JPH0429472A - Picture motion detector - Google Patents

Abstract

PURPOSE:To accurately detect only the frame where a moving object exists in a prescribed frame by synthesizing areas where the moving object exists in the prescribed frame and the preceding frame and areas where the moving object exists in the prescribed frame and the following frame. CONSTITUTION:A difference and threshold processing means 1 takes the difference between the signal of a picture of the n-th frame and that of a picture 10 of the (n-1)th frame and outputs the area, where the difference exceeds a threshold, as a motion area. This output is stored in a frame memory 2a for one frame period. Next, the processing means 1 takes the difference between the signal of a picture of the (n+1)th frame and that of the picture 11 of the n-th frame and outputs the area, where the difference exceeds the threshold, as a motion. An AND operating means 3 operates AND between a motion area 16 obtained by comparison between pictures of the (n-1)th frame and the n-th frame and a motion area 17 obtained by comparison between pictures of the n-th frame and the (n+1)th frame and outputs the result as an area 14 of the moving object of the n-th frame.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to an image processing device.The present invention relates to an image motion detection device configured to be able to detect only moving objects from a background.

Conventional technologyAs a device for detecting moving objects in images,
The apparatus shown in FIG. 11 stores the image signal of the previous frame in the frame memory 100, and when the image signal of the current frame is input, the difference/threshold means 101 i! The difference between the signals of each pixel in the previous frame and the current frame is calculated, and the pixels for which this difference value exceeds a predetermined value are output as motion areas. When an object moves in the current frame image (solid line), this device calculates the difference between the two frames to identify the area 102 that has movement in the image, as shown by diagonal lines in FIG. 12(b). However, the problem that the invention attempts to solve is that the conventional image motion detection device
As is clear from Figure (b) 102, an area 103 where a moving object exists in the grinding frame and an area 104 where a moving object exists in the current frame are simultaneously determined, and an area where a moving object exists in the current frame. 104
The present invention solves the above-mentioned conventional problems.It is an object of the present invention to provide an image motion detection device that can accurately detect only the area where a moving object exists in a predetermined frame. Another object of the present invention is to provide a background extraction device for extracting a background image from which moving objects have been removed from an input image. An object of the present invention is to provide an excellent residual object detection device capable of detecting only residual objects.The image motion detection device of the present invention detects only areas where moving objects exist in a predetermined frame. a means for comparing at least two or more frames of images and detecting an area where the signal values differ; a memory for storing the detection output for at least one frame;
the detection output and the output read from the memory may be combined and output as a motion area; or a plurality of means for comparing at least two or more frames of images and detecting areas with different signal values; means for synthesizing the detection outputs of the plurality of means and outputting it as a motion region, and using the output of the motion region as a control signal;
It has means for synthesizing one frame image from two frame images and outputting it as a background image, and comparing the synthesized one frame image with the at least two or more frame images to detect areas where signals differ. To at least one of the plurality of means In a motion detection device for an active image, even if the motion image has the characteristic of being a one-frame image used for comparison, the detection output is at least l
A memory for storing more than one frame, and a means for synthesizing the detection output and the output read from the memory and outputting it as a motion area, are used to determine the area where a moving object exists in a predetermined frame and the previous frame, and the area where a moving object exists in a predetermined frame and the previous frame. By compositing the areas where moving objects exist in the next frame, it is possible to remove areas where moving objects exist only in frames other than the predetermined frame, and accurately detect only the areas where moving objects exist in the predetermined frame. 4 Also, by means of combining two images in a moving area according to a signal, a background image when there is no moving object can be combined. By means of comparing this output with an image of a predetermined frame, a predetermined image can be synthesized. Although it is possible to accurately detect only the area where a moving object exists in a frame, the area where a moving object exists is detected in a predetermined frame and a frame other than the predetermined frame when using other means for comparing images.Detection outputs of multiple methods By combining the images and outputting them as a motion area,
Although it is possible to accurately detect only the area where a moving object exists in a predetermined frame, the means for comparing images for detecting the area where a moving object exists in a predetermined frame and a frame other than the predetermined frame is a method that detects a signal in a moving area in an initial state. The system operates to detect a moving area until a background image in the absence of a moving object is synthesized by a means for synthesizing two images according to the same method. It is possible to realize a background extraction device that extracts a removed background image, and by applying the background extraction device, a background image extraction means that extracts a background image with moving objects removed from an input image, and a background image extraction device that extracts a background image with moving objects removed from an input image, and a background image extraction device that extracts a background image without moving objects from an input image. A residual object detection device that detects a residual object from an input image can be realized by a frame memory that stores an image and a means that compares the output of the background image extracting means with the frame memory and detects an area where the signal value differs.

