KR101585022B1 - Streaming Data Analysis System for Motion Detection in Image Monitering System and Streaming Data Analysis Method for Motion detection - Google Patents

Abstract

According to the present invention, a streaming information analysis method used by an image monitoring system includes the steps of: generating syntax information in a compression bit-stream received from a camera; determining whether the type of a macro block in the syntax information is intra or inter; comparing the coefficient bit amount of the corresponding macro block with a threshold value, when the type of the macro block is intra; and determining an occurrence of movement in the corresponding block when the coefficient bit amount of the corresponding macro block exceeds the threshold value. The streaming information analysis system used by the image monitoring system includes: a camera compressing and transmitting the image information generated by recording an object in a monitoring area; a streaming information generation unit which generates the syntax information from the compression bit-stream received from the camera; a streaming information analysis unit using the syntax information generated by the streaming information generation unit to determine the movement of the object.

Description

TECHNICAL FIELD The present invention relates to a streaming information analysis system for motion detection in a video surveillance system and a streaming information analysis method for motion detection in a video surveillance system,

The present invention relates to a video surveillance system using a camera to detect and store motion by decoding a compressed stream. In general, the video surveillance system monitors the intruder's movements in a certain surveillance area. The surveillance system detects the presence of the intruder by receiving the photographed image and determining whether the subject is moving. Therefore, It has a very important meaning. In general, a video surveillance system decides whether a moving image is compressed by decoding and decrypting an image compressed and transmitted from a plurality of cameras in order to reduce storage space, and stores moving image information when it is determined to be a moving image.

A conventional technique related to the present invention is disclosed in Korean Patent Registration No. 10-1416957 (published on Apr. 201, 07, 09). FIG. 1 is a block diagram of a motion analysis method using the conventional video recording apparatus and the SVC video stream. 1, a conventional motion picture analysis method using an image recording apparatus and an SVC video stream includes a stream storage unit 110 for storing an SVC video stream received from at least one video camera, an SVC video stream received from at least one video camera, A stream selector 130 for selecting and outputting a video stream of a specific resolution or a specific frame rate from the SVC video stream stored in the SVC video stream storage 110, a decoder 130 for decoding the video stream output from the stream selector 130, Based on the analysis results of the object analyzing unit 170 and the object analyzing unit 170 for analyzing the motion or shape of the object in the extracted image, the object extracting unit 150 extracting the moving object from the decoded image, A stream selector 130 for controlling the operation of the stream selector 130 by determining the required resolution or frame rate, It will, which may include a section 190. In addition, the conventional image recording apparatus is for performing storage, display, image analysis, and the like of an image received from a video camera, and may be in the form of a DVR (Digital Video Recorder) or an NVR (Network Video Recorder). On the other hand, at least one video camera connected to the video recording apparatus via a network is installed at a plurality of locations and encodes the captured video into an SVC video stream to transmit the SVC video stream through the network. In addition, the stream storage unit 110 stores an SVC video stream received from at least one video camera. The stream storage unit 110 may store various types of storage media such as ROM, flash memory, RAM, Lt; / RTI > In addition, the SVC video stream received from one or more video cameras or the SVC video stream stored in the stream storage unit 110 may be selected by the stream selector 130 according to a specific resolution or a specific frame rate, The video decoder 130 separates the SVC video stream by resolution or frame rate, and can select and output a video stream having a specific resolution or a specific frame rate. In addition, the object extraction unit 150 may extract a moving object by decoding the video stream output from the stream selection unit 130. In this case, the object extraction may be performed by various methods including Background Subtraction Can be performed. In addition, the object analyzing unit 170 can analyze the motion or shape of the object in the extracted image. More specifically, when the object extracted by the object extracting unit 150 is extracted, The motion pattern of the object, the direction of movement, the velocity, and the shape of the object. When a detailed analysis of the analysis result of the object analysis unit 170 is required, the object analysis unit 170 can request a stream of a high-resolution or high frame rate to the stream selection control unit 190, 190 request the precise analysis of the SVC video stream received from at least one video camera or the SVC video stream stored in the stream storage unit 110 when a video stream of a high resolution or a high frame rate is requested from the object analyzing unit 170 It is possible to control the stream selection unit 130 such that a high-resolution or high-frame-rate video stream is output from the SVC video stream. For example, when the video camera is a security surveillance camera, the object extraction unit 150 can extract a moving object by decoding a low-resolution video stream of the SVC video stream received from a plurality of surveillance cameras. Then, the object analyzing unit 170 can analyze the motion pattern, the moving direction, the speed, and the shape of the object of the extracted object. If the moving object is determined to be a human, it is difficult to distinguish the face in the low resolution image , The object analysis unit 170 can request a high resolution image to the stream selection control unit 190. In addition, when the speed of the moving object is very fast, the object analyzing unit 170 requests the stream selecting unit control unit 190 for a high frame rate image. When the video camera is used for the overspeed camera, The object analyzing unit 170 extracts a moving object from the input unit 150 and analyzes the shape of the extracted object to determine whether it is an automobile. In this case, if the extracted object is determined as an automobile, the object analyzing unit 170 can analyze the speed of the automobile. If the analyzed speed of the automobile is equal to or higher than a predetermined speed, It can be requested to be selected. On the other hand, the resolution or the frame rate of the video stream outputted from the stream selector 130 can be set by the user depending on the use.

