KR101631023B1

KR101631023B1 - Neighbor-based intensity correction device, background acquisition device and method thereof

Info

Publication number: KR101631023B1
Application number: KR1020150047497A
Authority: KR
Inventors: 이승룡; 티엔현더
Original assignee: 경희대학교 산학협력단
Priority date: 2015-04-03
Filing date: 2015-04-03
Publication date: 2016-06-15

Abstract

The specification discloses a neighbor-based intensity correction device, a background acquiring device, and a method thereof. The neighbor-based intensity correction device comprises: an initial setting unit which separates a video stream into N series frames, and sets a first background based on a first frame among the series frames; and an acquiring unit which acquires an i^th background for an i^th frame, wherein i is a natural number greater than or equal to 2 and smaller than or equal to N. The acquiring unit includes: a different frame acquiring unit for acquiring different frames on a pixel basis from a first background, an (i-1)^th background, or the i^th frame; an image acquiring unit for acquiring a binary image for the different frames based on a constant threshold; a matrix acquiring unit for acquiring a steady matrix based on the acquired binary image; and a background acquiring unit for determining whether a neighbor-based intensity correction (NIC) algorithm is to be executed by comparing a robustness threshold with a minimum value of the steady matrix, and acquiring the i^th background for the i^th frame based on the determined result. According to the present invention, the background can be acquired at high accuracy by correcting intensities among doubt pixels using the NIC algorithm.

Description

BACKGROUND OF THE INVENTION Field of the Invention [0001] The present invention relates to a neural-based intensity correction device, a background acquisition device,

Background of the Invention Field of the Invention [0001] The present invention relates to a neighbors-based intensity correction apparatus, a background acquisition apparatus and a method thereof, and more particularly, to a neighbors-based intensity correction apparatus, a background acquisition apparatus and a method thereof for acquiring backgrounds of successive series of frames will be.

Recently, as the introduction of surveillance systems in various fields has become active, technological developments for improving the performance of surveillance systems are continuously being carried out.

Surveillance applications based on surveillance applications for human surveillance or traffic flow analysis can detect various objects such as people and vehicles from photographs, and can detect various objects from a video stream or a series of successive frames.

The techniques for detecting an object may include at least one of a background extraction technique, a phase difference frame extraction technique, and an optical flow technique.

The background extracting technique is a technique for detecting an object by extracting a different part between a current frame and a current background image as a most general technique. The different frame extracting technique is a technique for extracting different parts between frames to detect an object, Is a technique for gradually detecting the object by gradually changing the brightness type, thereby calculating the velocity distribution of the motion generated in the image.

Among background art techniques for detecting an object, there are various conventional arts. For example, there is a background extraction technique using a gaussian mixture model (GMM).

The background extraction technique using the Gaussian mixer model can improve the accuracy of object detection by eliminating roughness or noise phenomenon of video, and can detect an object by dividing an area of interest in the background.

On the other hand, the conventional art for extracting the background has a problem that a high calculation amount is required in the background modeling process and an object detection process, and the calculation time is long.

Korean Patent No. 101384627 (Apr. "How to Automatically Split Object Area in Video" Korean Patent No. 101374139 (Apr. "Object monitoring method through video convergence of surveillance system"

An embodiment of the present invention provides a neighbors-based intensity correction apparatus, background acquisition apparatus, and method for acquiring and updating a background for a series of consecutive frames in real time.

One embodiment of the present invention provides a method and apparatus for correcting intensity in a low computational amount by correcting intensity with respect to a dou- ble pixel whose minimum value of the Steady matrix is less than the robustness threshold value from a different frame, A background acquiring device, and a method thereof.

An embodiment of the present invention provides a neighbors-based intensity correction apparatus, a background acquisition apparatus, and a method thereof for acquiring a background with high accuracy by correcting the intensity between dirt pixels using a neighbor-based intensity correction algorithm.

