels. By adding 1/64 to the upper bits by the adder 31d 'in the arithmetic unit 31c', the same lower bits are added to the 8.times.8 pixel block, but the low-frequency components remain as they are. As a result,
Although a slight loss occurs in the signal component, there is no visual problem.

Generally speaking, in the relationship between image quality and compression, it is necessary to increase the efficiency of compression. However, the SN ratio, which is the signal-to-noise ratio, and the frequency characteristics greatly affect the compression efficiency. For the Y signal, a resolution equivalent to 7 bits is generally regarded as a limit that cannot be visually recognized unless an effect operation or an editing operation is performed. For this reason, when it is desired to increase the compression ratio, the least significant bit of the code string, LS
For the B2 bits, the LSB1 bit is separated and 2
Even if the bit number is rounded off, the image is almost safe. However, a surveillance camera requires a higher resolution (sharpness) than a general home use. It is necessary to vary the noise reduction and the frequency characteristic while keeping the compression ratio in balance without lowering the frequency characteristic and the resolution as much as possible.

Next, the operation of the above-described video signal recording monitoring device will be specifically described. Here, when the transfer speed of the recording medium 14 is about 2 Mbytes / sec, the image data is 720 × 480 pixels × 2 (Y signal and C signal) = 691.
2K bytes 691.2K bytes x 30 frames / sec = 20.74M
Since it is bytes / second, it is not realistic to record it on the recording medium 14 as it is. Therefore, for example, the NTSC signal is compressed to about one tenth of 2.07 Mbytes / sec by JPEG or wavelet conversion method (or it is compressed to 1.04 Mbytes / sec by dropping one field). By changing the sampling frequency to 1/2 in the horizontal direction and 1/2 in the vertical direction, the transfer rate can be reduced to about 1/4.

However, even if one field is dropped by the JPEG or wavelet conversion method and compressed to one tenth (= 1.04 Mbytes / sec), it is recorded on a 20 gigabyte (10 ⁹ ) bytes hard disk. ,
It is the best to record for about 5.3 hours. If the recording time is to be extended, it is necessary to perform intermittent recording and extend the time.

In FIG. 6, the data 25 of the image in which no moving object is detected is compressed at a certain arbitrarily determined interval (for example, once per second), and the entire screen (texture) is compressed and recorded on a recording medium. When the comparator 6c detects an area (object) 26b where the image changes, only the object 26b is compressed by the compressor 4 and recorded on the recording medium 14.

Next, the image 26 and the image 27 are compared, the moving direction and the moving amount (distance) of the moving object 26a, the position of the moving object 26a are determined by vector calculation, and the area (object) 27b where the image changes Is detected, only the object 27b is compressed by the compressor 4, and the recording medium 14
Recorded in. At that time, moving direction and moving amount of the detected moving object 26a (distance), the position of the moving object 26a that has been determined by the vector operation, the video camera CA ₁ to CA _n
The information such as the number is recorded together as a flag. The above description is for one video camera input,
When the number of input cameras is large, the comparison is performed in the memory for one image corresponding to each camera, and the recording operation is performed in the same manner.

In FIG. 1, the compressed image data is
The recording medium 14 or D temporarily passes through the bus line 10a.
The data is stored in the work area of the RAM 8 and is not recorded in the storage area of the recording medium 14. Further, when the transfer capability of the recording medium 14 is low or when data is transferred via the LAN I / F 13b, it is necessary to keep the transfer rate constant. As shown in FIG. 1, the CPU 6a always counts the amount of compressed data per unit time on the bus line 10a, and controls the compression coefficient 38 in FIG. 11 so that this data amount becomes substantially constant.

In wavelet-based transformation, an image is processed as a set of data without subdividing an image by using sub-band coding. As a result, block distortion and luminance distortion peculiar to JPEG and MPEG can be eliminated. In the wavelet transform, low-frequency components among sub-band encoded frequency components use the fact that they contain a lot of information, and, like DCT, quantization that gives many bits to components with a large amount of information is performed. Achieving effective compression. Since the wavelet transform uses sub-band coding, when the compression ratio is increased, high-frequency components are lost, resulting in an overall blurred image. On the other hand, DC
As described above, an image with a high compression ratio due to T has an extremely large amount of block distortion and mosquito noise, and cannot maintain an image quality level required for surveillance use and capable of identifying a person. On the other hand, for an image with a low compression ratio,
CT can obtain a sharp and sharp image.