EXAMPLE A first example of the present invention will be described below with reference to the drawings. FIG. 1 shows a block diagram of an image motion detection device according to a first embodiment of the present invention. 1st
In the figure, 1 is a difference/threshold processing means, 2, 2
a is a frame memory, 3 is an AND operation means, 45 is an input terminal, and 6 is an output terminal.4 The operation of this image motion detection device will be explained using FIG. 2. First, in FIG. 1o is n-17L/
11 is an n-frame image 12 is an n+1 frame image 13 is an n-1 frame moving object area 1
4 is the area of the moving object in the n frame. 15 is the area of the moving object in the n+1 frame. 16 is the movement area when comparing the images of the n-1 frame and the n frame. 17 is the area of the moving object in the n frame and the n+ frame.
The motion area when comparing images of one frame [18 is n-
This is the AND of the moving object area 13 of one frame and n+, the moving object area 15 of one frame.

First, the input terminal &2 of this device stores inter-frame image signals even if temporally continuous images are inputted to the frame memory 2i; When n frames of images 11 are directly input to the difference/threshold processing means 1 from the input terminal, n-1 frame images 10 are input from the frame memory 2. Difference/threshold processing means 1i; kn frame image 1
The difference between the signals of 1 and n-1 frame images 10 is taken, and the area where the difference exceeds the threshold value is output as a motion area.

As shown in FIG. 2(a) 16, this is the logical sum of the moving object area 13 of frame n-1 and the moving object area 14 of frame n. When an n+1 frame image 12 is input to the direct deflection difference/threshold processing means 1 from an input terminal, an n frame image 11 is input from the frame memory 2. Difference/
Threshold processing means 1 takes the difference between the signals of the n11 frame image 12 and the n frame image 11, and outputs the area where the difference exceeds the threshold as a motion area. This is shown in FIG. 2(b). As shown in 17, the number ζ is the logical sum of the moving object area 14 of n frames and the moving object area 15 of n frames.

When this output is input to the direct sputum logical product calculation means 3,
Even if the output of one frame before is inputted from the frame memory 2a, the AND calculation means 3! ;t,, a motion region 16 when comparing images of n-1 frame and n frame, and a motion region 17 when comparing images of n frame and n+1 frame.
The logical product of the moving object area 14 of the n-frame and the area 14 of the moving object of the n+1 frame is obtained by logical conversion. The logical sum of the moving object area 15 of the moving object area 15 and the moving object area 14 of n frames shows that the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is convex, and the shape of the moving object area is a convex shape, and Assuming that the moving direction of the moving object in the frame does not change, the moving object region 13 of the n-1 frame and the moving object region 1 of the n+1 frame
The logical product 18 of 5 is always included in the moving object area 14 of n frames. Therefore, even if the logical product calculation means 3 outputs exactly only the moving object area 14 of n frames, it can be easily calculated using a logical formula. Supplement the explanation. Assuming the output of the AND operation means 3 as the area 14a, Condition 1 Area 1 Condition 2 Area 1 Condition 3 Area 1 Condition 4 Area 1 4a = area 160 area 1 6 = area 13U area 1 7 = area 14U area 1 4 areas 13n area l From proof condition 1.2.3, area 14a = (area 13U area 1 area 15); area 14 days (area 1 From condition 4, 4) n (area 1 30 area 1 area 14a = area 14 From the above According to this embodiment, it is possible to accurately detect only the moving object region 14 of n frames.A second embodiment of the present invention will be described below with reference to the drawings, but FIG. This is a block diagram of the image motion detection device in the second embodiment. In FIG. 3, the same elements as in FIG. 1 are given the same numbers (2b is a frame memory, 7 is a switch, 8 and 8a are memories, and 9 is a reset signal input terminal. First, when temporally continuous image signals are input to the input terminal 5 shown in FIG. 3, the frame memory 2 stores the image signals between one frame and the difference. - Threshold processing means 1 takes the difference between the current frame image and the signal of the previous frame image from frame memory 2, and outputs the area where the difference exceeds the threshold value as a motion area. The processing means 1a calculates the difference between the current frame image and the signal of the image from the frame memory 2b, and outputs the area where the difference exceeds the threshold as a motion area to an output terminal 6.
Output to.