A background acquiring apparatus according to an embodiment of the present invention includes an initial setting unit for dividing a video stream into N series frames and setting a first background based on a first frame in the series of frames, (i is a natural number equal to or greater than 2 and equal to or less than N) acquiring a background, and the obtaining unit obtains a frame from the i < th > An image obtaining unit for obtaining a binary image of the image based on a constant threshold based on the obtained binary image, an image obtaining unit for obtaining a steady matrix based on the obtained binary image, The obtained matrix acquiring unit and the robustness threshold and the minimum value of the Steady matrix are compared And a background obtaining unit for determining whether to perform the neighborhood based intensity correction algorithm (NIC algorithm) and obtaining the i < th > background for the i < th & .

Wherein the background obtaining unit can select the first background or the (i-1) th background in the i < th > background when the minimum value of the Steady matrix is equal to or greater than the robustness threshold value, I < / RTI > background through the neighbor-based intensity correction algorithm if the robustness threshold is less than the robustness threshold.

According to the embodiment, the background obtaining unit may include a filtering unit for filtering at least one dou- ble pixel whose minimum value of the stadi-matrix is smaller than the robustness threshold value from the image frame, A mask generating unit for generating a mask for the i < th > frame and a mask for the i-th background from the pixel, a standard deviation arithmetic unit for calculating a standard deviation for each mask, And an intensity correction unit that corrects the intensity between the filtered douter pixels based on the deviation and acquires the i-th background for the i-th frame using the dou- ble-corrected dou- ble pixels.

The obtaining unit may further include an updating unit updating the obtained i-th background in the current background, and the different-frame obtaining unit may obtain a different frame for the updated current background and the (i + 1) -th frame.

According to an embodiment, the image obtaining unit may obtain the binary image based on the consistency threshold value that distinguishes the background area and the object area.

A neighbors-based intensity correction apparatus in accordance with an embodiment of the present invention includes a first background or an i-1 background in a series of N frames and at least one douter pixel based on a frame for an i- (I is a natural number equal to or greater than 2 and equal to or less than N), a mask for the first background or the i-1 th background from the filtered douter pixel, and a mask for generating a mask for the i- A standard deviation arithmetic unit for calculating a standard deviation for each of the masks, and a correction unit for correcting the intensity between the filtered douter pixels based on the calculated standard deviation, I < th > frame.

Wherein the filtering unit obtains a binary image for the frame based on a constant threshold value and acquires a Steady matrix based on the obtained binary image, wherein the minimum value of the Steady matrix is the at least You can filter on one douter pixel.

A background acquisition method according to an embodiment of the present invention includes dividing a video stream into a series of N frames, setting a first background based on a first frame in the series of frames, (I is a natural number equal to or greater than 2 and equal to or less than N), and acquiring the i-th background includes acquiring the first background or the (i-1) Obtaining a binary image for the frame based on the constant threshold, obtaining a Steady matrix based on the obtained binary image, and determining a value of the robustness threshold and the value of the Steady matrix Determining whether to perform the neighbor-based intensity correction algorithm by comparing the minimum values, On the basis of whether the embodiment includes the step of obtaining the i-th background for the i-th frame.

Wherein when performing the neighborhood-based intensity correction algorithm, acquiring the i < th > background comprises filtering the at least one douter pixel with the minimum value of the Steady matrix less than the robustness threshold from the image frame Generating a mask for the first background or the i-th background from the filtered douter pixel and a mask for the i < th > frame, computing a standard deviation for each mask, Correcting the intensity between the filtered douter pixels based on the computed standard deviation, and obtaining the i-th background for the i-th frame using the dou- ble-corrected dou- ble pixels.

A neighbors-based intensity correction method in accordance with an embodiment of the present invention includes a first background or an i-1 background in a series of N frames, and at least one douter pixel based on an image frame for the ith frame Generating a mask for the first background or the i-th background from the filtered douter pixel and a mask for the i < th > frame from the filtered dou- ble pixels, filtering (i is a natural number not less than 2 and not more than N) Calculating a standard deviation for each of the masks, and correcting the intensity between the filtered dummy pixels based on the calculated standard deviation, and correcting the intensity between the filtered dummy pixels based on the calculated standard deviation, And acquiring the i-th background.

