WO2019124634A1

WO2019124634A1 - Syntax-based method for object tracking in compressed video

Info

Publication number: WO2019124634A1
Application number: PCT/KR2018/002556
Authority: WO
Inventors: 이성진; 정승훈; 배현성; 이현우
Original assignee: 이노뎁 주식회사
Priority date: 2017-12-20
Filing date: 2018-03-04
Publication date: 2019-06-27
Also published as: KR20190074900A; KR102015082B1

Abstract

The present invention relates to a technology which extracts from a compressed video, generated by a CCTV camera for example, a region having meaningful movement, that is, a moving object region, in image block units of the compressed video on the basis of syntax, such as motion vectors and coding types, without having to resort to complex image processing as in the conventional technology, and tracks and provides, normalized, movement of the moving object region to a video surveillance system. According to the present invention, a moving object region in a CCTV-filmed video can be rapidly tracked without having to resort to complex processing such as decoding, down-scale resizing, differential image acquisition, image analysis, and the like of a compressed video, and thus anti-crime effect and post-crime evidence gathering by means of a video surveillance system can be improved. The result of moving object tracking is provided to the video surveillance system after mathematical normalization, and thus compatibility with a variety of equipment can be assured.

Description

A syntax-based object tracking method for compressed images

The present invention relates generally to techniques for effectively performing object tracking from compressed images such as H.264 AVC and H.265 HEVC.

More specifically, the present invention relates to a method and an apparatus for detecting a moving image of a compressed image generated by a CCTV camera, that is, an area in which there is a significant motion, Based on a syntax such as a motion vector and a coding type, and tracking motion of the moving object region to provide normalization to a video control apparatus.

In recent years, it is common to construct a video surveillance system that uses CCTV for crime prevention and post-evacuation. With many CCTV cameras installed in each region, the images generated by these CCTV cameras are displayed on the monitor and stored in the storage device. When the controller finds a scene where a crime or accident occurs, it immediately responds appropriately and, if necessary, retrieves the images stored in the storage for subsequent evidence.

By the way. It is a reality that the number of control personnel is very short compared with the installation status of CCTV cameras. To effectively perform video surveillance with such a limited number of people, it is not sufficient to simply display the CCTV image on the monitor screen. It is preferable to detect the movement of an object existing in each CCTV image and to process the CCTV image so as to be effectively detected by adding something to the corresponding area in real time. In this case, the controller does not monitor the entire CCTV image with a uniform attention, but can monitor the CCTV image centered on the object motion.

On the other hand, the image sensing system adopts compressed image for efficiency of storage space. In recent years, as the number of CCTV cameras is rapidly increasing and high-definition cameras are installed, complicated image compression techniques of high compression ratio such as H.264 AVC and H.265 HEVC are adopted.

In a camera apparatus for generating moving image data, a compressed image is generated according to one of these technical specifications, and the apparatus for reproducing the moving image receives the compressed image, and if the compressed image is received, As shown in FIG. Conventionally, in order to determine the presence or absence of movement in a CCTV image to which an image compression technique is applied, a process of decoding a compressed image to obtain a reproduced image, that is, an original image in which a decompressed image has been obtained, is then processed.

1 is a block diagram showing a general configuration of a moving picture decoding apparatus according to the H.264 AVC technical standard. Referring to FIG. 1, a moving picture decoding apparatus according to H.264 AVC includes a syntax analyzer 11, an entropy decoder 12, an inverse transformer 13, a motion vector calculator 14, a predictor 15, a deblocking filter 16).

These hardware modules process compressed images sequentially, decompress them, and restore the original image data. At this time, the parser 11 parses the motion vector and the coding type for the coding unit of the compressed image. Such a coding unit is generally an image block such as a macroblock or a sub-block, but may be implemented not exactly in accordance with a technical standard.

2 is a flowchart showing a process of performing object tracking from a compressed image in a conventional image analysis solution.