The logical product operation means 3ζ calculates the logical product of the motion regions output from the difference/threshold processing means 1 and the difference/threshold processing means 1a, and outputs it as the region of the moving object. According to the output of the product calculation means 3, the input image from the memory 8 is output in the area other than the area of the moving object, and the image from the ζ friend memory 8a is output to the frame memory 2b in the area of the moving object.A The memory 8 is an input terminal. The memory 8a stores the input image signal from the frame memory 2b.

In addition, when a reset signal is input from the switch 7 CL reset signal input terminal 9, the input image from the memory 8 is output regardless of the output of the AND calculation means 3.The frame memory 2b, the memory 8, 8
To further explain the operation of a, in the moving object area, the switch 7 outputs the image signal from the memory 8a to the frame memory 2b.This is the signal stored in the frame memory 2b. In an area other than the moving object area, the input image signal from the input terminal 5 is stored in the frame memory 2b, and in the moving object area, the signal previously stored in the frame memory 2b is retained, but the image is stored in the memory 8.8aζ. Difference/threshold processing means l, 1 for signal strength of a certain pixel
a. The area of the moving object is processed by the AND operation means 3, and the result is input to the switch 7. The input signal and the output signal of the frame memory 2b are stored in this image motion detection device. The operation will be explained using Fig. 5 of the fourth plate. Fig. 4 is the image input to this motion detection device. The first image input in the initial state is taken as frame 0.
, up to 5 frames are shown. In the image, a moving object 19 is moving at a constant speed in the direction of an arrow 20.

Figure 5 (1) shows the processing process for the input image in Figure 4. The output (c) of the value processing means 1a is the output fi (
d) is an image stored in the frame memory 2b.4 First, when the O frame is input, the difference/threshold processing means 1, 1a, (a), (b)
Then, nothing is output (- Therefore, the AND operation means 3
(C) But nothing is output L% Therefore, when the 0 frame image is stored as it is in the frame memory 2b (d) and 1 frame is input next, the difference/threshold processing means 1. Ia, (a), (b)
At any rate, the logical sum of the moving object areas of frame 0 and frame 1 is output. Therefore, the logical product calculation means 3 (
Even if the logical sum of the moving object areas of frame 0 and frame 1 is output in c), the previous signal is retained in the area ζ of the logical sum of the moving object areas of frame 0 and frame 1 in frame memory 2b(d). Therefore, even though the moving object image of O frames is stored as is in the frame memory 2b, when two frames are inputted next, the difference/threshold processing means 1 (a) uses frames 1 and 2.
Even if the logical sum of the moving object area of the frame is output, the difference/
The threshold processing means 1a (b) outputs the logical OR of the moving object area of frame 0 and frame 2. Therefore, the output (C) of the logical product calculation means 3 outputs the moving object area of frame 2 and the area of moving object of frame 0. The logical sum of and the logical product of the moving object area of one frame is output a frame memory 2b
In (d), ζ Since 2 frames of images are input except for the area output by the logical product operation means 3, the area is the logical product of the θ frame and the area of the moving object in 1 frame. When 3 frames are input next even though the image of
Even if the logical sum of the moving object area of the frame is output, the difference/
The threshold processing means 1a(b) calculates I-! , The logical product of the moving object area of frame 0 and frame 1, and the logical sum of the moving object area of frame 3 are output. Therefore, the output (c) of the AND calculation means 3 is the logical sum of the moving object area of 1st and 3rd frames and the logical product of the moving object areas of O frame, 1st frame, and 2nd frame. In the frame memory 2b (d), images of 3 frames are input manually except for the area output by the logical product calculation means 3, so only the area of the logical product of the moving object areas of frame 0, frame 1, and frame 2, A 0-frame moving object image is stored.

Next, when 4 frames are input, difference/threshold processing means 1 (a) has 3 frames and 4 frames.
The logical OR of the moving object area of the frame is output. Difference/
The threshold processing means 1a(b) outputs the logical product of the moving object area of two frames, O frame and 1 frame, and the logical sum of the moving object area of 4 frames. Therefore, the output (C) of the AND calculation means 3 is the logical sum of the moving object area of CL4 frame and the logical product of the moving object area of 0 frame, 1 frame, 2 frame, and 3 frame. As shown in Fig. 5, when the logical product of the moving object area of the O frame and the 3rd frame becomes an empty set, even if only the moving object area of the 4th frame is detected accurately, the logical product calculation means in the frame memory 2b (d) Since the image of 4 frames is input except for the area output in step 3, it is the area of the logical product of the moving object area of frame 0, frame 1, frame 2, and frame 3 (the image of the moving object of frame 0) If the logical product of the moving object regions of frame 0 and frame 3 becomes an empty set, the background image from which the moving object has been removed is stored.