That is, when the transfer capability of the recording medium 14 is high or when the external transmission band of the LAN I / F 13b is sufficiently wide, the recording by the JPEG conversion method using DCT is performed, and When the transfer capacity is low, LA
When the external transmission band of the N I / F 13b is narrow, recording can be performed by using compression of the wavelet transform method. The conversion method can be switched also by the compression ratio. In other words, when performing long-time recording on the recording medium 14, the compression ratio must be increased.
If the EG conversion method is used and the compression ratio is set high, block distortion occurs, which is not suitable for image recording in monitoring applications. Therefore, when recording is performed for a long time with a high compression ratio, by switching to the wavelet transform method and performing recording, an overall blurred impression is obtained. An image quality level at which identification is possible can be maintained, and recording can be performed for a long time.

The description returns to the recording operation. The compressed image data is temporarily stored in the recording medium 14 or the DRAM 8
And is always recorded.
When recording at 30 frames / sec without intermittent recording, all data is transferred from the work area to the storage area of the recording medium 14. In the case of performing intermittent recording, the image data stored in the work area can be intermittently transferred to the storage area of the recording medium 14 by specifying frame skipping. Since data is always recorded in the work area, for example, it is possible to send stored data only when the moving amount of the moving object exceeds a certain set value, and it is possible to detect a region where the image changes, for example. Recording can only be performed if It is also possible to transfer the recording data from the situation before the area where the image has changed is detected to the storage area.

In the reproduction operation of the recorded image data, the image data stored in the storage area of the recording medium
The compressed data managed by the PU 6a and corresponding to the input channel of each video camera is read. The compressed image data read from the recording medium 14 is an I / F
The data is transferred to the JPEG or wavelet decoder 5 via the bus 10a via the bus 13a. Basically, the texture image of the specified channel is expanded, and the image of the object is expanded on the expanded image, and the expanded image is reproduced. The channel is specified by the operation unit 11b such as a manual.
An instruction is sent to a.

Next, shortening of the search time will be described with reference to FIGS. 20 and 21. FIG. The compressed image, grouped per video camera CA ₁ to CA _n the number frames as in Figure 20, with the management table 53 shown in FIG. 21 for managing the group 51a~51n on DRAM8 of FIG . Each of the groups 1 to n shown in FIG.
a to 54n, 56a to 56n, image data (texture) 55a to 55n obtained by compressing the entire screen, and image data (objects) 57a to 57n obtained by compressing an area where the image changes accompanying the image data. The head of each group has search information 52a necessary for search.

In the management table, as shown in FIG.
Address information 58 on the recording medium 14 of each group
a to 58c, time information 59a to 59c indicating recording start time and recording end time in the group, flags 60a to 60c indicating presence or absence of a moving object (object), and the like.
Further, the search information 52a to 52n includes information such as the moving direction, moving amount, size, position on the screen, and recording time of the moving object (object) of each image data of the frame.

For example, in a case where a search is performed using only time information in monitoring outside the business hours of a store, a relevant group is narrowed down from the time information in the management table, and the time is shortened by looking at the time information in the group. Searchable on time. Further, when searching for an intruder outside business hours, the group can be narrowed down by the time information and a flag indicating the presence or absence of a moving object (intruder). More detailed search by position and size is also possible.

As described above, when there is no intruder, since the flag of the moving object is not set, it is possible to instantly confirm that there is no abnormality, and when there is an intruder, the corresponding image can be browsed at high speed. Becomes possible.