Referring to FIG. 2, in the conventional art, a compressed image is decoded according to H.264 AVC and H.265 HEVC (S10), and the frame images of the reproduced image are downscaled to a small image, for example, 320x240 (S20). At this time, downscaling is performed to reduce the processing burden in the subsequent process. Then, differential images are obtained for the resized frame images, and the moving object is extracted through the image analysis (S30). Then, a moving path of the moving object is identified through an image analysis on a series of frame images (S40).

To extract moving objects in the prior art, compressed image decoding, downscaled resizing, and image analysis are performed. These are very high complexity processes, and therefore, the capacity of simultaneous processing of one image analysis server is considerably limited in the conventional video control system. Currently, the maximum CCTV channel that can be covered by a high performance image analysis server is typically a maximum of 16 channels. Since a large number of CCTV cameras are installed, a number of image analysis servers are required in the video monitoring system, which causes problems such as increase in cost and difficulty in securing physical space.

It takes a considerable amount of budget to construct and maintain a large-scale video control system, and a corresponding utility value is required. The basic direction of such a demand is to prevent crime and secure evidence of crime. Accordingly, it provides a high-level detection function that allows the video control system to detect a specific situation in which the experience is itself a problem, in addition to simply recording the surrounding scene or informing the existence of the moving object from the scene. There is a need. At this time, efficient implementation techniques are also required in view of the realistic problems of system construction cost and physical space.

It is an object of the present invention to provide a technique for effectively performing object tracking from compressed images such as H.264 AVC and H.265 HEVC.

Particularly, it is an object of the present invention to provide an image processing apparatus and a method for processing a moving image in a region where there is a significant motion for a compressed image generated by, for example, a CCTV camera, Extracting the motion vector based on a syntax such as a motion vector and a coding type, and tracking motion of the moving object region to provide normalization to the video control device.

According to an aspect of the present invention, there is provided a method of tracking a syntax-based object for a compressed image, the method comprising: a first step of obtaining a motion vector and a coding type for a coding unit by parsing a bitstream of the compressed image; A second step of acquiring a motion vector accumulation value for a first time preset for each of a plurality of image blocks constituting a compressed image; A third step of comparing the accumulated value of the motion vector with a preset first threshold value for a plurality of image blocks; A fourth step of marking an image block having a motion vector accumulation value exceeding a first threshold value as a moving object region; A series of coordinates of the tracking object moving object region is acquired over a series of image frames of the compressed image in relation to the moving object region (hereinafter referred to as the 'tracking object moving object region') specified by the user operation as a tracking object, And a fifth step of providing the coordinate sequence to the video control device.

In the present invention, an image block constituting a compressed image may include a macro block and a sub-block.

In this case, the fifth step is a step 5a for newly issuing and allocating a unique ID when the moving object area is in an ID unassigned state; A step 5b of setting a specific moving object area (hereinafter, referred to as 'tracking object moving object area') as a tracking object according to a user operation; A fifth step c) of identifying a unique ID (hereinafter, referred to as 'tracking object unique ID') allocated to the tracking object moving object area; A fifth step of sequentially calculating square coordinates of a moving object area to which a tracking object unique ID value is assigned for a series of image frames constituting a compressed image and setting the coordinates as a coordinate sequence of the tracking object moving object area; A fifth step of normalizing the series of rectangular coordinates included in the coordinate sequence of the tracking object moving object region according to the resolution of the compressed image; A fifth step of providing a coordinate sequence of the moving object region to be traced, which has been normalized, to a video control apparatus; And a fifth step (g) of revoking the assigned unique ID when the moving object area disappears in the series of image frames.

In this case, the rectangular coordinates include upper left coordinates (x, y), horizontal axis length (dx), and vertical axis length (dy) of a rectangle formed to virtually include the moving object region, The upper left x coordinate and the horizontal axis dx are divided by the horizontal resolution x_res of the compressed image to reflect the horizontal resolution x_res and the vertical resolution y_res, It is preferable to perform division processing with the resolution (y_res).