In the conventional technique as described above, image data transmitted from a camera is usually in the form of a compressed bitstream, so image decoding must be performed to obtain a pixel value. Therefore, in the conventional art as described above, image decoding requires a great deal of computation, which increases the load on the system. Accordingly, an object of the present invention is to detect and store motion by analyzing streaming information without decoding a compressed stream received from a camera.

According to another aspect of the present invention, there is provided a method for analyzing streaming information, comprising the steps of: generating syntax information in a compressed bitstream received from a camera; determining whether a macroblock type is Intra or Inter, Comparing a coefficient bit amount of the corresponding macroblock with a threshold value if the macroblock type is Intra, and generating a motion in the corresponding block if the coefficient bit amount of the corresponding macroblock exceeds a threshold value And a step of judging whether or not there is a difference between the measured values. According to another aspect of the present invention, there is provided a streaming information analysis system comprising: a camera for compressing and transmitting image information of an object in a surveillance area; a streaming information generator for generating syntax information in a compressed bitstream received from a camera; And a streaming information analyzing unit for analyzing the motion of the object using the syntax information generated by the streaming information generating unit.

In the streaming information analysis system for motion detection and the streaming information analysis method for motion detection in the video surveillance system of the present invention configured as described above, since the motion of the object is determined based only on the syntax information included in the received compressed stream, the decoding process is omitted It is possible to drastically reduce the burden on the system for computation. Further, another effect of the present invention is to solve the problem of insufficient storage space of the system because only the motion image can be stored.

1 is a block diagram of a conventional video recording apparatus,
FIG. 2 is a flow chart of a motion analysis process using a conventional SVC bitstream,
FIG. 3 is a block diagram of a decoding structure of an H.264 codec according to an embodiment of the present invention.
FIG. 4 is a block diagram of a streaming information analysis system for motion detection in the video surveillance system of the present invention,
FIG. 5 is a flowchart illustrating a method for analyzing streaming information for motion detection in a video surveillance system according to the present invention.
FIG. 6 is a flowchart illustrating a process of determining whether a motion is generated in a macroblock using syntax information applied to the present invention.
FIG. 7 is a diagram showing eleven neighboring macroblocks and an 8-way motion vector category compared with a current macroblock for checking temporal / spatial correlation.

A streaming information analysis system for motion detection and a streaming information analysis method for motion detection in the video surveillance system of the present invention having the above-described objects will be described with reference to FIG. 3 to FIG.

FIG. 3 is a block diagram of a decoding structure of an H.264 codec according to an embodiment of the present invention. 3, the streaming information generation for generating whether or not a motion of a macroblock is generated generates streaming information through an entropy decoding unit 20 and a syntax decoding unit 30, and performs inverse quantization, inverse transform, The entry prediction, the motion compensation, the deblocking filter, and the frame buffer are not executed. Therefore, the complex computation process is omitted, and the load of the system is drastically reduced, and the problem of the lack of the system storage space can be solved.