[0116]

As described above, according to the present invention, the redundancy of the JPEG method and the compression of the wavelet transform method can be reduced by sufficiently eliminating redundancy so as to minimize the amount of data to be compressed per unit time. Compares the input image from the same camera between the previous and next frames, detects the area where the image changes from the comparison result, compresses only the area where the image changes, and records it on the recording medium. Therefore, the data amount can be sufficiently reduced. Further, in a recording medium having the same capacity, a recording time which is ten and several times longer can be obtained, and efficient recording can be performed. At the same time, the search method has been improved, and instead of the conventional simple search using the time code at the time of recording, screen information such as the moving direction and moving amount (distance) of the moving object and the position of the moving object are added as flags. , You can search instantly.

Further, since images are compared before being compressed, the image data is constantly monitored, and if there is a movement, the image data is stored in a recording medium such as an image memory for a predetermined time. . Also, the obtained motion data is used for monitoring purposes, and only when the temporal change exceeds a predetermined value, the compressed image data is recorded on the recording medium from the stored memory. Recording can be performed only when the image that changes is changed, and since it is always recorded in the memory, important moments can be ascertained from image signals captured by surveillance cameras, including before and after necessary scenes Can be recorded and played back.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a video signal recording monitoring device according to the present invention.

FIG. 2 shows a video switcher / signal processing unit and an A /
It is a block diagram which shows a D converter part in detail.

FIG. 3 is a block diagram of a synchronization signal generation genlock circuit and a system clock generation circuit of FIG. 1;

FIG. 4 is a flowchart illustrating a motion vector determination of a moving object.

FIG. 5 is a flowchart illustrating motion vector detection.

FIG. 6 is an image diagram used for describing an image detection method input from a camera.

FIG. 7 is a block diagram illustrating a JPEG encoder.

FIG. 8 is a block diagram illustrating a JPEG decoder.

FIG. 9 is a block diagram illustrating a wavelet encoder.

FIG. 10 is a block diagram illustrating a wavelet decoder.

FIG. 11 is a block diagram illustrating a JPEG or wavelet encoder of FIG. 1;

FIG. 12 is a block diagram illustrating a JPEG or wavelet decoder of FIG. 1;

FIG. 13 is a block diagram illustrating a Hadamard transform process.

FIG. 14 is a diagram showing a structure of a sub-band coating filter.

FIG. 15 is a screen diagram for explaining a wavelet transform of an image.

FIG. 16 is a block diagram showing another configuration of the memory of FIG. 1;

FIG. 17 is a block diagram of a noise reduction unit.

18 is an explanatory diagram of the operation of the nonlinear amplification processing circuit of FIG. 17;

19 is a block diagram when the noise reduction unit of FIG. 17 is digitized.

FIG. 20 is a diagram modeling a storage form of compressed image data in a recording medium.

FIG. 21 is a modeled view of a management table for managing and searching for image data of a recording medium.

[Explanation of symbols]

1 video switcher / signal processing device, 1a video switcher unit, 1b noise reduction, 1c
‥ Y / C separation unit, 2 ‥‥ A / D converter, 4 ‥‥ JPEG
Or a wavelet encoder, 5 @ JPEG or wavelet decoder, 6a @ CPU, 6
b {DMA controller, 6c comparator, 14}
‥ recording medium, 16 ‥‥ D / A converter, 17 ‥‥ video signal processing / output unit, CA ₁ to CA _n ‥‥ video camera, 2
9 high-pass filter (HPF), 30 low-pass filter (LPF), 31 nonlinear amplification processing circuit, 3
5 DCT calculator, 36 wavelet separation filter, 48 Hadamard transform processing unit, 50 a horizontal low band pass filter, 50 b horizontal high band pass filter, 50 c vertical low band pass filter,
50d ‥‥ vertical high-pass filter, 50e, 50
f ‥‥ Decimation filter

