KR101496407B1 - Image process apparatus and method for closed circuit television security system - Google Patents

Abstract

The present invention relates to a security CCTV image processing apparatus and an image processing method, in which an object and a background are separated by binarizing an object from a monochrome image subjected to image processing to white and a background to black, Determining the center coordinates of the object whose effectiveness has been determined, generating a motion region with reference to the center coordinates of the object, tracking the center coordinates of the object, and determining the center coordinates of the object as the movement region And stores the image in case of escaping. According to the present invention, only a valid image required for security can be stored by accurately determining the degree of movement of an object with reference to the movement region, and by performing image processing only on a movement region formed around the moving object, Speed can be improved.

Description

TECHNICAL FIELD [0001] The present invention relates to a security CCTV image processing apparatus and an image processing method,

The present invention relates to an image processing apparatus and an image processing method, and more particularly, to a security CCTV image processing apparatus and image processing method for identifying an object moving from an image and storing the image.

Recently, image processing technology has been applied in various fields, and a representative example thereof is applied to a security system.

That is, various image processing techniques for real-time processing and storing vast amount of data received through a camera of a security system have been proposed.

Hereinafter, the conventional image processing technique will be described.

1 is a flowchart of a conventional image processing technique.

Referring to FIG. 1, when an image transmitted from a camera is received (S 1), a comparison with a reference image is performed every time a current image is input (S 2) If the current image and the reference image are the same, the process returns to step S1. If the current image and the reference image are not the same, the current image is stored in the storage (S4).

That is, every frame is not subjected to image processing every time an image is input from the camera, and image processing is performed only on the current image in which a change in the situation has occurred, thereby reducing data throughput and storage amount. Such an image processing technique may be useful particularly in a place requiring security (a place where the image change is small).

Conventionally, as a method of comparing a reference image and a current image, if a pixel-to-pixel comparison method of comparing the reference image and the current image one by one with each pixel value is used, for example, It requires a high-performance image processing apparatus because it needs to compare 270,000 or more data in 1/30 second. Therefore, real-time image processing speed may be limited depending on the image processing apparatus.

On the other hand, there is a disadvantage that an unnecessary image is stored by judging that the current image has been changed (misoperation) in response to a change in ambient brightness and the like even though the situation has not substantially changed. In other words, image processing technology that recognizes the exact situation change is not enough yet.

Therefore, in order to reduce the vast amount of data received through the camera of the security system, a new technique of image processing technology for storing only valid images required for security will be needed.

Korean Patent Laid-Open Publication No. 10-2011-000037485 (published on April 23, 2011)

SUMMARY OF THE INVENTION Accordingly, the present invention has been made to solve the above-mentioned problems of the prior art, and it is an object of the present invention to provide a security system, which is capable of minimizing a large amount of data received through a camera of a security system, The present invention also provides a security CCTV image processing apparatus and an image processing method for setting a moving area and storing an image only when the moving area is out of the moving area.

According to another aspect of the present invention, there is provided a security CCTV image processing apparatus comprising: an object discrimination unit for converting a color image transmitted from a camera into a monochrome image to discriminate between an object and a background; A validity determining unit for determining the validity of the object in the pixel; A center coordinate determination unit for generating center coordinates of the object; And an object motion determining unit determining whether to store the image by determining whether the center coordinates of the object are out of the movement area.

In this case, the object determining unit may include a monochrome image converting unit for converting a monochrome image into a color image, and a binarizing unit for binarizing the object by defining white as the background and black as the background, A center coordinate tracking unit for tracking the center coordinates of the object, a motion determination unit for determining whether the center coordinates of the object exist inside the movement area, And a judgment unit.

Meanwhile, the security CCTV image processing method of the present invention includes the steps of: binarizing an object from a monochrome image subjected to image processing to white and a background to black to divide an object and a background; Determining validity of the object at each pixel from the binarized image; Determining center coordinates of an object whose validity of the object has been determined; Generating a motion region based on the center coordinates of the object; Tracking the center coordinates of the object; And storing the image when the center coordinates of the object are out of the movement area.

As described above, according to the security CCTV image processing apparatus and image processing method of the present invention, it is possible to accurately determine whether an object is moved by setting a movement region, so that only valid images required for security can be stored. Therefore, the amount of storage can be reduced.