Also, an object tracking method according to the present invention includes: a) a step of identifying a plurality of adjacent image blocks (hereinafter, referred to as 'neighboring blocks') around a moving object region; Comparing a motion vector value with a predetermined second threshold value for a plurality of neighboring blocks; Further comprising: marking a neighboring block having a motion vector value exceeding a second threshold as a moving object region; D) marking a neighboring block having a coding type of an intra picture among a plurality of neighboring blocks as a moving object region; Marking a predetermined number or less of unmarked image blocks surrounded by the moving object area as a moving object area by performing interpolation on the plurality of moving object areas.

Meanwhile, a computer-readable nonvolatile recording medium according to the present invention records a program for executing a syntax-based object tracking method on a compressed image, such as the above, in a computer.

According to the present invention, since the moving object region is extracted from the CCTV image without performing the complicated processing such as decoding, downscaling resizing, differential image acquisition, and image analysis on the CCTV compressed image, There is an advantage that performance improvement can be obtained.

Also, according to the present invention, it is possible to quickly track a moving object area in a CCTV shot image without performing complex processing such as decoding, downscaling, differential image acquisition, and image analysis on a compressed image, And the effect of securing the prevention and follow-up evidence. At this time, the moving object tracking result is mathematically normalized and then provided to the video control device, thereby ensuring compatibility with various equipment.

1 is a block diagram showing a general configuration of a moving picture decoding apparatus;

2 is a flowchart showing a process of performing object tracking from a compressed image in the prior art;

3 is a flowchart showing an overall process of performing object tracking from a compressed image in accordance with the present invention;

4 is a flowchart showing an embodiment of a process of detecting valid motion from a compressed image in the present invention.

5 is a diagram illustrating an example of a result of applying a valid motion region detection process according to the present invention to a CCTV monitoring screen.

Figures 6 and 7 are partially enlarged views of the main part of Figure 5;

FIG. 8 is a flowchart illustrating an example of a process of detecting a boundary region for a moving object region in the present invention. FIG.

9 is a view showing an example of a result of applying a boundary region detection process according to the present invention to a compressed image.

Figs. 10 and 11 are partially enlarged views of the main part of Fig. 9; Fig.

FIG. 12 is a diagram illustrating an example of a result of summarizing a moving object region through interpolation in the present invention; FIG.

Figs. 13 and 14 are partially enlarged views of the main part of Fig. 12; Fig.

15 is a flowchart showing an embodiment of a process of tracking and identifying a tracking object moving object region specified by a user from a compressed image in the present invention.

16 is a diagram illustrating an example in which a unique ID is assigned to a moving object area in the present invention.

17 is a diagram illustrating an example in which rectangular coordinates are set in a moving object area in the present invention.

Hereinafter, the present invention will be described in detail with reference to the drawings.

3 is a flow chart illustrating the overall process of performing object tracking from a compressed image in accordance with the present invention. The object tracking process according to the present invention may well perform an image analysis server in a system for handling a series of compressed images, such as a CCTV image control system.

In the present invention, a bitstream of a compressed image is parsed without decoding a compressed image, and syntax information such as a macroblock and a sub-block, preferably a motion vector, And the coding type (Coding Type) information. The obtained moving object area does not accurately reflect the boundary line of the moving object as shown in the image attached to this specification, but has a high processing speed and high reliability. Then, in the present invention, an operation of tracking a specific moving object area designated by the controller based on the obtained moving object area is performed.

According to the present invention, the moving object region can be extracted and the object tracking can be performed without decoding the compressed image. However, the apparatus or software to which the present invention is applied should not perform the operation of decoding the compressed image, and the scope of the present invention is not limited thereto.

Hereinafter, the concept of a process of tracking an object from a compressed image according to the present invention will be described with reference to FIG.

Step S100: First, an effective motion that is substantially meaningful from the compressed image is detected based on the motion vector of the compressed image, and the image area in which the valid motion is detected is set as the moving object area.

