JP3185362B2 - Moving object detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving object detecting apparatus for detecting a moving object from continuously input moving images.

[0002]

2. Description of the Related Art As a method of detecting a moving object from a moving image, a method of extracting a change by taking a difference between an input image and an image input a predetermined period (eg, one frame period), There is a method in which a background image that does not include a moving object is stored in the apparatus in advance, and a moving object is extracted by calculating a difference from an input image that is continuously input.

[0003] However, in the technique of obtaining a difference between an input image and a previously input image, since only a change due to time differentiation is extracted, it is difficult to detect an image of a moving object only from the value. . For example, when a plain white circular area moves in front of a uniform black background, two crescent-shaped areas are obtained at the front and rear in the traveling direction. It is difficult to extract a region.

The moving object detection method based on the difference between an input image and a background image that does not include a moving object stored in advance has a problem when the camera is stationary or the background does not change. No, but the camera moves
In the case of panning, or in the case where lighting changes due to the sun being hidden by clouds outdoors or the like, the background image changes every moment, so that correct results cannot be obtained. In order to cope with such a case, a method of synthesizing a background image from an input image has been proposed.
No. 0 describes a background image generation method by adding an input image and a background image at that time at a predetermined ratio.

[0005]

However, in the above-described method, it is necessary to appropriately determine the mixing ratio between the input image and the background image. In general, the time constant at which the background image is updated is determined by changing the time constant of the background image. The speed must be set sufficiently higher than the change speed and sufficiently lower than the moving speed of the target. Therefore,
Various conditions such as the moving speed of the object and the changing speed of the background are constant,
There is no problem if it is known in advance, but when these conditions are not constant, for example, when the background changes rapidly, the background part is erroneously detected as a target, or conversely, the speed of the target decreases. In such a case, an erroneous operation occurs such that the target is not detected. As described above, when various conditions such as the moving speed of the target and the changing speed of the background are not constant, it is extremely difficult to uniquely set each constant, and there is a problem that the number of malfunctions increases.

In view of the foregoing, it is an object of the present invention to provide a moving object detecting apparatus which is less likely to malfunction even when the moving speed of the object or the changing speed of the background changes.

[0007]

According to the present invention, there is provided an image input means, a background image storage means, a background image generating means, and a moving body area detection for extracting a moving body area by taking a difference between the background image and the input image. Means and speed and size of moving object area are detected
The moving object detecting apparatus includes a moving body speed / size detecting means for performing the calculation and a coefficient determining means for determining a ratio of addition of the input image and the background image in the background generating means.

[0008]

[0009]

[0010]

According to the present invention, the moving object region is extracted by the difference between the input image and the background image, and the moving object speed and the moving object speed are calculated.
The speed of the moving body region obtained by the size detecting means and
By using the size , the coefficient determining means determines the addition ratio of the input image and the background image in the background generating means, and generates the next background image using the coefficient, so that
The background image is generated by mixing the input image and the background image at a mixing ratio according to the characteristics of the detected moving body region.

[0011]

[0012]

[0013]

FIG. 1 is a block diagram showing a moving object detecting apparatus according to a first embodiment of the present invention. In FIG. 1, reference numeral 101 denotes an image input unit; 113, a frame interval setting unit for setting a frame interval for inputting an image;
3 is a multiplication unit, 104 is an addition unit, 105 is a background storage unit that stores the generated background image, and 106 is an image input unit 101
A difference unit 107 for calculating a difference between an input image from the image processing unit and a background image from the background storage unit 105; 107, a time axis filter processing unit for absorbing a change in the time axis direction from the result obtained by the difference unit 106 Denotes an area detection unit for detecting an area from the result, and 109 denotes an image display unit. Also, 110
Is a speed / size detection unit for detecting the speed and size of the detected moving body region; 111 is a coefficient determination unit for determining coefficients in the multiplication units 102 and 103 based on the detected speed and size; Is a coefficient storage unit for storing. The operation of the moving object detection device according to the first embodiment configured as described above will be described below.