Next, when 5 frames are input, difference/threshold processing means 1 (a) has 4 frames and 5 frames.
Even if the logical sum of the moving object area of the frame is output, the frame memory 2b outputs the background image from which the moving object has been removed. The area is output accurately.

Therefore, in the output (c) of the AND calculation means 3, only the area of the moving object in 5 frames is accurately detected.

In the frame memory 2b(d), since the previous signal is retained in the area output as the moving object area by the AND operation means 3, 1'=
Only the background image with moving objects removed is stored and retained.

If only the background image from which moving objects have been removed is stored in the L frame memory 2b (d) in the subsequent frames, then in the next frame, the difference/threshold processing means 1a (b) will calculate the moving objects in that frame. Even if the area is output accurately, the difference/threshold processing means 1 (a) outputs the logical sum of the area of the moving object in that frame and the previous frame. Even if only the area of the moving object in that frame is detected accurately, in the frame memory 2b(d), the area outputted as the area of the moving object by the AND calculation means 3 retains the previous signal, so Only the background image with objects removed is stored and retained.

Therefore, by repeating this operation, the frame memory 2b (d) retains the background image from which the moving object has been removed, and the output (C) of the logical product calculation means 3 retains only the area of the moving object in that frame. Accurately detected.

Second embodiment of the present invention Frame memory 2b Strength 1
If the configuration is such that data writing can be controlled on a pixel-by-pixel basis and the previous data remains when writing is stopped, switch 7 will stop writing when there is an output from the motion area.
When there is no output of the motion area, only the data from memory 8 needs to be written. Therefore, in this case, memory 8a
Also, as is clear from the explanation of the second embodiment, there is no need for frame memory 2b in FIG. 3 of the second embodiment.
By extracting the output of A block diagram of a background extraction device according to a third embodiment of the invention is shown in FIG.

The operation of each element inside this third embodiment of the present invention is completely the same as that of the second embodiment of the present invention. However, from the output terminal 6, the output of the frame memory 2b in FIG. 6 is output as a background image.

Next, a background extraction device according to a fourth embodiment of the present invention will be explained. FIG. 7 is a block diagram of a background extraction device according to a fourth embodiment of the present invention. In FIG. 7, 200 is a first background extraction means, 201 is a second background extraction ratio, and 202 is a switch. 203 is an input terminal, and 204 is an output terminal. Ascaris First background extraction means 200 and second background extraction means 2
02 and 4 respectively have the same configuration as the background extraction device of the third embodiment of the present invention.
Even if a video signal is input at a speed of 30 frames/sec.
The first background extraction means 200 performs the same operation as the background extraction device of the third embodiment of the present invention (\ input image signal outputs the background image from which moving objects have been removed to the switch 202. From the 30 frames/second image signal, only one frame image is transferred to the second image at 10 second intervals.
It is output to the background extraction means 201. Second background extraction means 20
1ζ Performs the same operation as the background extraction device according to the third embodiment of the present invention (＼Outputs a background image obtained by removing moving objects from the image signal input at 10 second intervals to the output terminal 204)4 This device number For example, if a moving object temporarily stands still relative to the background and does not stand still for at least 10 seconds, it is possible to output a background image with the moving object removed. If you do not stand still for at least 10 seconds, a background image with moving objects removed will be output.It is obvious that you can freely set this time. Although it is easily possible to construct the hojo in three or more stages, the background extracting device according to the fifth embodiment of the present invention will be explained below. In FIG. 8, which is a block diagram of the background extraction device of the embodiment, the same elements as in FIG. 6 include:
Also, in Figure 8, 210 is the switch control means, 211.212.213.214.
．． 215 and 216 are switches, 203 is an input terminal,
204 is an output terminal, 2c, 2d, 2e, 2
f is a frame memory.

