KR101726690B1 - System for background substraction - Google Patents

Abstract

According to an embodiment of the present invention, there is provided a background difference system and method, comprising: a target presence / absence determination unit for determining whether a current status signal is equal to or greater than a threshold value relating to movement of a target; A target state determination unit for determining whether the previous state signal is equal to or more than a threshold value when the current state signal is less than or equal to a threshold value; A weight setting unit for setting a weight of an adaptive filter algorithm for differentiating the background in the observation area based on the information of the current state signal and the previous state signal; A background differential system is provided.

Description

Background Differential System {SYSTEM FOR BACKGROUND SUBSTRACTION}

The present invention relates to a background differential system and method with differential weights.

More particularly, the present invention relates to a background differential system and method that enable more accurate detection of a target by observing a motion state of a target and differentiating weights when an adaptive filter algorithm is used.

The video surveillance system is a system that installs cameras in various places where surveillance is needed and transmits the images acquired from the cameras to a monitor or stores them in a storage device to perform real-time surveillance and post-search.

In recent years, as the demand for video surveillance has increased in society as a whole, intelligent image analysis techniques using motion detection technology, face recognition technology, and object tracking technology have been developed for people in order to solve the labor shortage phenomenon Is being developed.

The target tracking technique is to detect the position and the relative speed by calculating the time difference between the transmission and reception of pulses by receiving the impulse transmission signal coming into the target. Therefore, it is necessary to study algorithms that can effectively remove background noise and extract only desired target pulses because the remaining signals except for the target receiving pulse act as background noise and cause many problems in position tracking.

Many methods for background noise removal, ie, background difference algorithm, have been studied, and the adaptive filter type background difference algorithm is most widely used.

On the other hand, IR-UWB (impulse Radio Ultra Wideband) technology is a technology that can complement the existing camera-based surveillance system by enabling accurate position recognition / tracking function of several tens cm in indoor and outdoor with low cost and low power consumption It is expected.

The present invention uses a previous state signal of a target to determine whether or not the target is stopped so that the weight of the current state signal is reflected in the background difference calculation process when the target stops, And to prevent a decrease in the number of pixels.

According to an exemplary embodiment of the present invention, a target presence / absence determination unit that determines whether a size of a current state signal based on an ultra wideband (UWB) communication in a viewing region is equal to or greater than a threshold value for motion of a target, ; A target state determiner for determining whether a size of the previous state signal based on the UWB communication is equal to or greater than a threshold value when the current state signal is less than or equal to a threshold value; A weight setting unit for setting a weight of an adaptive filter algorithm for differentiating the background in the observation area based on the information of the current state signal and the previous state signal; A background differential system is provided.

In the present invention, the target presence / absence determining unit determines that the target is stopped or does not exist in the observation region when the size of the current state signal is less than or equal to a threshold value.

In the present invention, the target presence / absence determining unit may determine that the target is moving in the observation area when the size of the current state signal is equal to or greater than the threshold, and when the target is determined to be moving, 1 weighting value.

In the present invention, the target state determination unit determines that the target exists in the observation area and stops when the size of the current state signal is less than or equal to the threshold value and the size of the previous state signal is equal to or greater than the threshold value. And sets the weight as a second weight.

In the present invention, the target state determination unit determines that a target does not exist in the observation area when the size of the current state signal is less than or equal to a threshold value and the size of the previous state signal is less than or equal to a threshold value, The weight is set as the first weight.

In the present invention, the first weight is larger than the second weight.

According to the present invention, when a target stops, information on a current state signal is lowered to a weight that is reflected in a background difference calculation process by judging whether or not the target is stopped by utilizing a previous state signal of the target, Can be prevented from being reduced.

1 is a diagram schematically illustrating an example of a background differential system according to an embodiment of the present invention.
2 is a diagram showing an image obtained by applying an existing method and an embodiment of the present invention.
3 is a diagram showing a configuration of a background differential server.
4 is a flowchart illustrating an algorithm of a background difference method according to an exemplary embodiment of the present invention.
5 is a graph comparing the signal pulse magnitudes of the conventional method of the present invention and the background difference method according to an embodiment of the present invention.
FIG. 6 is a graph of a position tracking using an adaptive filter algorithm and an algorithm of a background difference method according to an embodiment of the present invention.

The following detailed description of the invention refers to the accompanying drawings, which illustrate, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It should be understood that the various embodiments of the present invention are different, but need not be mutually exclusive. For example, the specific shapes, structures, and characteristics described herein may be implemented by changing from one embodiment to another without departing from the spirit and scope of the invention. It should also be understood that the location or arrangement of individual components within each embodiment may be varied without departing from the spirit and scope of the present invention. Therefore, the following detailed description is not to be taken in a limiting sense, and the scope of the present invention should be construed as encompassing the scope of the appended claims and all equivalents thereof. In the drawings, like reference numbers designate the same or similar components throughout the several views.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings in order to facilitate a person skilled in the art to which the present invention pertains.