Wherein the filtering comprises obtaining a binary image for the frame based on a constant threshold and obtaining a Steady matrix based on the obtained binary image, wherein the minimum value of the Steady matrix is less than a robustness threshold And filtering the at least one dummy pixel.

An embodiment of the present invention can acquire and update the background for a succession of frames in real time.

One embodiment of the present invention provides a method and apparatus for correcting intensity in a low computational amount by correcting intensity with respect to a dou- ble pixel whose minimum value of the Steady matrix is less than the robustness threshold value from a different frame, Can be obtained.

An embodiment of the present invention can acquire background with high accuracy by correcting the intensity between dirt pixels using a neighborhood-based intensity correction algorithm.

1 is a block diagram illustrating a background acquisition apparatus according to an embodiment of the present invention.
2 is a block diagram showing the acquiring unit of FIG.
FIG. 3 is a block diagram showing the background acquiring unit of FIG. 2. FIG.
4 is a flowchart illustrating a background acquisition method according to an embodiment of the present invention.
5 is a flow chart illustrating an i-th background acquisition method for the i-th frame of Fig.
6 is a flow chart illustrating the neighborhood-based intensity correction algorithm of FIG.
7 is a flowchart illustrating a background acquisition method according to another embodiment of the present invention.
8 is a flow chart illustrating the neighborhood-based intensity correction algorithm of FIG.
9 is an example of generating masks for the first background or the (i-1) th background and masks for the i-th frame.
10 is an example of correcting three channels existing between the (i-1) th background and the i-th frame.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings and accompanying drawings, but the present invention is not limited to or limited by the embodiments.

In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear. The terminology used herein is a term used for appropriately expressing an embodiment of the present invention, which may vary depending on the user, the intent of the operator, or the practice of the field to which the present invention belongs. Therefore, the definitions of these terms should be based on the contents throughout this specification.

1 is a block diagram illustrating a background acquisition apparatus according to an embodiment of the present invention.

Referring to FIG. 1, the background acquiring apparatus 10 includes an initial setting unit 100 and an acquiring unit 200.

The initial setting unit 100 separates the video stream into N series of frames and sets the first background based on the first frame in the series of frames.

More specifically, the initialization unit 100 can separate the video stream into N consecutive series of frames, and since there is no initial value for the background, The first background of the unit can be set, and can be expressed by the following [Expression 1].

[Equation 1]

here,

Is a first background in pixel units,

Is a first frame in pixel units, and x and y are pixels.

The acquiring unit 200 acquires the i-th background for the i-th frame. Here, i is a natural number of 2 or more and N or less.

More specifically, the acquiring unit 200 can acquire the background corresponding to each of the N frames in real time. Hereinafter, the acquisition unit 200 will be described in more detail with reference to FIG.

2 is a block diagram showing the acquiring unit of FIG.

2, the obtaining unit 200 includes a phase difference obtaining unit 210, an image obtaining unit 220, a matrix obtaining unit 230, and a background obtaining unit 240.

The different-frame obtaining unit 210 obtains a first background or an (i-1) -th background and a picture frame in units of pixels from the i-th frame.

For example, the different-frame acquiring unit 210 may acquire a frame-by-pixel image from the first background and the second frame, and may acquire a frame-by-pixel image from the i-1-th background and the i- have.

Also, the different-frame acquiring unit 210 can acquire a frame of a gray scale unit, and can be expressed as shown in [Equation 2], and the different frame includes at least one of an object region and a background region can do.

[Equation 2]

here,

Is an i < th >

The image acquiring unit 220 acquires a binary image of a phase based on a constant threshold.

More specifically, the image acquiring unit 220 may acquire a binary image of a different frame based on a background threshold and a constant threshold value used to distinguish the object region.

Also, the image obtaining unit 220 can distinguish the object region and the background region by comparing the contrast threshold value and the pixel value of the different frame, and can be expressed as [Equation 3].

[Equation 3]

here,

Tc is a constant threshold value, and the constant threshold value may be a range of a predetermined pixel value or a range of a predetermined gray scale.