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H03M 7/46 H04N 7/18 D 5L096 H04N 5/765 5/91 K 5/781 5/781 510L 7 / 30 520A 7/32 7/133 Z 7/18 7/137 Z F term (reference) 5C053 FA11 FA14 GB17 GB22 GB26 GB36 HA29 KA04 KA24 KA25 LA14 5C054 AA01 CA04 CC02 CH03 CH09 EA05 EB01 EB05 EB07 EC07 EJ05 FC13 FE07 FE07 FE07 FE07 GB01 GD05 HA18 5C059 KK03 KK04 KK08 KK22 KK34 KK37 LA01 MA22 MA23 MA24 ME02 ME05 NN03 NN24 NN28 RA01 RC16 RC17 SS00 TA62 TB04 TC12 TC13 TD05 UA05 UA09 UA29 UA30 UA32 UA35 5D044 AB07 CC04 DE07 ABC BC11 BC16 BD03 5L096 AA02 AA06 BA02 CA02 GA08 GA17 HA03 LA05

Claims (7)

[Claims] An image pickup signal obtained by obtaining a plurality of subject images through a plurality of image pickup means is digitized, and the digitized image data is stored in a temporary storage means and then recorded on a recording medium. A video signal recording monitoring device, wherein the temporary storage means has a capacity of at least one image data or more as one unit, and has a total capacity of one or more inputs of the plurality of imaging means, and the temporary storage. Comparing means for reading out the image data before and after the imaging signal of the same imaging means written in the means from the temporary storage means for comparison, and detecting means for detecting a change area of the image data from the comparison output of the comparing means And compression means for compressing the change area of the image data by a high-efficiency encoding method, wherein the image data compressed by the compression means and the detection extracted by the detection means are provided. Broadcast the video signal recording monitoring apparatus characterized by comprising recording on the recording medium. 2. A video signal recording apparatus which digitizes an image signal obtained through at least one or more image pickup means, stores the digitized image data in a temporary storage means, and then records the digitized image data on a recording medium. A monitoring device, wherein the capacity is at least one image data or more as one unit, and the total storage capacity is the number of inputs of the at least one or more imaging means plus one or more capacities; Comparing means for reading and comparing the image data before and after written in the temporary storage means from the temporary storage means for comparison, and movement amount extraction means for extracting the movement amount of the moving object of the image data from the comparison output of the comparison means; Threshold comparing means for comparing the magnitude of the motion vector amount calculated by the moving amount extracting means with a preset threshold, and an output level of the threshold comparing means Video signal recording monitoring apparatus characterized by starting the recording onto the recording means when it exceeds a predetermined level. 3. A video signal recording apparatus for digitizing an image signal obtained through at least one or more image pickup means, storing the digitized image data in a temporary storage means and then recording the digitized image data on a recording medium. A monitoring device, wherein the capacity is at least one image data or more as one unit, and the total storage capacity is the number of inputs of the at least one or more imaging means plus one or more capacities; Comparing means for reading and comparing the image data before and after written in the temporary storage means from the temporary storage means for comparison, and movement amount extraction means for extracting the movement amount of the moving object of the image data from the comparison output of the comparison means; A flag is attached to each output from the detecting means for detecting the change area of the image data and the moving amount extracting means for extracting the moving amount of the moving object, and the recording medium A video signal recording monitoring device characterized in that the video signal is recorded on a video signal recording device. 4. The video signal recording monitoring apparatus according to claim 1, further comprising a noise reduction unit configured to reduce noise of a high frequency component of an image signal of the imaging unit. 5. The video signal recording monitoring apparatus according to claim 1, wherein said compression means includes a Hadamard transform means as a high-frequency component reduction means of a wavelet transform method. 6. A temporary storage unit for digitizing an image signal obtained by imaging an object by an imaging unit by an analog-to-digital conversion unit, and recording the digitized image data; a JPEG conversion unit; a wavelet conversion unit; Means for changing the compression ratio of both the conversion means, wherein the JPEG is determined by the band condition of the transmission path when transmitting image data or the compression ratio when compressing the image data.
A video signal recording monitoring apparatus characterized in that the method is switched to the conversion method or the wavelet conversion method, and is recorded on a recording medium to increase the compression efficiency. 7. A means for varying a sampling clock in a vertical and horizontal video section at the time of the analog-digital conversion, wherein the JPEG conversion method or the JPEG conversion method is performed according to a sampling rate or a compression rate at the time of the analog-digital conversion. 7. The video signal recording monitoring apparatus according to claim 6, wherein the compression efficiency is increased by switching the wavelet transform method.