To do this, the motion vector and coding type of the coding unit of the compressed image are parsed according to a moving picture compression standard such as H.264 AVC and H.265 HEVC. At this time, the size of the coding unit is generally 64 x 64 pixels to 4 x 4 pixels and can be set to be flexible.

Accumulates the motion vectors for a predetermined time period (e.g., 500 msec) for each image block, and checks whether the accumulated motion vector accumulated value exceeds a preset first threshold value (e.g., 20). If such an image block is found, the valid motion is found in the corresponding image block and marked as a moving object area. Accordingly, even if a motion vector occurs, if the cumulative value for a predetermined time does not exceed the first threshold value, it is assumed that the image change is negligible and ignored.

Step S200: Next, the boundary region is detected based on the motion vector and the coding type for the moving object region detected in the previous step (S100). For this purpose, when a plurality of image blocks neighboring the image block marked as the moving object region are examined and the motion vector is generated over a second threshold value (for example, 0) or the coding type is an intra picture, Mark the block as a moving object area. In this process, the image block is substantially a block of the moving object area detected in step S100.

If an effective motion is found and the image block has some motion in the vicinity of the moving object area, it is marked as a moving object area because it is likely to be a lump with the previous moving object area. In addition, in the case of intra-picture, since there is no motion vector, determination based on motion vectors is impossible. Accordingly, the intra picture adjacent to the image block already detected as the moving object region is estimated as a lump together with the previously extracted moving object region.

Step S300: Interpolation is applied to the moving object area detected in the previous steps S100 and S200 to arrange the fragmentation of the moving object area. In the previous process, since it is judged whether or not the moving object region is determined in units of image blocks, in reality, there is an image block which is not marked as the moving object region in the middle even though it is one moving object (for example, As shown in FIG.

Accordingly, if one or a small number of non-marking image blocks are surrounded by a plurality of image blocks marked with the moving object area, they are further marked as a moving object area. In this way, it is possible to make a plurality of divided moving object areas to be united. The effect of such interpolation can be clearly shown by comparing FIG. 9 and FIG.

Step S400: The moving object region is quickly extracted based on the syntax (motion vector, coding type) of the coding unit for the compressed image through the above process. In step S400, when the control agent requests the tracking of the object while specifying the moving object area through a mouse operation or the like on the monitor screen using the extracted result of the moving object area, the moving path of the moving object area in the compressed image Track it. Since this process is important for real time, the present invention can be preferably applied. In the video control system, if there is something that looks suspicious in the judgment of the control personnel, it is aimed to increase the crime prevention effect by tracking and marking it. In addition, object tracking information can be useful in securing evidence.

To this end, in the present invention, a series of image frames (for example, a sequence of 30 frames per second) of a compressed image in association with a moving object region (hereinafter referred to as a 'tracking object moving object region' To acquire a series of coordinates for the tracked moving object region over the tracked moving object region. Then, a series of coordinates thus obtained, that is, a coordinate sequence, is provided to the video control device as a tracking result for the moving object area. In the video control apparatus, the moving object region corresponding to the corresponding coordinates in each image frame is displayed and displayed on the controller in a prominent manner.

A specific process of performing object tracking from the compressed image will be described later in detail with reference to FIG.

FIG. 4 is a flowchart illustrating an embodiment of a process for detecting valid motion from a compressed image in the present invention, and FIG. 5 is a view illustrating an example of a result of applying the effective moving area detection process according to the present invention to a CCTV monitoring screen.

Step S110: First, the coding unit of the compressed image is parsed to obtain a motion vector and a coding type. Referring to FIG. 1, the moving picture decoding apparatus performs a syntax analysis (header parsing) and a motion vector operation on a stream of a compressed image according to a moving picture compression standard such as H.264 AVC and H.265 HEVC. Through this process, the motion vector and the coding type are parsed for the coding unit of the compressed image.