Further, according to the present invention, image processing speed can be improved by performing image processing only on a moving region formed around a moving object.

According to the present invention, only an image required for security can be stored, and thus an efficient security system can be constructed.

1 is a flowchart of a conventional image processing technique.
2 is a block diagram of a control circuit of a security CCTV image processing apparatus according to an embodiment of the present invention.
3 is a flowchart illustrating a security CCTV image processing method according to an exemplary embodiment of the present invention.
4 is a view showing a stored current image data folder.
5 shows an example in which the image upload fails when the time delay is not set.
Figure 6 is an illustration of an image that is incompletely uploaded.
7 is a diagram showing a completely uploaded image.
8 is a comparison chart of difference images according to brightness sensitivity.
9 is a diagram showing an example of an access to an erroneous index.
10 is a diagram showing a result of removing noise from a difference image with a reference image.
11 is a conceptual diagram showing a method of expressing coordinate values of an image.
12 is a graph showing the result of calculation of coordinates of upper left and upper right coordinates.
FIG. 13 is a diagram showing the result of calculation of the coordinates of the lower left and right lower center coordinates.
14 is a diagram showing the object center coordinates and the movement area.
Fig. 15 is a diagram for judging the distance calculation in the movement area and the intruder detection.
16 is a view showing the main page of the homepage.
17 is a diagram showing a sensor information output page.
18 is a diagram showing a home control page.
19 is a diagram showing a captured picture output page.
20 is a diagram showing a first screen when an application is executed.
21 is a view showing the first screen of the homepage.
22 is a diagram showing a sensor information confirmation page.
23 is a diagram showing a home control page.
24 is a view showing a photographing page at the time of the click.

Hereinafter, a security CCTV image processing apparatus and an image processing method according to the present invention will be described in detail with reference to the accompanying drawings.

2 is a block diagram of a control circuit of a security CCTV image processing apparatus according to an embodiment of the present invention.

Referring to FIG. 2, an object discriminating unit 1 for discriminating between an object and a background by converting a color (RGB) image transmitted from a camera (not shown) into a monochrome image, a validity discriminating unit A center coordinate determination unit 3 for generating center coordinates of the object, and an object motion determination unit 4 for determining whether the center coordinates of the object deviate from the movement region and storing the image .

The object discriminating section 1 includes a monochrome image converting section 11 for converting a monochrome image into a color image, and a binarizing section 12 for binarizing the object by defining the object as white and the background as black.

The object movement determining unit 4 includes a movement area generating unit 41 for generating a movement area around the center coordinates of the object, a center coordinate tracking unit 42 for tracking the center coordinates of the object, And a motion determiner 43 for determining whether or not coordinates exist in the movement area.

Here, the movement region is preferably defined as a circle centered on the center of the object.

3 is a flowchart illustrating a security CCTV image processing method according to an exemplary embodiment of the present invention.

Referring to FIG. 3, when an image is transmitted from the camera (S11), the monochrome image converting unit 11 of the object determining unit 1 converts the color (RGB) image into a monochrome image (S12) The binarization unit 12 of the binarization unit 1 binarizes an object from a monochrome image to white and a background to black to distinguish the object from the background (S13).

Then, the validity determination unit 2 determines the validity of the object in each pixel from the binarized image (S14). That is, if a white pixel is detected while searching all the pixels, the data of eight surrounding pixels of the white pixel are examined, and if the number of pixels represented by white is five or more, the middle pixel is determined to be a valid pixel, It is determined and removed. The noise is processed in black.

Subsequently, the center coordinate determination unit 3 determines the center coordinates of the object whose effectiveness has been determined (S15).

Subsequently, the motion-area generating unit 41 of the object-motion determining unit 4 generates a motion region around the center coordinates of the object (S16). When the motion region is generated, it is preferable to perform only the image processing for the motion region in order to improve the image processing speed.

Thereafter, the center coordinate tracking unit 42 of the object motion determination unit 4 tracks the center coordinates of the object (S17), and the motion determination unit 43 of the object motion determination unit 4 determines that the center coordinates of the object are If it is determined that the center coordinates of the object are out of the movement area, the image is stored (S19). On the other hand, when it is determined that the center coordinates of the object do not deviate from the movement area, it is regarded as the background (S20).

Hereinafter, a specific embodiment of the security CCTV image processing apparatus and method configured as described above will be described.