In the image input unit 101, the frame interval s set in advance by the frame interval setting unit 113
Then, for example, the input video signal from the camera is converted into digital data and supplied to the multiplication unit 102, the difference unit 106, and the image display unit 109. The multiplication unit 102 multiplies each pixel of the input signal by the coefficient stored in the coefficient storage unit 112 and inputs the result to the addition unit 104. On the other hand, the multiplication unit 103 multiplies each pixel of the background image stored in the background storage unit 105 by the coefficient stored in the coefficient storage unit 112 and inputs the result to the addition unit 104. The adding unit 104 adds the values obtained by the multiplying unit 102 and the multiplying unit 103, and stores the sum at a position corresponding to each pixel in the background storage unit 105. Therefore, the coefficient of the multiplication unit 102 stored in the coefficient storage unit 112 is a (0 <a <1), and the coefficient of the multiplication unit 103 is (1−a).
Then, based on the input signal (X in the figure) and the contents of the background storage unit 105 (Y in the figure), the output of the adding unit 104 (Z in the figure) is given by: Z = a × X + (1-a) × Y (1) is obtained. That is, each pixel value in the background storage unit 105 is updated by Expression (1). For example, a = 0
In the case of, the expression (1) becomes Z = Y, and the background storage unit 105
Does not change at all. Conversely, when a = 1, the expression (1) becomes Z = X, so that the content of the background storage unit 105 is always rewritten to the input image.

The difference unit 106 calculates the absolute value of the difference between each pixel between the input image from the image input unit 101 and the background image from the background storage unit 105. Decrease. FIG. 2 shows an example of the configuration of the time axis filter processing section 107.

Here, the operation of the time axis filter processing section 107 will be described with reference to FIG. The result obtained by the difference unit 106 is input to the multiplication unit 201, and the multiplication unit 201 multiplies the coefficient stored in the coefficient storage unit 203 and supplies the result to the addition unit 204. On the other hand, the multiplication unit 202 multiplies the content stored in the result storage unit 205 by the coefficient stored in the coefficient storage unit 203 and supplies the result to the addition unit 204. The addition unit 204 adds the results of the multiplication units 201 and 202 and stores the result in the result storage unit 205.
Here, the coefficient stored in the coefficient storage unit 203 determines the time constant of the time axis filter, and the coefficient determined in advance may be stored, or may be changed according to the situation. You may be able to. Thus, the detection result averaged in the time axis direction is generated in the result storage unit 205.

The area detection unit 108 detects the area of the moving object based on the result of the time axis filter processing unit 107. FIG. 3 shows an example of the configuration of the area detection unit 108. Hereinafter, the operation of the area detection unit 108 will be described with reference to FIG.
The detection result obtained by the time axis filter processing unit 107 is first smoothed by the smoothing filter unit 301, and a result that is somewhat discrete is collected. As the smoothing filter 301, a normal low-pass filter may be used. However, when the degree of smoothing is increased, the regions are easily united, but distant portions are easily connected.
It is necessary to select in advance those with appropriate characteristics. After the smoothing filtering process, binarization is performed by a binarization processing unit 302 using a predetermined threshold. This threshold value is set, for example, by selecting from several types at the time of initial setting. The labeling processing unit 303
The quantified areas are numbered so that the distant areas can be treated as different ones. However, the labeling processing may be performed by a known method, and thus the description is omitted here. The result obtained by numbering the moving object area obtained as described above is displayed on the image display unit 109 and the speed / size detection unit 11.
0 is given.

In the image display unit 109, the image input unit 101
Is combined with the result of the area detection unit 108, for example, the brightness of the pixel at the position corresponding to the portion where the result of the area detection unit 108 is 0 in the input image is reduced, and the other pixels are To display the detection result.