Step S120: The motion vector accumulation value for a preset time (for example, 500 ms) is obtained for each of the plurality of image blocks constituting the compressed image.

This step is presented with the intent to detect any valid motion that is substantially meaningful from the compressed image, such as a running car, a runner, or a crowd fighting with each other. The shaking leaves, the ghost appearing for a while, and the shadows that change slightly due to the reflection of light are prevented from being detected because they are moving objects, but they are meaningless objects.

To this end, the motion vector accumulation value is obtained by accumulating the motion vectors in units of one or more image blocks for a preset predetermined time (for example, 500 msec). At this time, the image block is used as a concept including a macro block and a sub-block.

Steps S130 and S140: The motion vector accumulation value is compared with a preset first threshold value (e.g., 20) for a plurality of image blocks, and an image block having a motion vector accumulation value exceeding the first threshold value, Lt; / RTI >

If an image block having a cumulative motion vector value of more than a predetermined value is found, a significant motion, that is, a valid motion is detected in the corresponding image block, and is marked as a moving object region. For example, in a video control system, the degree of movement is such that the control personnel are worthy of interest. On the contrary, if the cumulative value for a predetermined time period is small enough to not exceed the first threshold value even if a motion vector occurs, the change in the image is estimated to be insignificant and insignificant, and ignored in the detection step.

Step S150: The moving object region is displayed on the reproduction screen of the compressed image so as to be distinguished from the general image. FIG. 5 is a diagram illustrating an example of a result of applying a valid motion region detection process to a CCTV monitoring screen. In FIG. 5, a plurality of image blocks indicating a cumulative motion vector value exceeding a first threshold value are marked as a moving object region, Line box. Figs. 6 and 7 are enlarged views of main parts in Fig. 5. Fig.

5 to 7, the sidewalk block, the road, and the shadowed portion are not displayed as the moving object area, while the walking people and the traveling car are displayed as the moving object area. In this specification, the moving object region is represented by a thick line block, but it is more preferable that the CCTV monitor screen expresses the moving object region in a color that the controller can identify immediately.

FIG. 8 is a flowchart illustrating an embodiment of a process of detecting a boundary region for a moving object region in the present invention. FIG. 9 is a diagram illustrating an example of a result of applying a boundary region detection process to a compressed image. 10 and 11 are enlarged views of main parts in FIG.

Referring to FIGS. 5 to 7, it can be seen that the moving object is not properly marked and only a part of the moving object is marked. In other words, if you look at a person walking or a car in motion, you can find that not all of the objects are marked, but only some of the blocks are marked. Furthermore, it is also found that a plurality of moving object areas are marked for one moving object. This means that the judgment criterion of the moving object region adopted in the previous (S100) is very useful for filtering out the general region, but it is very strict.

Therefore, it is necessary to detect the boundary of the moving object by looking around the moving object area.

Step S210: First, a plurality of adjacent image blocks are identified centering on the image block marked as the moving object region by the previous step (S100). These are referred to herein as " neighboring blocks ". These neighboring blocks are portions that are not marked as a moving object region according to S100. In the process of FIG. 8, a more detailed look at the neighboring blocks will be made to see if there are any neighboring blocks that can be included in the boundary of the moving object region.

Step S220: S230: The motion vector value is compared with a preset second threshold value (e.g., 0) for a plurality of neighboring blocks, and a neighboring block having a motion vector value exceeding the second threshold value is marked as a moving object region do. If there is a motion that is located adjacent to the recognized moving object region, which is substantially effective, the moving image block is likely to be a lump of the moving object region ahead of the moving object region. Therefore, this neighboring block is also marked as a moving object area.

Step S240: Also, among the plurality of neighboring blocks, marking that the coding type is intra picture is marked as the moving object area. In the case of an intra picture, since there is no motion vector, it is basically impossible to judge whether motion exists in the neighboring block based on the motion vector. In this case, the intra picture adjacent to the image block already detected as the moving object region is safer to maintain the setting of the extracted moving object region.