The video program in which the security CCTV image processing method of the present invention is executed has two functions. The first function is 'View Current Image' to check the image input from the camera, and the second function is 'Check for intruders'.

First, if you execute the first function 'View Current Image', the image input from the camera is output to a new window named 'Shooting'. At this time, as shown in FIG. 4, when the 'view current image' button is clicked, the image captured from the camera at the time of clicking the button is captured and stored in the server as 'capture.jpeg'.

In this case, when the 'view current image' button is clicked, the image program must store the current image in the server path before the connected page is outputted. By intentionally executing the loop statement at the beginning of the page to match this time, It is necessary to generate a time delay. If the time delay is not generated, the image load may not be properly performed as shown in FIG.

At this point, how much of the loop should be done depends on the server specification. For example, in case of Intel (R) Core (TM) i5 750 2.67GHz, DDR3 4G RAM, it can be confirmed that the upload is executed properly when the loop is executed from 1 to 9,999,999. On the other hand, when the loop is performed from 1 to 999,999, it can be seen that the image upload is incomplete about 2 to 3 times in 10 times as shown in FIG.

When the iterative statement is used to perform the loop from 1 to 9,999,999, the image input from the camera is properly captured and output to the web browser as shown in FIG.

On the other hand, the second function of the video program is the 'intruder identification' function. The image input from the camera is divided into odd-numbered frames and even-numbered frames, and difference images of two images are successively performed. If a reference image is registered and a difference image is obtained from the reference image, a still image is continuously stored in the server even if the position of the camera is shifted or the intensity of the light is small relative to the reference image. It should not happen that images are stored continuously in this limited space. In order to compensate for this problem, it is possible to compensate for this problem by performing continuous difference image operation by inputting odd and even numbers in the real-time input image rather than the difference image operation with the reference image.

First, the sensitivity of brightness is considered in obtaining the difference image.

Since the image is generated in two dimensions, it approaches the pixel value through the double loop. The absolute value obtained by subtracting the value of the data located at the same coordinates from the two images is stored in the variable value. If the value is greater than or equal to the specified constant value LITE_SENSITIVITY, the value of the subCount variable is incremented by one. If the value stored in this subCount is equal to or greater than MOVEMENT_SENSITIVITY (motion sensitivity), it is determined that image_01 and image_02 are different images, and the result of the difference image operation is output.

The reason why the difference between the two images is discriminated by using a constant called LITE_SENSITIVITY is that the brightness value varies from time to time and it is very sensitive to the change.

As shown in FIG. 8, when the sensitivity is set to 20, it can be confirmed that the object can be separated from the background most perfectly. When the sensitivity is too low, it is difficult to detect a proper object because a lot of background noise occurs. If the sensitivity is set too high, it can be confirmed that the object can not be detected properly due to a change in the brightness value. Through this experiment, it can be concluded that it is necessary to set the brightness sensitivity appropriately according to the experimental environment.

First, image variables of the same size as the difference image are generated, and the color values are all initialized to 0 (black). We then use a double loop to access the secondary image data, but adjust the index value appropriately to avoid searching the border pixels. This is to prevent an index access error because it is necessary to search 8 pixels around the pixel while searching for the pixel. For reference, FIG. 9 shows an example of the access to the wrong index.

In this way, the image data is searched as a double loop, and the validity test is performed on the pixels that are estimated as valid data. This is because we have not yet removed the noise, but merely searched for a pixel that is supposed to be an object, so we need a step that determines whether this pixel represents valid information or noise.

If a condition of 50% or more is detected, it is judged that five or more of the nine pixels should be detected to be a valid pixel. If the number of pixels is less than 5, the pixel is judged as noise and considered as background. The result of the experiment by removing the noise in this manner is shown in FIG.

On the other hand, eliminating the noise makes it possible to obtain more precise information desired by the user in the background. In the center coordinate calculation method of the present invention, the search is performed in units of rows and columns on a black background to examine the frequency of objects represented by white, and the coordinates of rows and columns having the highest frequency are obtained. The coordinates of the horizontal and vertical centers of the image. The coordinates of the image are the starting point (0, 0) at the upper left corner, the width index value is increasing at the right, and the height index value is decreasing at the lower right. That is, coordinate values are expressed as shown in FIG.