1 is a diagram schematically illustrating an example of a background differential system according to an embodiment of the present invention.

Referring to FIG. 1, the camera 100 system of the present invention includes a plurality of surveillance cameras 100, a background differential server 300, and a communication environment between the plurality of cameras 100 and the background differential server 300 .

Referring to FIG. 1, a plurality of surveillance cameras 100 exchange data with a background differential server 300 through a communication network 200. Specifically, the surveillance cameras 100 communicate with the background difference server 300 through a communication channel (DCOM), and deliver live view video data to the background difference server 300 through a video data channel (DIMA). Of course, only one surveillance camera 100 other than the plurality of surveillance cameras 100 may communicate with the background differential server 300, and one surveillance camera 100 or a plurality of surveillance cameras 100 may communicate with a plurality of background differences And may communicate with the servers 300 and the like. Here, the communication network 200 forming the communication channel DCOM and the video data channel DIMA may be any means capable of transmitting and receiving data or commands by wire or wireless. For example, the communication network 200 may be an ultra wideband (UWB), particularly an IR-UWB (Impulse Radio Ultra Wideband) communication network, which transmits a large amount of digital data over a wide spectrum frequency in a short- have.

As described above, in order to accurately detect the surveillance target, the background difference that leaves only the image of the target should be preceded by removing the background from the images acquired by the surveillance camera 100. [

A number of methods have been studied with the existing background difference algorithm, and the background difference algorithm of the adaptive filter method is the most widely used. The adaptive filter method is an algorithm for restoring the current position signal of the target in the background noise using the difference between the previous signal and the current signal by the moving target. The background difference algorithm uses the following equation.

[Equation 1]

In Equation (1), Bk is the size of the currently estimated background noise,? Is the weight, Bk-1 is the previously estimated background noise magnitude, and xk is the size of the currently received signal. Depending on the weight α between Bk-1 and xk, the adaptive filter algorithm determines the performance between stability and agility.

The larger the α value, the more quickly the change waveform can be reflected due to the movement of the target. However, since the amount of information of the received signal at the time of stopping is reflected in the background difference calculation process, On the other hand, as the value of α is smaller, stable position information can be maintained even when the target is stopped. However, when moving after stopping, the position information during the stop period is reflected in the background noise calculation for a long time. This shape can be a big problem in the position detection process even after the movement of the target.

In other words, the biggest problem of the conventional adaptive filter-based background difference algorithm is that when the target is stopped, the amount of change between the previous signal and the current signal is very small and information representing the position of the target can be reflected in the background difference calculation process to be. This can be a problem in position measurement by reducing the pulse size of the target and lowering the detection probability. This problem is caused by the fact that, when applied to an actual monitoring system, detection and tracking can not be performed when an unauthorized person is latent in the current position.

Therefore, it is necessary to use a background difference algorithm that applies a different weight according to the state of the target

The background differential system according to an embodiment of the present invention uses the previous state information of the target to solve the problem to determine whether the target is stopped or not. Accordingly, when the target stops, the weight of the current received signal information reflected in the background difference calculation process is lowered, thereby preventing the pulse size representing the position information from being reduced.

2 is a diagram showing an image obtained by applying an existing method and an embodiment of the present invention.

FIG. 2 (a) is a diagram showing an existing background difference result when there is an object moving in the image. A general background subtraction method uses a method to detect a moving region in an image and to exclude a region having no change in the image. (a), it can be seen that the rest of the image except for the person walking is removed.

FIG. 2 (b) is a diagram showing a conventional background difference result when the target is stopped in the image. The conventional background difference method has a problem that if the target is stopped for a predetermined time in the image, the target is misjudged as a background because the background difference method determines the area that does not change with time as the background. (b), it can be seen that, when the person in the image is stationary, the resultant person of the background difference is also removed.

FIG. 3C is a diagram illustrating a background difference result that can detect a still object by applying a background difference method according to an embodiment of the present invention. According to an embodiment of the present invention, when it is determined that a target is moving in a previous image of an image to be detected, the target can be detected without discriminating a stopped target by applying a weight differentially. (c), even if a moving target enters a range of an image and stops for a predetermined time, a stopped target can be detected by applying a background difference system according to an embodiment of the present invention.

3 is a diagram showing the configuration of the background difference server 300. As shown in FIG.

Referring to FIG. 3, the background difference server 300 of the present invention includes a target existence determination unit 310, a target state determination unit 320, and a weight application unit 330.

First, the target presence determining unit 310 sets a threshold for determining the presence or absence of a target and detects whether or not a signal exceeding a threshold value is found in the observation region. If a signal above the threshold is found, it is assumed that there is a target moving within the observation area.