4 is a block diagram of a streaming information analysis system for motion detection in the video surveillance system of the present invention. 4, the streaming information analysis system for motion detection in the video surveillance system of the present invention is applicable to the H.264 codec, and includes a camera 10 for compressing and transmitting image information of an object in a surveillance region, And a streaming information analysis unit 70 for determining the motion of the object using the syntax information generated by the streaming information generation unit 40. The streaming information generation unit 40 generates syntax information in the compressed bitstream received from the streaming information generation unit 40, It is characterized by. The streaming information generator 40 includes an entropy decoder 20 for decoding a compressed bitstream and a syntax decoder 30 for decoding syntax information in a decoded bitstream.

5 is a flowchart illustrating a method of analyzing streaming information for motion detection in the video surveillance system of the present invention. 5, the method for analyzing streaming information for motion detection in the video surveillance system of the present invention includes generating (S11) syntax information in a compressed bitstream received from a camera (S11) And determining whether a motion of a macroblock is generated (S12).

FIG. 6 is a control flowchart illustrating a step of determining whether a motion is generated in a macroblock using syntax information applied to the present invention. 6, the step of determining whether a motion is generated in a macroblock using the syntax information according to the present invention includes determining whether a macroblock type is Intra in syntax information (S21) (S22) comparing whether a coefficient bit amount of the corresponding macroblock is larger than a threshold value IntraTh if the coefficient bit amount of the corresponding macroblock is greater than a threshold value; S23). ≪ / RTI > If it is determined in step S21 that the type of the macroblock is not Intra, it is determined to be Inter and it is determined whether the coefficient bit amount is larger than the threshold value InterTh (S31). If the coefficient bit amount is larger than the threshold InterTh in step S31, (S32) of determining whether a size (MVSize) is larger than a threshold value MVTh, and a step (S32) of determining whether a spatial-temporal correlation exists when a motion vector size (MVSize) is larger than a threshold value MVTh If there is temporal spatial correlation in step S33 and in step S33, it is determined that motion in the macroblock has occurred (S23) and the process ends. If the coefficient bit amount is smaller than the threshold InterTh in step S31 or the motion vector size MVSize is smaller than the threshold MVTh in step S32 or there is no spatial-temporal correlation in step S33, It is determined that there is no motion in the block.

In the above, IntraTh and InterTh can be expressed as the following Equation (1).

-1), where

= QP : 1< QPIntraTh = Scaleparam x SenLevel / (

-1), where

= QP: 1 < QP

-SenLevel) x 2, where

=2 : 2>QP (1) InterTh = 51 x 2 - (

-SenLevel) x 2, where

= 2: 2 > QP (1)

In Equation (1), QP is a quantization coefficient, and Scaleparam is an adjustment value for determining a threshold value, which is determined to be 3,600 in the present invention. Also, SenLevel means motion detection sensitivity and has an integer value from 1 to 10. The SenLevel can be changed in consideration of the complexity of a camera image to be sensed, the size of a minimum motion to be sensed, Coefficient.

The motion vector size MVSize is determined by the sum of the horizontal axis absolute magnitude MVx and the vertical axis magnitude MVy of the motion vector as shown in Equation (2) below.

(2)

(2)

The threshold value MVTh is a value determined by the motion detection sensitivity SenLevel and can be determined as shown in Table 1 in the present invention.

SenLevel One 2 3 4 5 6 7 8 9 10 MVTh 54 74 110 160 234 316 448 662 926 1240

As described above, when the coefficient bit amount of the macroblock is larger than the threshold InterTh and the size of the motion vector is larger than MVTh, it is checked whether the macroblock is spatiotemporally correlated. The decision conditions for checking the spatio-temporal correlation of the corresponding macroblock are as follows: 1. The motion vector category should be the same as that of the neighboring macroblock; 2. The neighboring macroblock and the motion vector size (MVSize) should be similar; 2. If there are more than 6 motion-macroblocks satisfying 2., it is determined that there is a temporal-spatial correlation.

FIG. 7 is a diagram showing eleven neighboring macroblocks and an 8-way motion vector category compared with a current macroblock for checking temporal / spatial correlation. In FIG. 7A, a total of eleven macroblocks including four neighboring macroblocks belonging to the current frame and colocated macroblocks belonging to the previous frame and six neighboring macroblocks are compared, and the temporal / spatial correlation determination It is determined by conditions. 7B shows the 8-direction motion vector category. If the motion vector category of the current macroblock and the motion vector category of the current macroblock are the same and the motion vector size (MVSize) is similar, It is determined that the motion has the same motion as the macroblock. In the present invention, when six or more of the 11 neighboring macroblocks have the same motion as the current macroblock, it can be determined that the current macroblock has temporal correlation. In addition, whether or not the motion vector sizes of the neighboring macroblocks compared with the current macroblocks are similar to each other is determined by the following equation (3) and it is determined that they are similar if they satisfy the following equation (3) will be.