According to the embodiment, the image obtaining unit 220 may obtain a binary image including a background region when the i-th binary image is 0-bit, and obtain a binary image including the object region when the i-th binary image is 1-bit .

The matrix acquisition unit 230 acquires a steady matrix based on the obtained binary image.

The Steady matrix may include a comparison value between pixels of the binary image, and can be expressed as [Equation 4].

[Equation 4]

here,

, It may be an object area,

, It may be a background area.

Also,

Is an i-th Steady matrix, and the initial value of the Steady matrix

Lt; RTI ID = 0.0 > 0, < / RTI >

Can be expressed as " (5) "

[Equation 5]

Here, p and q are frame sizes.

The background acquisition unit 240 determines whether to perform a neighbor based intensity correction algorithm (NIC algorithm) by comparing the robustness threshold and the minimum value of the Steady matrix, I < th > frame for the i < th >

Here, the robustness threshold value is a threshold value used for determining whether or not to perform the background estimation (e.g., intensity correction, etc.), and includes a threshold value including the intensity of the backgrowth pixel adaptively reflected according to the surrounding environment Lt; / RTI >

According to the embodiment, when the minimum value of the Steady matrix is equal to or greater than the robustness threshold value, the background obtaining unit 240 may select the first background or the i-1 background in the i-th background.

More specifically, when the minimum value of the Steady matrix is equal to or greater than the robustness threshold value, the background acquiring unit 240 acquires the first background or the (i-1) th background in the i-th background Can be selected.

For example, when the minimum value of the Steady matrix is equal to or greater than the robustness threshold value, the background acquiring unit 240 can determine that there is no difference between the background of the first background or the i-1 background and the background of the i-th frame The first background or the (i-1) th background can be selected in the i-th background without the intensity correction for the background.

According to an embodiment, the background acquiring unit 240 can acquire the i-th background through the neighbor-based intensity correction algorithm when the minimum value of the Steady matrix is smaller than the robustness threshold value. Hereinafter, with reference to FIG. 3, the neighbor-based intensity correction algorithm will be described in detail.

FIG. 3 is a block diagram showing the background acquiring unit of FIG. 2. FIG.

3, the background acquisition unit 240 may include a filtering unit 241, a mask generation unit 242, a standard deviation calculation unit 243, and a strength correction unit 244. [

The filtering unit 241 can filter from the image frame to at least one doubt pixel whose minimum value of the Steady matrix is smaller than the robustness threshold value.

According to the embodiment, the filtering unit 241 can perform filtering on the dummy pixels included in the background area or not included in the background area, and can be expressed as in the following equation (6).

[Equation 6]

here,

The dirt pixel may be included in the object area, and may be included in the stadi matrix value less than zero.

The mask generation unit 242 may generate masks for the first background or the (i-1) th background from the filtered dou- ble pixels and a mask for the i-th frame.

According to the embodiment, the mask generator 242 may generate two masks, each of which may be a mask for a first background or an (i-1) -th background and a mask for an i-th frame,

Size,

Wow

As shown in FIG.

The standard deviation calculator 243 can calculate the standard deviation of each mask, and can be expressed by the following equation (7).

[Equation 7]

here,

Is the standard deviation, and M is the mask (

), &Lt; / RTI >

May be the intensity of the m-th pixel,

May be an average of the intensities, and can be expressed as [Eq. 8].

[Equation 8]

The intensity corrector 244 can correct the intensity between the filtered douter pixels based on the computed standard deviation and obtain the i-th background for the ith frame using the dou- ble-corrected dou- ble pixels.

For example, the intensity corrector 244 may obtain the i-th background for the i-th frame via (Equation 9).

[Equation 9]

here,

Is the standard deviation for the i < th > frame mask,

Is the standard deviation for the first background mask or the i-1 < th > background mask, and the i < th &

) Can be calculated based on each standard deviation.

Also, the intensity corrector 244 can obtain the standard deviation for the i-th frame through the following equation (10) and correct the intensity for each pixel through the obtained standard deviation.