The problem is that a signal of a threshold value or more is not found in the observation area. In this case, there is no significant target in the observation area or the target that has entered the observation area is stopped for a predetermined time. Therefore, it is necessary to determine the state of the target to confirm whether the target is not present or the target is stopped.

The target state determination unit 320 determines whether the target is stopped in the current observation area using the previous state information of the target when the target presence determination unit 310 determines that the signal exceeding the threshold value is not found. The target state determination unit 320 utilizes the previous state information to determine whether the target is stopped or not in the observation area.

When the target is stationary, most signal sizes that have gone through the existing background subtraction algorithm are reduced to noise level levels that are much lower than the threshold. As the size of the signal becomes smaller, it becomes difficult to determine whether the target is in the stop state or not within the measurement range. Therefore, the target state determination unit 320 utilizes the previous state information for discrimination between the two cases.

The target state determination unit 320 determines the state of the current signal x _k and the previous signal x _{k -m} and determines whether or not there is a signal greater than or equal to the threshold value in the previous signal. Here, the previous signal x _{k -m} denotes the final state information of the previous signal, and m is usually set to a value of 5 to 10. The threshold value is a threshold set by the target presence / absence determining unit 310.

The target state determination unit 320 estimates that the target is in a stationary state when a signal of a threshold value or more is observed in the previous signal and if the target signal does not exist in the previous signal, State.

Next, the weight application unit 330 sets a weight to be high if there is a target moving within the observation area or there is no target according to the determination result of the target presence determination unit 310 and the target state determination unit 320, If it is stopped, the weight is set low.

That is, when the target presence determining unit 310 determines that there is a moving target, the weight applying unit 330 applies the first weight 1. Further, when the state of the target is regarded as 'no target exists' (environment in which only noise exists), the weight is increased by applying the first weight? 1 so that the target can be quickly responded to the appearance of the target.

Lastly, when the target state determiner 320 regards the state of the target as 'stopped', the weight applying unit 330 applies the second weight? 2 to lower the weight to prevent the pulse size from being reduced.

The operations of the target presence / absence determiner 310, the target state determiner 320, and the weight applying unit 330 may be summarized by the following algorithm (Equation 2).

&Quot; (2) "

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

Referring to FIG. 4, first, the target presence / absence determining unit 310 determines whether there is a signal of a threshold value or more in the observation region of the current signal (S1).

If it is determined that there is more information than the threshold value, it is determined that there is a target moving in the current position in the observation area because it is detected that there is a target that is moving even in the current state (S2), and the weight is set as the first weight (S3).

Next, when the target presence determination unit 310 determines that there is no more information than the threshold value in the observation area, the target state determination unit 320 confirms the final state information of the previous signal (S4). As described above, the final state information of the previous signal can utilize the information stored in the buffer.

Next, the target state determination unit 320 determines whether a signal of a threshold value or more is found in the final state information of the previous signal (S5).

Next, the target state determination unit 320 determines that the target does not exist in the observation area if no information exceeding the threshold is found in the final state information of the previous signal (S6). If the target does not exist in the observation area, the weight applying unit 330 sets the weight as a first weight, which is a high weight (S7).

Alternatively, when the target state determination unit 320 finds a signal of a threshold value or more in the final state information of the previous signal, the target state determination unit 320 determines that the target is stationary in the observation region (S8). If the target has stopped in the observation area, the weight applying unit 330 sets the weight as a second weight having a low weight (S9).

Hereinafter, an experimental example for demonstrating a background difference method using an algorithm according to an embodiment of the present invention will be described. To demonstrate the performance of the algorithm, we used NVA-R640 from Novelda, a device that transmits and receives IR-UWB pulse waveforms. During a total of 512 frames of observation time, the target moves within the measurement range of 512 samples. The target starts from outside the measurement range, stops at around 180 samples, and then moves out again.

5 is a graph comparing the signal pulse magnitudes of the conventional method of the present invention and the background difference method according to an embodiment of the present invention.

FIG. 5A shows a case where a high weighting factor (? = 0.1) is applied in the conventional adaptive filter algorithm. Also, FIG. 5B shows a case where a low weight value (? = 0.005) is applied in the conventional adaptive filter algorithm. 5 (c) shows an example in which a differential weight is applied according to an embodiment of the present invention.

5 (a) and 5 (b) show the results of problems that may occur depending on the weight value of the background differential algorithm using the conventional adaptive filter.

Referring to FIG. 5A, if the weight is large, it is confirmed that the pulse size of the target gradually decreases as the current received signal information is reflected in the background difference calculation when the target stops. Conversely, referring to FIG. 5 (b), it can be seen that if the weight is small, the residual image of the stop position remains at a considerable size when the target is stopped and moved.