From this device ζ To the two background extraction means 200 and 201 of the background extraction device according to the fourth embodiment of the present invention Frame memory 2.
Even if the parts except 2b are shared, the switch control means 210 and the switch 2
11, 212, 213, 2, 4, 215, and 216 are provided. First, a moving image signal is input from the input terminal 203 at a rate of 30 frames/second. At this time, the switch control means controls the switch 211 to output the input signal from the switch 203. Also switches 212, 213, 214
, 215, to allow the signal path passing through the frame memories 2c and 2d to pass, and also controls the switch 216 to stop the output. At this time, this device is the fourth embodiment of the present invention.
The switch control means 41 performs the same operation as the first background extraction means 200 of the background extraction apparatus of the embodiment and outputs a background image from which moving objects are removed from the input image signals t, X, to the switches 216 and 211 every 10 seconds. You can control the switch 211 and output the signal from 215 only for the frame, or switch 212, 213.
214.215, control 'aL,, frame memory 2e,
The signal path passing through 2f can also be passed through switch 2.
16 and outputs the signal from switch 215 to terminal 2.
At this time, this device performs the same operation as the second background extracting means 201 of the background extracting device according to the fourth embodiment of the present invention. Furthermore, the background image from which moving objects have been removed is output to the output terminal 204.The power of this device is the skin of one frame that performs the same operation as the second background extraction means 201 of the background extraction device of the fourth embodiment of the present invention. The same operation as the first background extraction means 200 of the background extraction device of the fourth embodiment is suspended. There is no practical problem.

Although the present device has a two-stage background extraction means, it is also easy to imagine that it could be constructed with three or more stages. Embodiment 4 East A force that shares the parts of two background extraction means except for the frame memory (similarly, 3
Regarding a device configured with background extracting means of more than one stage, it is easy to share the parts of the background extracting means except for the frame memory. Examples will be described with reference to the drawings. FIG. 9 is a block diagram of an image residual object detection device according to a sixth embodiment of the present invention. In FIG. 9, 30 is a frame memory, 31 is a difference/threshold processing ratio, 32 is a switch, 33 is a stationary background extraction means.

FIG. 10 is a route diagram for explaining the operation of the image residual object detection device in the sixth embodiment.4First, +;! is stored in the frame memory 30. , Jio A background image in a normal state shown in Figure (a) is stored. ζ to the static background extraction means 33 as shown in FIGS. 10(b) and 10(c) Moving objects 34, 3
4. A moving image including the artifact 35 is input. The still background extraction means 33 performs the same operation as the background extraction device of the fifth embodiment of the present invention, and extracts a still background image from which moving objects are removed from the input moving image, as shown in FIG. 10(d). The difference/threshold processing means 31 extracts only the normal background image (FIG. 10(a)) read out from the frame memory 30, as shown in FIG. The switch 32 calculates the difference from the still background image shown in FIG. From the still background image outputted from the still background extracting means 33 shown in f) in FIG. and output. By the above-described operation, only the leftover object 35 can be detected from the input moving image including the moving object 34, 34 and the leftover object 35.

This could also be applied to security systems, such as detecting dangerous objects left behind in public places.

Note that in the first, second, third, fourth, fifth, and sixth embodiments, the three input images are composed of temporally consecutive frames as moving images. There is no need for it to be visually recognized as a moving image like a television signal (for example,
It is also possible to operate each of the embodiments using images that are outputted from an imaging device at a rate of 30 frames/second, sample only one frame per second, and increase the interval of 1 second between each frame. be. Also, by doing this,
In the first, second, third, fourth, and fifth embodiments, the amount of calculation processing required per unit time of the apparatus of each embodiment can be reduced.
Difference/Threshold Processing Means 1, 1 a U This takes the difference between the signals of two input frames of images, and outputs the area where the difference exceeds the threshold as a motion area, and the output is binary. This does not necessarily have to be a binary value. It calculates and outputs L, which is a problem to achieve the purpose of the present invention (
In addition, in the first, second, third, fourth, and fifth embodiments, the logical product calculation means 3 outputs the logical product of the two input motion areas.For example, as described above, The difference/threshold processing means 1, 1a, instead of thresholding the difference between the signals of two frames of images.
Based on the magnitude of the difference, an evaluation function indicating the possibility of a motion region is calculated and outputted.Instead of the logical product calculation means 3 calculating the logical product of two motion regions, it calculates the possibility of two motion regions. Assuming that the product of the evaluation functions shown is L, t,
In addition, it is easy to imagine variations of such similar processing methods, and there are slight differences in performance depending on the processing method. This makes it possible to realize an excellent image motion detection device that can accurately detect the area of a moving object. Similarly, it is possible to realize a background extraction device that extracts a background image from which moving objects are removed from an input image. Furthermore, it is possible to realize an excellent leftover object detection device that can detect only leftover objects from an input image that includes not only leftover objects but also moving objects.