[Equation 10]

Where k is the background pixel, a is the intensity for the background pixel, l is the object pixel, and b is the intensity for the object pixel.

According to an embodiment, the intensity correction unit 244 may correct the three channels (RGB channels, red green blue channel) existing between the first background or the (i-1) th background and the i- I < / RTI > background motion for the frame.

2, the obtaining unit 200 may further include an update unit 250 that updates the obtained i-th background in the current background.

According to the embodiment, the differential frame acquiring unit 210 can acquire a differential frame for the updated current background and the (i + 1) -th frame.

Therefore, the background acquiring device 10 according to the embodiment of the present invention can acquire backgrounds of frames of successive power in real time.

In addition, the background acquiring device 10 may acquire a foreground corresponding to N backgrounds based on the obtained N backgrounds.

4 is a flowchart illustrating a background acquisition method according to an embodiment of the present invention.

Referring to FIG. 4, in step 410, the background acquisition apparatus separates the video stream into N series frames, and sets a first background based on the first frame in the series of frames.

More specifically, in step 410, the video stream can be divided into N consecutive series of frames, and since there is no initial value for the background, the first frame in a series of frames You can set the background.

In step 420, the background acquisition device acquires the i-th background for the i-th frame. Here, i is a natural number of 2 or more and N or less.

More specifically, in step 420, the background corresponding to each of the N frames can be obtained in real time. Hereinafter, step 420 will be described in more detail with reference to FIG.

5 is a flow chart illustrating an i-th background acquisition method for the i-th frame of Fig.

Referring to Fig. 5, in step 510, the background acquiring device acquires a first background or an i-1 background and a picture frame in units of pixels from the ith frame.

More specifically, in step 510, a pixel-by-pixel image frame from the first background and the second frame may be obtained, and a pixel-by-pixel image frame may be obtained from the i-1-th background and the i-th frame.

The background acquisition device, in step 520, obtains a binary image for the phase-based frame based on the constant threshold value.

More specifically, at step 520, a binary image for a phase-difference frame may be obtained based on a constant threshold used to distinguish the background region and the object region.

In step 530, the background acquisition device acquires the Steady matrix based on the obtained binary image. The Steady Matrix may contain a comparison between the pixels of the binary image.

In step 540, the background acquisition device determines whether to perform the neighbor-based intensity correction algorithm by comparing the robustness threshold value and the minimum value of the steady matrix, and determines whether the i-th background .

According to the embodiment, in step 540, the background acquiring apparatus may select the first background or the i-1 background in the i-th background when the minimum value of the Steady matrix is equal to or greater than the robustness threshold value.

In addition, the background acquiring device may acquire the i-th background through the neighbor-based intensity correction algorithm when the minimum value of the Steady matrix is smaller than the robustness threshold value at step 540. Hereinafter, with reference to FIG. 6, the neighbor-based intensity correction algorithm will be described in detail.

6 is a flow chart illustrating the neighborhood-based intensity correction algorithm of FIG.

Referring to FIG. 6, in step 610, the background acquisition apparatus may filter from at least one frame to at least one dummy pixel whose minimum value of the Steady matrix is smaller than the robustness threshold value.

The background acquisition device may generate a mask for the first background or the i-1 th background from the filtered douter pixel and a mask for the i < th > frame in step 620, and in step 630, Can be calculated.

In addition, the background acquiring device may, in step 640, correct the intensity between the filtered douter pixels based on the computed standard deviation and acquire the ith background for the ith frame using the intensity-corrected douter pixel have.

According to an embodiment, the background acquisition device may further comprise updating the obtained i-th background in the current background, obtaining an updated current background and a different image frame for the (i + 1) -th frame, The foreground corresponding to the N backgrounds can be obtained based on the N backgrounds.

7 is a flowchart illustrating a background acquisition method according to another embodiment of the present invention.

7, the background acquiring device may receive an input for a video stream in step 711, and may separate the video stream into N series of frames in step 712, and in step 713, The first background can be set based on the first frame.

The background acquisition device may acquire a current background that includes a first background or an i-1 background based on the first background at step 714 and may acquire a current frame from the N consecutive frames at step 715 have.