FIG. 5C is a graph of a result of applying the algorithm of the background difference method according to an embodiment of the present invention. When the target is out of the measurement range (noise only environment), it increases the weight to move quickly to respond to the change of the target. When the target stops, the weight is lowered so that the size of the pulse does not decrease. Let's do it.

FIG. 6 is a graph of a position tracking using an adaptive filter algorithm and an algorithm of a background difference method according to an embodiment of the present invention.

In order to compare the performance of the background difference algorithm using the conventional adaptive filter and the algorithm of the background difference method according to an embodiment of the present invention, the location tracking according to the movement of the target is performed as shown in FIG.

FIG. 6A shows a case where a high weighting factor (? = 0.1) is applied to the conventional adaptive filter algorithm. Also, FIG. 6B shows a case where a low weight (α = 0.005) is applied in the conventional adaptive filter algorithm. 6 (c) is an example in which a differential weight is applied according to an embodiment of the present invention.

Table 1 shows the variance of the distance error between the actual position Rn and the measured position Mn for the observation time frame n = (1, N) of the experimental equipment.

Existing algorithm
(α = 0.1) Existing algorithm
(α = 0.005) The proposed
algorithm

2616.34 4134.64 106.75

In the case of FIG. 6 (a), as compared with the motion of the actual target, it can be confirmed that correct measurement is not possible from about 200 frames after the stopping time, as the signal size becomes smaller.

In the case of FIG. 6 (b), the position information is stably maintained during the stop period due to the low weight, but the stop position remains as an obstacle to the position measurement. It can be confirmed that the movement information of the target can not be correctly tracked after about 400 frames.

On the contrary, in the case of FIG. 6 (c), which is an experimental result according to an embodiment of the present invention, it can be confirmed that correct position tracking is performed by applying the differential weight according to the target situation.

As described above, the present invention makes it possible to detect a target more precisely by differentiating the weights of the adaptive filter method, which is one of the background differential algorithms for observing the motion state of the target. The present invention proposes a background subtraction method capable of detecting a still object by correcting the current state information through the previous state information of the target in order to overcome the error of the position measurement due to the state information of the wrong target during the weight reflection process .

The specific acts described in the present invention are, by way of example, not intended to limit the scope of the invention in any way. For brevity of description, descriptions of conventional electronic configurations, control systems, software, and other functional aspects of such systems may be omitted. Also, the connections or connecting members of the lines between the components shown in the figures are illustrative of functional connections and / or physical or circuit connections, which may be replaced or additionally provided by a variety of functional connections, physical Connection, or circuit connections. Also, unless explicitly mentioned, such as " essential ", " importantly ", etc., it may not be a necessary component for application of the present invention.

The use of the terms " above " and similar indication words in the specification of the present invention (particularly in the claims) may refer to both singular and plural. In addition, in the present invention, when a range is described, it includes the invention to which the individual values belonging to the above range are applied (unless there is contradiction thereto), and each individual value constituting the above range is described in the detailed description of the invention The same. Finally, the steps may be performed in any suitable order, unless explicitly stated or contrary to the description of the steps constituting the method according to the invention. The present invention is not necessarily limited to the order of description of the above steps. The use of all examples or exemplary language (e.g., etc.) in this invention is for the purpose of describing the present invention only in detail and is not to be limited by the scope of the claims, It is not. It will also be appreciated by those skilled in the art that various modifications, combinations, and alterations may be made depending on design criteria and factors within the scope of the appended claims or equivalents thereof.

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 embodiments, but, on the contrary, Those skilled in the art will appreciate that various modifications and changes may be made thereto without departing from the scope of the present invention.

Accordingly, the spirit of the present invention should not be construed as being limited to the above-described embodiments, and all ranges that are equivalent to or equivalent to the claims of the present invention as well as the claims .

100: Surveillance camera
200: Network
300: background differential server
310: Target presence / absence determination unit
320: target state determination unit
330: weight application unit

Claims (6)

A target presence determining unit that determines that the target is moving within the observation area if the size of a current state signal based on an ultra wideband (UWB) communication is greater than or equal to a threshold value for movement of the target within an observation area;
A target state determiner for determining that the target exists in the observation area and stops if the size of the previous state signal based on the UWB communication is greater than or equal to a threshold value when the current state signal is less than or equal to a threshold value;
And a weight setting unit for setting a weight of an adaptive filter algorithm for differentiating the background in the observation area based on the information of the current state signal and the previous state signal,
Wherein the weight setting unit sets the weight as a first weight when it is determined that the target is moving in the observation area and sets the weight as a second weight lower than the first weight when it is determined that the target is stopped Set the background differential system. The method according to claim 1,
Wherein the target presence / absence determining unit determines that a target does not exist in the observation region if the size of the previous state signal is less than or equal to a threshold value,
Wherein the weight setting unit sets the weight to the first weight when it is determined that no target exists in the observation area. delete delete delete delete

Images