[Brief explanation of drawings]

FIG. 1 is a block diagram of the image motion detection device according to the first embodiment of the present invention. FIG. 2 is a route diagram for explaining the operation of the image motion detection device according to the first embodiment of the present invention.
FIG. 5 is a route diagram for explaining the operation of the image motion detection device in the second embodiment of the present invention.
FIG. 6 is a block diagram of an image background extraction device according to a third embodiment of the present invention. FIG. 7 is a block diagram of an image background extraction device according to a fourth embodiment of the present invention. FIG. 9 is a block diagram of the image background extraction device according to the fifth embodiment of the present invention; FIG.
The diagram is a diagram for explaining the operation of the residual object detection device according to the sixth embodiment of the present invention. FIG. 11 is a block diagram of a conventional image motion detection device. FIG. 12 is a conventional image motion detection device. This is a route map for explaining the operation of the device.
...Difference/threshold processing ratio 2, 2a, 2b,
2c, 2d, 2a, 2f-frame memory, 3... logical product calculation means, 5... input terminal, 6
...Output terminal, 7...Switch, 8, 8a...
・Memory, 10...n-1 frame image 11-n frame image 12...n+1 frame image 13...
- Moving object area of n-1 frame 14... Moving object area of n frame [15... Moving object area of n+1 frame 16... When comparing n-1 frame image and n frame image Motion area 17...Motion area 18 when comparing n frame image and n+1 frame image...
- Logical product of the area of the moving object in the n-1 frame and the area of the moving object in the n+1 frame 3o...Frame memory,
31... Difference/threshold processing half-recessed 32... Switch, 33... Still background extraction means, 34a, 34b
... Movement risk 35... Remaining track 36... Output area of difference/threshold processing means 31 100... Frame memory, 101... Difference/threshold processing means, 1
02... Motion area when comparing the previous frame image and current frame image @ 103... Area of moving object in previous frame* 104... Area of moving object in current frame 20
0...First background image extraction means, 201...Second background image extraction means, 202...Switch, 203...
Input terminal, 204... Output terminal, 210... Switch control half-recessed 211, 212, 213, 214, 21
5, 216...Switch, name of agent Patent attorney Shigetaka Awano and 1 other person

Claims (8)

[Claims] (1) Means for comparing at least two or more frames of images and detecting areas with different signal values, a memory for storing the detection output for at least one frame or more, and an output read from the detection output and the memory. An image motion detection device characterized by comprising means for determining and outputting a motion area. (2) Based on the detection output and the output read from the memory,
2. The image motion detection apparatus according to claim 1, wherein the means for determining and outputting a motion region calculates a logical product of two outputs, determines this as a motion region, and outputs the result. (3) a plurality of means for comparing at least two or more frames of images and detecting regions with different signal values; a means for determining and outputting a motion region based on detection outputs of the plurality of means; and a means for determining and outputting a motion region; has a means for synthesizing one frame image from at least two frame images, using the output of the at least two frames as a control signal, and comparing the synthesized one frame image with the at least two or more frames of the signal. An image motion detection device characterized in that one frame image is used for comparison of at least one of a plurality of means for detecting different regions. (4) The means for determining and outputting a motion region based on the detection outputs of the plurality of means calculates a logical product of the plurality of outputs, determines this as a motion region, and outputs it. The described image motion detection device. (5) The means for synthesizing one frame image from at least two frame images by using the output of the motion area as a control signal, uses the output of the movement area as a binary control signal, and according to the value of the control signal. 4. The image motion detection apparatus according to claim 3, wherein the image of two frames are switched and combined into one frame of image. (6) a plurality of means for comparing at least two or more frames of images and detecting regions with different signal values; a means for determining and outputting a motion region based on detection outputs of the plurality of means; and a means for determining and outputting a motion region; has means for synthesizing one frame image from at least two frame images and outputting the synthesized one frame image as a background image, using the output of the at least two frames as a control signal, and converting the synthesized one frame image into the at least two or more frames. 1. A background image extraction device for an image, characterized in that one frame image is used for comparison of at least one of a plurality of means for comparing and detecting different areas of signals. (7) The means for synthesizing one frame image from at least two frame images by using the output of the motion area as a control signal, uses the output of the movement area as a binary control signal, and according to the value of the control signal. 7. The background image extracting apparatus for images according to claim 6, wherein the images of two frames are switched and combined into one frame image. (8) Background image extraction means for extracting a background image from which moving objects have been removed from an input image; a frame memory for storing at least one frame image; and a comparison between the output of the background image extraction means and the frame memory. , a residual object detection device characterized by having means for detecting areas where signal values differ.