The result of calculating the center coordinate value with the function defined above is as shown in Fig. 12 (upper left corner, upper right corner) and Fig. 13 (lower left corner, lower right corner). First, let's analyze the result of calculating coordinate values when an object is detected at the top of the image. The size of the image used in the experiment is 640 pixels in width and 480 pixels in height.

Referring to FIG. 12, the results of calculating the coordinates of the upper left corner and the upper right corner can be confirmed. First, when the image is viewed on the row, the pixel having the highest pixel density is the row in which the finger is displayed. At this time, the coordinates correspond to the height of the image. When comparing two images, it can be confirmed that the height of the fingers is similar, and the calculation result is also very similar. However, since there is a large difference in abscissa, the highest density in the image in terms of the thermal unit is the left end and the right end, respectively, so that the left upper left figure should be close to 0 and the upper right figure should be close to 640 . As a result of the experiment, it can be confirmed that the experimental data corresponding to this is appeared.

Referring to FIG. 13, as a result of the experiment when the object is detected at the lower end, it can be visually confirmed that the object with the highest frequency is the object under the finger have. Since the coordinates at this time are the ordinate of the object and are located at the lower end, a number close to 480 should be displayed. Then, when you see the object in the column unit, the part with the finger and the one holding the fist is the most frequently displayed in the same column, so the coordinates at this time are the abscissa of the object. Since the width of the image is 680 pixels and it is located a little away from the left end, it is roughly known that it is within about 100 coordinates. Experimental results show that the center coordinates of the object located at the lower left are calculated as 470 in the vertical direction and 70 in the horizontal direction, which is similar to the roughly predicted value. The coordinates of the center of the object located at the lower right corner should similarly be looked up on the first row to find the place where the white pixel is most frequently displayed. The coordinates at that time are the ordinate of the object. Compared with the figure at the bottom left, we can see that the high-frequency row is located at a slightly higher elevation and it can be predicted that the ordinate of the object will be measured slightly lower. There may be room for chaos, but it should be recalled that the origin of the image is located at the upper left corner, and the vertical coordinate value increases rather than decreases. Experimental results show that the vertical coordinate of the object located at the bottom right is slightly lower than the vertical coordinate of the object located at the bottom left, which confirms that it matches the predicted result. The abscissa is the column with the highest frequency when viewed in terms of columns, with the finger and the fisted part together, and the coordinates at this time are the abscissa of the object. Since we are located at the right end, it is predicted that the abscissa will be close to 640, and we can confirm that the center abscissa of the object located at the lower right corner is 565. It can be confirmed that this is also in agreement with the results predicted before the experiment. Taking the experimental results of FIGS. 12 and 13 together, it can be confirmed that the center coordinates can be calculated properly.

Then, the process of determining the intruder is performed.

Referring to FIG. 14, after calculating the center coordinates, the movement area should be set based on a certain distance based on the center coordinates of the object. First, when the first motion is detected, assign the height center point and the width center point of this object to the variables declared as ref_h_point and ref_w_point, respectively. Based on this information, it is necessary to set a movement area by drawing a circle having a certain distance as a radius. Then, in the frame input continuously from the camera, the center of height and the center of the width are similarly computed, and the current motion is displayed in a color different from the reference point by substituting into the variable declared as center_h_point and center_w_point.

OpenCV supports built-in functions that draw a circle on an image. Here is the function called cvCircle (). That is, it includes 'Variable in which the image to be drawn is stored', 'Center coordinates of circle', 'Radius', 'Line color', 'Line thickness', 'Line shape', 'Move from center point'.

In the above function, the movement area is displayed in red, and the object in which the current motion is detected is displayed in green. The results of the experiment are shown in Fig.

Referring to FIG. 15, when a palm is displayed for the first time in the background, a center point is obtained, a circle is drawn as much as the set area based on the center point, and then the movement of the palm is continuously captured to output the corresponding center coordinates in green. If the movement of the palm is detected inside the red circle, it is considered to be the moving background, not the intruder. However, if the palm movement is outside the red circle, it is regarded as the intruder by judging that the new object moves instead of the background with constant movement , And stores the image at that point in time. At this time, it is preferable to perform only the image processing for the moving region in order to improve the image processing speed.

The security CCTV image processing apparatus and image processing method of the present invention presented in the present invention can perform remote monitoring and control in a PC, a smart phone, and the like.