(3)

(3)

In the above equation (3), MVSize _curr Denotes the motion vector size of the current macroblock, MVSiz &lt; RTI ID = 0.0 &gt; _adj Represents the motion vector size of the comparison macroblock. Also, MVScale is a value determined by the motion detection sensitivity SenLevel, and can be determined and applied as shown in Table 2 in the present invention.

SenLevel One 2 3 4 5 6 7 8 9 10 MVScale 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0.1

As described above, the present invention can detect the occurrence of motion in a specific frame or a specific area of an image based on a motion generation confirmation method of a macroblock. That is, when a part or whole area of an image is designated as a motion detection area according to a user's input setting in a video surveillance system, it is checked whether a motion of a macroblock included in the corresponding area occurs or not, It can be judged that the motion of the user has occurred.

10: camera, 20: entropy decoding unit,
30: syntax decoding unit, 40: syntax information generating unit,
70: streaming analysis unit, 110: stream storage unit,
130: stream selection unit, 150: object extraction unit,
170: Object analysis unit, 190: Stream selection control unit

Claims (22)

delete delete delete A streaming information analysis system for motion detection of a video surveillance system,
The streaming information analysis system for motion detection comprises:
A camera 10 for compressing and transmitting image information of an object in a surveillance area;
An entropy decoding unit 20 for decoding the streaming information without executing the inverse quantization, inverse transform, entry prediction, motion compensation, deblocking filter, and frame buffer for the compressed bitstream by generating the syntax information in the compressed bitstream received from the camera And a syntax decoding unit (30) for decoding the syntax information in the decoded bit stream.
And a streaming information analyzing unit 70 for determining the motion of the object based on the syntax information generated by the streaming information generating unit. The streaming information analyzing unit 70 analyzes the streaming information of the intra- And if the coefficient bit amount of the corresponding macroblock is larger than the threshold value, if the coefficient bit amount of the corresponding macroblock is larger than the threshold value, it is determined that the motion is generated in the corresponding macroblock , Determines whether the coefficient bit amount is larger than the threshold value InterTh when the macroblock type is Inter among the generated syntax information. If the motion vector amount MVSize is larger than the threshold value MVTh when the coefficient bit amount is larger than the threshold value InterTh, And if the motion vector size (MVSize) is greater than the threshold value (MVTh) Temporal correlation between the macroblock and the macroblock, and determines that a motion in the macroblock occurs when temporal / spatial correlation exists.
5. The method of claim 4,
The video surveillance system includes:
Wherein the H.264 codec is a streaming information analysis system for motion detection.
5. The method of claim 4,
The streaming information analysis system for motion detection comprises:
When a part or whole area of an image is designated by a user's input setting in the video surveillance system, it is checked whether or not motion of the macroblocks included in the corresponding area is generated, and it is judged that motion has occurred in the corresponding area Wherein the streaming information analyzing system comprises:
delete delete delete A streaming information analysis method for motion detection of a video surveillance system,
The method for analyzing streaming information for motion detection,
Decoding streaming information without performing inverse quantization, inverse transform, entry prediction, motion compensation, deblocking filter, and frame buffer in a compressed bitstream received from a camera;
Reconstructing syntax information in the decoded bitstream;
Determining whether the macroblock type is Intra or Inter in the syntax information (S21);
If the macroblock type is Intra, comparing (S22) whether the coefficient bit amount of the corresponding macroblock is larger than a threshold value;
Determining that motion has occurred in the macroblock when the coefficient bit amount of the corresponding macroblock is larger than a threshold value (S23);
If it is determined that the type of the macroblock is not Intra (S21), it is determined that Inter is greater than a threshold (InterTh) (S31);
(S32) determining whether the motion vector magnitude (MVSize) is larger than the threshold value (MVTh) when the coefficient bit amount is greater than the threshold value (InterTh) in step S31;
If it is determined in step S32 that the motion vector magnitude MVSize is greater than the threshold value MVTh, it is checked whether there is a spatial-temporal correlation.
And determining (S23) that a motion in a macroblock occurs when temporal / spatial correlation exists in step S33 (S23).
Here, MVSize is the size of the motion vector, MVx is the absolute value of the horizontal axis of the motion vector, and MVy is the absolute value of the vertical axis of the motion vector.
11. The method of claim 10,
The threshold value may be set to &
IntraTh = Scaleparam x SenLevel / (