The speed / magnitude detecting unit 110 calculates the moving speed v of the moving object on the screen and the size w in the moving direction of the moving object based on the frame interval s set by the frame interval setting unit 113 and the result of the area detecting unit 108. Is detected. FIG. 4 shows an example of the configuration of the speed / size detection unit 110. The result of the area detection unit 108 is supplied to the center-of-gravity position detection unit 401, and the center of gravity of the detected area is detected. A barycentric position storage unit 402 stores the barycentric position detected in the immediately preceding image, and stores the barycentric position in the previous image detected by the barycentric position detecting unit 401. The speed detection unit 403 detects the speed v of the moving object from the center of gravity position detected by the center of gravity position detection unit 401 and the previous center of gravity position stored by the center of gravity position storage unit 402. Further, the size detection unit 404 detects the size w of the moving object from the center of gravity position detected by the center of gravity position detection unit 401 and the previous center of gravity position stored by the center of gravity position storage unit 402.

Here, the operation of the speed detector 403 and the size detector 404 will be described with reference to FIG. s
Considering an example in which an image is input once in a frame, the result of the area detection unit 108 in the n-th frame is shown in FIG.
If it becomes as shown in (a), the center of gravity position detection unit 401
, The coordinates (x ₁ , y ₁ ) of the center of gravity G ₁ of the area 501 are detected. If the result of the area detection in the (n + s) -th frame is as shown in FIG. 5B, the coordinates (x ₁ , y ₁ ) of G ₁ are stored in the center-of-gravity position storage unit 402, and the center-of-gravity position detection is performed. in section 401, the center of gravity G ₂ of the coordinate region 502 (x _2,
y ₂ ) is detected. Using these results and the frame interval s given by the frame interval setting unit 113, the speed detection unit 403 uses the moving speed v (pixels / s frame).
Is determined by equation (2).

V = √ (| x₁-X_Two|^Two+ | Y₁-Y_Two|^Two(2) Next, FIG. 5C shows the operation of the size detector 404.
It will be described using FIG. The size detection unit 404 detects the center of gravity position.
Center of gravity G detected by protrusion 401_TwoAnd center of gravity storage
Center of gravity G stored in 402₁Of the moving object using
The size w in the direction is obtained. That is, for each detection area
Center of gravity G₁And G_TwoG on a straight line 503 connecting ₁To G_TwoToward
The existence of the region 502 in the direction perpendicular to the straight line 503
After detecting the area 502 first, the area 5 is finally detected.
Calculate the distance to detect 02 and move the moving object
Of size w. Center of gravity G of each detection area₁And G_Two
Is expressed by equation (3), and on the straight line 503
G₁To G_TwoThe coordinates of the point to be examined toward (x_Three, Y_Three)When
Then, x_ThreeAnd y_ThreeSatisfies equation (4). Also, this point
The straight line that is perpendicular to the straight line 503 is expressed by equation (5).
Then, on this straight line, the coordinates of the point to check for the existence of the detection area
(X_Four, Y_Four), X_FourAnd y_FourSatisfies equation (6).

(X ₁ −x ₂ ) (y−y ₁ ) = (y ₁ −y ₂ ) (x−x ₁ ) (3) (x ₁ −x ₂ ) (y ₃ −y ₁ ) = (y ₁₋ y ₂ ) (x ₃ −x ₁ ) (4) (y ₁ −y ₂ ) (y−y ₃ ) = (x ₁ −x ₂ ) (x−x ₃ ) (5) (y ₁ −y ₂ ) (y ₄ −y ₃ ) = (x ₁ −x ₂ ) (x ₄ −x ₃ ) (6) In this manner, the straight line 503 is perpendicular to the straight line 503 from G ₁ toward G _2. The area 502 is first detected at a point 505 on the straight line 504, and finally the straight line 5 is detected.
If it is detected at a point 507 on the line 06, the coordinates of the intersections 508 and 509 of the straight line 503 and the straight lines 504 and 506 are (x ₅ , y ₅ ) and (x ₆ , y ₆ ), respectively, and w Is found.