First, a web server is constructed and a homepage is created, and a login screen can be registered or logged in as shown in FIG.

When logging in, the measured values (voltage, current, active power, reactive power, apparent power, power factor, temperature, and humidity) received from the sensor are displayed as shown in FIG. That is, the present invention enables control of doors, windows, and electric lamps as well as monitoring thereof by using a PC and a smart phone.

When the user clicks 'Control device', the screen is moved to the page as shown in FIG. 18, the current state of the home is displayed, and the doors and bulbs of the home can be remotely controlled.

If the user clicks 'View Current Image', as shown in FIG. 19, the camera at the click point captures and displays the photograph. You can check the status of your house by installing a camera in your house. In other words, you can check the current status of the house to see if the light bulb is on, the door is open, or there is an intruder.

Meanwhile, the present invention can also be implemented through a smart phone.

When you go out, you can go to the homepage to check the house status and control it so that it can be linked to the smart phone. I created an application to access the homepage from the smartphone. When I run the application, I go to the homepage main screen. This allows you to check and control the state of the house at any time, even when you are away from home. I installed an apk file on my smartphone and created an application called MyHome. When the application is executed, the first screen (FIG. 20) is displayed. When the user clicks the click button, the user is connected to the home page main screen as shown in FIG. 21, and can perform functions such as checking sensor information, controlling the device, and viewing a current image, as shown in FIGS.

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, but, on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the invention.

1: object discrimination unit
11: B &
12: binarization unit
2: validity discrimination unit
3: center coordinate determination unit
4: object motion discrimination unit
41:
42: center coordinate tracking unit
43:

Claims (3)

An object discrimination unit including a monochrome image converting unit for converting a color image transferred from a camera into a monochrome image, and a binarizing unit for binarizing the object by defining white as a background and black as a background;
A validity determining unit for determining the validity of the object in the pixel;
A center coordinate determination unit for generating center coordinates of the object; And
A center coordinate tracking unit for tracking the center coordinates of the object, and a determination unit for determining whether the center coordinates of the object exist within the movement area An object motion determining unit that determines whether or not to store an image;
Wherein the validity determination unit checks surrounding pixels when a white pixel is detected, and determines that the effective pixel is a white pixel if the number of white pixels is greater than or equal to the set number;
The center coordinate determination unit detects the frequency of the object represented by white in the row and column units on the black background and determines the coordinates of the row and column having the highest frequency as the center coordinates;
The object movement determining unit may determine,
Numbered frame and an even-numbered frame, and performs a difference image operation of two images continuously between the odd-numbered frame and the even-numbered frame,
Wherein the difference image operation increases the count value when the absolute value between data located at the same coordinates of the odd frame and the even frame is greater than a preset brightness sensitivity, and when the count value is greater than a preset motion sensitivity, And outputs a result,
Wherein the movement region defines a circle centered on the center of the object,
Wherein when the moving area is generated, only the image processing for the circle is performed.
delete A method for performing security CCTV image processing using the apparatus of claim 1,
Separating an object and a background by binarizing an object from a monochrome image subjected to image processing to white and a background to black;
Determining validity of the object at each pixel from the binarized image;
Determining center coordinates of an object whose validity of the object has been determined;
Generating a motion region based on the center coordinates of the object;
Tracking the center coordinates of the object; And
And storing the image when the center coordinates of the object deviate from the movement area,
In the step of determining the effectiveness of the object, if a white pixel is detected, surrounding pixels are checked to determine that the white pixel is a valid pixel if the number of white pixels is greater than the set number;
In the step of determining the center coordinates, the frequency of an object represented by white is detected by detecting row units and column units on a black background, respectively, and coordinates of rows and columns having the highest frequency are determined as center coordinates;
In the step of generating the moving region,
Numbered frame and an even-numbered frame, and performs a difference image operation of two images continuously between the odd-numbered frame and the even-numbered frame,
Wherein the difference image operation is performed by calculating an absolute value obtained by subtracting the value of data located at the same coordinates of the odd frame and the even frame from each other and increasing the count value when the absolute value is greater than a predetermined brightness sensitivity, Determines a different image and outputs a result when it is larger than a preset motion sensitivity,
Wherein the movement region defines a circle centered on the center of the object,
When the moving region is created, performs only image processing on the circle.

Images