In addition, the background acquiring device may acquire a frame in units of pixels from the i < th > frame and the first background or i-1 background in step 720, and in step 730, A binary image can be obtained.

Further, in step 740, the background acquiring device may acquire a Steady matrix based on the obtained binary image. The Steady Matrix may contain a comparison between the pixels of the binary image.

In addition, the background acquisition device may determine whether to implement the neighborhood-based intensity correction algorithm by comparing the robustness threshold value and the minimum value of the steady matrix at step 750.

The background acquisition device may acquire the i-th background through the neighbor-based intensity correction algorithm if the minimum value of the Steady matrix is less than the robustness threshold at step 760. Hereinafter, with reference to FIG. 8, the neighbor-based intensity correction algorithm will be described in detail.

8 is a flow chart illustrating the neighborhood-based intensity correction algorithm of FIG.

8, the background acquiring device may filter, at step 810, at least one dou- ble pixel whose minimum value of the Steady matrix from the different frame is less than the robustness threshold, and at step 815, It is possible to determine whether or not the intensity correction is performed by the number (n) of dummy pixels.

The background acquisition device may generate a mask for the first background or an i-1 background from the filtered douter pixel and a mask for the i < th >

9 is an example of generating a mask for a first background or an i-1 th background and a mask for an i-th frame, wherein the back-grained acquisition device calculates, at step 820, a difference between the phase difference 303 and the robustness threshold value The background masks 304 and 306 corresponding to the first background or the i-1 background 910 and the frame masks 305 and 307 corresponding to the i-th frame 302 are generated can do.

The background acquisition device may calculate the standard deviation for each mask in step 830 and may, in step 840, correct the intensity between the filtered dirt pixels based on the computed standard deviation.

10 illustrates an example of correcting three channels existing between an i-1 th background and an i-th frame, in which the background acquiring device acquires the first background or the i-1 background 1010 and the i- It is possible to correct the three channels (RGB channel, red green blue channel) existing between the frame 1020 and acquire the motion of the i-th background for the i-th frame.

Referring back to Fig. 7, the background acquiring device may update the ith background acquired in step 770 in the current background.

According to an embodiment, in step 714, the background acquisition device may acquire the updated current background, acquire the i + 1 frame in step 715, and update the current background and the i + 1 The image for the frame can be obtained.

The background acquisition device may separate the background of the current frame based on the updated current background at step 780 and may acquire the foreground based on the isolated background at step 790. [

The method according to an embodiment may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions to be recorded on the medium may be those specially designed and configured for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. For example, it is to be understood that the techniques described may be performed in a different order than the described methods, and / or that components of the described systems, structures, devices, circuits, Lt; / RTI > or equivalents, even if it is replaced or replaced.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

10: Background Acquisition Device
100: initial setting section
200:
210:
220: Image acquisition unit
230: Matrix acquisition unit
240: background acquisition unit
250: Update section
241:
242: mask generation unit
243: Standard deviation calculator
244:

Claims

An initial setting unit for dividing a video stream into N series frames and setting a first background based on a first frame in the series of frames; And
An acquiring unit (i is a natural number of 2 or more and N or less)
Lt; / RTI >
The obtaining unit
An image frame acquiring unit for acquiring a first background image or an (i-1) th background image and a picture-by-pixel image frame from the i-th frame;
An image acquiring unit that acquires a binary image of the image based on a constant threshold for distinguishing the background area and the object area;
A matrix acquiring unit for acquiring a steady matrix based on the obtained binary image; And
Determining whether to perform a neighbor-based intensity correction algorithm (NIC algorithm) by comparing a robustness threshold including a strength of a background pixel and a minimum value of the steady matrix, I < / RTI > background for the i < th > frame based on the determined implementation,
/ RTI >

The method according to claim 1,
The background acquiring unit
If the minimum value of the Steady matrix is equal to or greater than the robustness threshold value, the first background or the (i-1) th background is selected in the i-th background
Background Acquisition Device.