[0023] w = √ In ^{_{(| x 5 -x 6 | 2}} + | | y 5 -y 6 2) (7) the coefficient determining unit 111, detected by the speed and magnitude detector 110, the speed of the moving object The coefficient a is determined from v and the magnitude w of the moving direction of the moving object. Hereinafter, the determination method will be described. The number of frames required for the moving object to move by the size w in the moving direction of the moving object is (w × s /
v) Frame. On the other hand, from equation (1), since the background image is rewritten by (1 / a) operations, s × (1 / a)
Rewrite with frame. Therefore, the coefficient a for which the background image is rewritten in the time required for the moving object to move by the size of the moving direction is as follows: a = v / w (8) The coefficient a thus determined and (1-a)
Is stored in the coefficient storage unit 112, and the multiplication units 102 and 103
Is the coefficient of multiplication in. Therefore, the input image and the background image are always mixed at such a rate that the background image is rewritten in a time required for the detected moving object to move by the size of the moving direction, so that the background image is automatically set. Will be.

As described above, according to the present embodiment, the background image is generated by changing the mixture ratio between the input image and the background image based on the moving speed of the detected area. Background generation can be performed. In addition, since time axis filter processing is performed when detecting a moving object,
Temporal fluctuation of the detection area due to noise, disturbance, or the like can be suppressed, and stable detection can be performed.

FIG. 6 is a block diagram showing a moving object detecting apparatus according to a second embodiment of the present invention. 6, the same components as those in FIG. 1 are denoted by the same reference numerals. 601 is an area setting unit for setting a detection area, 602 is a position detection unit for detecting the position of an input pixel in an image,
Reference numeral 603 denotes an area determining unit that determines which area the input pixel belongs to, 605 denotes a coefficient determining unit that determines coefficients of the multiplying units 102 and 103, 604 denotes a coefficient storage unit that stores the coefficient for each area, and 606 denotes a coefficient storing unit that stores the coefficient for each area. Adder for adding the difference result, 60
Reference numeral 7 denotes a change amount storage unit for storing the addition result, and 608, an image display unit for displaying the result. The operation of the moving object detection device according to the second embodiment configured as described above will be described below.

The area setting unit 601 divides a screen into a plurality of processing units. For example, as shown in FIG. 7, if the input image is 640 pixels in the horizontal direction (x direction) and 480 pixels in the vertical direction (y direction), and the horizontal direction is divided into 10 and the vertical direction is divided into 5, one area size is obtained. Is 64 (pixels) x 96 (pixels)
Becomes

The output of the image input unit 101 is supplied to a multiplication unit 102
, A difference unit 106, a position detection unit 602, and an image display unit 608.
Supplied to The position detection unit 602 detects the position of the input pixel on the screen, that is, the coordinate value of the input pixel. For example, the position in the vertical direction can be realized by counting up the counter reset by the vertical synchronizing signal with the horizontal synchronizing signal, and the counter resetting by the horizontal synchronizing signal with respect to the horizontal position is also input by one pixel. It can be realized by counting up every time.

The area judging section 603 judges which area the input pixel belongs to, and outputs an area number. For example,
As shown in FIG. 7, it is assumed that the input image has 640 pixels in the horizontal direction (x direction) and 480 pixels in the vertical direction (y direction).
When the division and the vertical direction are divided into five, when the position of the input pixel is, for example, x = 300 and y = 200, it is the position indicated by the point 701 and belongs to the area 702. When numbers are assigned to the respective regions in order from the upper left, the region 702 becomes the 25th.

On the other hand, the result of the difference unit 106 is
Is input to The change amount storage unit 607 includes the area determination unit 60
3 for storing the content of the change amount for each area specified by
Send the result to 5 and reset all contents to 0. The change amount of the area (the area to which the input pixel belongs) specified by the area determination section 603 is read from the change amount storage section 607 and added to the addition section 606. In the adding unit 606, the difference unit 1
06 and the change amount of the area to which the input pixel belongs in the change amount storage unit 607 are added, and the change amount storage unit 607 is again added.
Is written to. That is, the difference result is added to the content for the area to which the input pixel belongs, and the total sum of the difference values in each area is finally stored in the change amount storage unit 607.