-1), where

= QP: 1 &lt; QP
The method of claim 1, further comprising:
Where QP is the quantization coefficient, Scaleparam is the adjustment value for determining the threshold value, and SenLevel is the motion detection sensitivity.
12. The method of claim 11,
The SenLevel,
The SenLevel may be changed in consideration of the complexity of the camera image to be sensed, the size of the minimum motion to be sensed, and the sensing period, A method for analyzing streaming information for motion detection.
Where SenLevel is the motion detection sensitivity.
12. The method of claim 11,
In the QP,
3600. A method for analyzing streaming information for motion detection.
Where QP is the quantization factor.
12. The method of claim 11,
The threshold value may be set to &
InterTh = 51 x 2 - (

-SenLevel) x 2, where

= 2: 2 &gt; QP
The method of claim 1, further comprising:
Where QP is the quantization factor.
12. The method of claim 11,
The method for analyzing streaming information for motion detection,
If the coefficient bit amount is smaller than the threshold value InterTh in step S31 or the motion vector size MVSize is smaller than the threshold value MVTh in step S32 or a spatial- temporal correlation process is performed in step S33, If there is no motion, it is determined that there is no motion in the macroblock.
Where MVSize is the motion vector size and MVTh is the threshold.
12. The method of claim 11,
The step (S33) of confirming whether there is a spatial-temporal correlation exists,
1. Motion vector category should be the same as neighboring macroblock. 2. Motion vector size and MVSize should be similar to each other. 3. Motion macroblocks satisfying 1 and 2 should be 11 And if the number of macroblocks is 6 or more, it is determined that temporal and spatial correlation exists.
15. The method of claim 14,
The SenLevel,
The SenLevel may be changed in consideration of the complexity of the camera image to be sensed, the size of the minimum motion to be sensed, and the sensing period, A method for analyzing streaming information for motion detection.
Where SenLevel is the motion detection sensitivity.
15. The method of claim 14,
In the QP,
3600. A method for analyzing streaming information for motion detection.
Where QP is the quantization factor.
16. The method of claim 15,
The motion vector size (MVSize)

The method comprising the steps of: (a)
Here, MVSize is the size of the motion vector, MVx is the absolute value of the horizontal axis of the motion vector, and MVy is the absolute value of the vertical axis of the motion vector.
16. The method of claim 15,
The threshold value (MVTh)
MVTh = 54 if SenLevel = 2, MVTh = 74 if SenLevel = 2, MVTh = 110 if SenLevel = 3, MVTh = 160 if SenLevel = 4 and SenLevel = 5 if SenLevel = If MVTh = 234, and if SenLevel is 6, MVTh = 316. If SenLevel is 7, MVTh = 448 and if SenLevel is 8, MVTh = 662. If SenLevel is 9, MVTh = 926 and if SenLevel is 10, The method comprising the steps of:
Where MVTh is the threshold value and SenLevel is the motion detection sensitivity.
17. The method of claim 16,
Whether or not the motion vector size (MVSize) is similar to the neighboring macroblock,

(MVSize _curr represents the motion vector size of the current macroblock, and MVSize _adj represents the motion vector size of the comparison macroblock)
The method of claim 1, further comprising:
Here, MVScale is a value determined by the motion detection sensitivity SenLevel.
22. The method of claim 21,
The MVScale includes:
MVScale = 0.5 if SenLevel is 2, MVScale = 0.4 if SenLevel is 2, MVScale = 0.4 if SenLevel is 3, MVScale = 0.3 if SenLevel is 4, SenLevel = 5 if SenLevel is 4, If MVScale = 0.2, MVScale = 0.2 if the SenLevel is 8, MVScale = 0.1 if the SenLevel is 9, MVScale = 0.1 if the SenLevel is 8. If MVScale = The method of claim 1, further comprising:
Here, MVScale is a value determined by the motion detection sensitivity SenLevel.

Images