The method according to claim 1,
The background acquiring unit
If the minimum value of the Steady matrix is less than the robustness threshold value, acquiring the i < th > background through the neighbor based intensity correction algorithm
Background Acquisition Device.

The method according to claim 1,
The background acquiring unit
A filtering unit for filtering at least one dou- ble pixel whose minimum value of the Steady matrix is smaller than the robustness threshold value from the image frame;
A mask generator for generating a mask for the first background or the (i-1) th background from the filtered dummy pixel and a mask for the i-th frame;
A standard deviation arithmetic unit for calculating a standard deviation for each of the masks; And
I < th > frame for the i < th > frame by using the dou- ble pixels corrected for the intensity based on the calculated standard deviation,
/ RTI >

The method according to claim 1,
The obtaining unit
I < th > background in the current background,
Further comprising:

6. The method of claim 5,
The phase difference obtaining unit
The updated current background and the image for the (i + 1) -th frame are acquired
Background Acquisition Device.

The method according to claim 1,
The image obtaining unit
Acquiring the binary image based on the constant threshold value
Background Acquisition Device.

A filtering unit for filtering at least one dummy pixel based on a first background or an (i-1) th background in a series of N frames and an image for the i-th frame, wherein i is at least 2 Number);
A mask generator for generating a mask for the first background or the (i-1) th background from the filtered dummy pixel and a mask for the i-th frame;
A standard deviation arithmetic unit for calculating a standard deviation for each of the masks; And
I < th > frame for the i < th > frame using the dou- ble-corrected dirt pixel, and correcting the intensity between the filtered dirty pixels based on the calculated standard deviation,
Based intensity correction device.

9. The method of claim 8,
The filtering unit
Obtaining a binary image for the frame based on a constant threshold value for distinguishing a background region and an object region, obtaining a Steady Matrix based on the obtained binary image, and determining a minimum value of the Steady Matrix as a strength To filter the at least one dummy pixel that is smaller than the robustness threshold value
Neighbor-based intensity correction device.

Separating the video stream into a series of N frames and setting a first background based on the first frame in the series of frames; And
I < th > frame (i is a natural number of 2 or more and N or less)
Lt; / RTI >
The step of acquiring the i < th >
Obtaining a first background or an (i-1) th background and a picture frame in units of pixels from the i-th frame;
Obtaining a binary image for the image based on a contrast threshold value that distinguishes the background area and the object area;
Obtaining a Steady matrix based on the obtained binary image; And
Determining whether to perform a neighbor-based intensity correction algorithm by comparing a robustness threshold value including a strength of a background pixel and a minimum value of the steady matrix, and determining whether the i < th > Steps to acquire background
/ RTI >

11. The method of claim 10,
Wherein if performing the neighborhood-based intensity correction algorithm, acquiring the i < th >
Filtering from at least one dou- ble pixel whose minimum value of the Steady matrix is less than the robustness threshold value from the image frame;
Generating a mask for the first background or the (i-1) th background from the filtered douter pixel and a mask for the i-th frame;
Computing a standard deviation for each of the masks; And
Correcting the intensity between the filtered douter pixels based on the computed standard deviation and obtaining the ith background for the i < th > frame using the dou- ble-corrected dou- ble pixels
/ RTI >

Filtering a first background or an i-1 background in a series of N frames and at least one douter pixel based on a frame for an i-th frame, wherein i is a natural number of 2 or more and N or less; );
Generating a mask for the first background or the (i-1) th background from the filtered douter pixel and a mask for the i-th frame;
Computing a standard deviation for each of the masks; And
Correcting the intensity between the filtered douter pixels based on the computed standard deviation, and obtaining an i-th background for the i-th frame using the dou- ble-corrected dou- ble pixels
Based intensity correction method.

13. The method of claim 12,
The filtering step
Obtaining a binary image for the frame based on a constant threshold value for distinguishing a background region and an object region, obtaining a Steady Matrix based on the obtained binary image, and determining a minimum value of the Steady Matrix as a strength The at least one dummy pixel being smaller than a robustness threshold value that includes
Neighbor - based intensity correction method.