In the coefficient determining section 605, the change amount storing section 60
The coefficients in the multiplication units 102 and 103 are determined on the basis of the sum of the difference values in each area supplied from 7. Since the region where the sum of differences is large indicates that the motion region is large, the mixture ratio of the input image is increased (the coefficient a is increased), and in the region where the sum of differences is small, the mixture ratio of the background image is increased. (Coefficient a is reduced). FIG. 8 shows an example of the relationship between the sum of the change amounts in each region and the coefficient a. That is, assuming that there are half of the pixels having a change amount of 100 in each region in each region, the total sum of the change amounts in this case is 307200. The value is changed linearly to １1, and is fixed to 1 in the case of more than 〜1.

In the coefficient storage section 604, a coefficient determination section 605
Are stored for each region, and the coefficients of the region to which the input pixel belongs (the region determined by the region determining unit 603) are supplied to the multiplying units 102 and 103.

As described above, according to the present embodiment, the input image is divided into a plurality of regions, and the mixing ratio between the input image and the background image in each region is changed based on the difference result for each region. Since the background image is generated, highly accurate background generation can be performed with a simple configuration.

In the present embodiment, the coefficient determining section linearly determines the value below a predetermined value and sets a constant value above the predetermined value. However, the present invention is not limited to this. Alternatively, an appropriate conversion table may be created in advance, and the conversion may be performed using the table.

In the first and second embodiments, the threshold value of the difference portion, the threshold value of the number of pixels of the background change judging portion, and the threshold value of the rate of increase in the number of pixels are fixed values. It may be changed according to the situation.

[0035]

As described above, according to the present invention,
Even when the moving speed of the target or the changing speed of the background changes, erroneous detection can be reduced, and the practical effect is large.

[Brief description of the drawings]

FIG. 1 is a block diagram of a moving object detection device according to a first embodiment of the present invention.

FIG. 2 is a block diagram illustrating an example of a configuration of a time axis filter unit in FIG. 1;

FIG. 3 is a block diagram illustrating an example of a configuration of an area detection unit in FIG. 1;

FIG. 4 is a block diagram showing an example of a configuration of a speed / size detection unit in FIG. 1;

FIG. 5 is an explanatory diagram for explaining an operation of a speed / magnitude detection unit in FIG. 1;

FIG. 6 is a block diagram of a moving object detection device according to a second embodiment of the present invention.

FIG. 7 is an explanatory diagram for explaining the operation of FIG. 6;

FIG. 8 is an explanatory diagram for explaining an operation of a coefficient determination unit in FIG. 6;

[Explanation of symbols]

 Reference Signs List 101 Image input unit 102, 103 Multiplication unit 104, 605 Addition unit 105 Background storage unit 106 Difference unit 107 Time axis filter processing unit 108 Area extraction unit 109, 608 Image display unit 110 Speed / size detection unit 111, 605 Coefficient determination unit 112, 604 Coefficient storage unit 113 Frame interval setting unit 601 Area setting unit 602 Position detection unit 603 Area judgment unit 607 Change amount storage unit

Continuation of the front page (56) References JP-A-1-211083 (JP, A) JP-A-2-71380 (JP, A) JP-A-60-168276 (JP, A) (58) Fields studied (Int .Cl. ⁷ , DB name) G06T ^7/ 00-7/60 G06T 1/00 G08B 13/00-15/02 H04N 7/32 H04N 11/04

Claims (2)

(57) [Claims] 1. An image input means, a background image storage means for storing an image serving as a background, and an input image obtained by the image input means and a background image stored in the background image storage means for each pixel. Background image generation means for obtaining a new background image by adding a predetermined ratio to the background image and updating the contents of the background image storage means; and moving the difference between the background image and the input image to extract a moving body region. and body region detection unit, extracts of
Mobile speed to detect the speed and size of the moving mobile area
Degree / size detection means, the speed and the size
Et al., Moving object detection apparatus being characterized in that a coefficient determination means for determining the rate of addition of said background image and said input image in the background generation means. 2. The method according to claim 1, wherein the result of the moving body region detecting means is temporally flat.
Characterized by having a time axis filter processing means for smoothing
The moving object detection device according to claim 1.