Step S250: The moving object region is displayed on the reproduction screen of the compressed image so as to be distinguished from the general image. FIG. 9 is a diagram showing an example of a result applied to the boundary region detection process according to the present invention. In the above process, a plurality of image blocks marked as a moving object region are displayed as thick line boxes on a monitor screen. Referring to FIGS. 10 and 11, in the vicinity of the moving object area shown in FIG. 6 and FIG. 7, the moving object area is further expanded in FIGS. 10 and 11 to cover the entire moving object Can be found.

FIG. 12 is a diagram illustrating an example of a result of summarizing a moving object region through interpolation in the present invention, and FIGS. 13 and 14 are enlarged views of main parts in FIG.

Step S300 is a process of organizing the division of the moving object region by applying interpolation to the moving object region detected in the previous steps S100 and S200. Referring to FIGS. 9 to 11, a non-marking image block is found between moving object areas indicated by blocks. If there are non-marking image blocks in the middle, it is difficult to judge whether they are objects to be considered as individual moving objects or as a mass. In particular, since it is displayed mottled on the monitor screen of the CCTV video control system, it is difficult for the control personnel to grasp it immediately. Furthermore, if the moving object area is fragmented, the result of step S400 may become inaccurate, and in particular, the process of step S400 becomes complicated because the number of moving object areas becomes large.

Accordingly, in the present invention, if one or a small number of non-marking image blocks surrounded by a plurality of image blocks marked as a moving object region exist, they are marked as a moving object region, which is called interpolation. 9 and 12, all non-marking image blocks existing between the moving object areas are marked as moving object areas. This makes it possible to derive a more intuitive and accurate moving object detection result for reference by the control personnel.

15 is a flowchart illustrating an embodiment of a process of tracking and identifying a tracking object moving object region specified by a user from a compressed image in the present invention.

As described above, the present invention extracts a moving object region based on syntax information that can be directly obtained from a coding unit of a compressed image. It is not necessary to decode a compressed image of the conventional technique to acquire and analyze a difference image with respect to the original image, thereby achieving a processing speed improvement of up to 20 times according to the inventor's test. However, this approach has the drawback of being less accurate. There is a conceptual difference in that it does not extract the moving object itself but extracts a block of the image block which is assumed to contain the moving object. Reflecting these differences, the present invention adopts a different approach from the conventional technique in the process of tracking a specific object designated by the controller in the CCTV shot image over time.

Hereinafter, an embodiment of the object tracking process adopted in the present invention will be described in detail.

Step S410: First, if a moving object region that is not assigned an ID is found to handle the moving object region as one object, an Unique ID is newly issued and assigned. That is, in the previous process, the chunks of connected image blocks marked as moving object area are treated as one object (object). In order to implement this in the software processing process, a unique ID is assigned to a moving object area (a block of image blocks) and managed.

Accordingly, it is preferable that the following process is performed based on the Unique ID assigned to the moving object area in FIG. 16 shows an example in which a unique ID is assigned to a moving object area.

On the other hand, in step S410, it is necessary to determine whether or not the chunks of the image blocks marked as the moving object region are the same before and after the series of image frames. This is because it is possible to judge whether or not the Unique ID has been previously assigned to the moving object area being handled.

The present invention does not deal with the contents of the original video image but checks whether or not the video block is the moving object area, so that it is impossible to precisely check whether or not the mass of the moving object area in the preceding and subsequent video frames is identical. That is, since the contents of the image included in the image are not grasped, the change can not be identified, for example, when the cat is replaced by a dog between the front and back frames at the same point. However, it is very unlikely that the time interval between frames is very short and that the observation object of the video control system moves at normal speed.

Accordingly, in the present invention, it is assumed that those having a ratio or number of image blocks superimposed between chunks of the moving object region in the preceding and following frames are equal to or larger than a predetermined threshold value. According to this approach, even if the contents of the original image are unknown, it can be determined whether a specific moving object area is moving, a new moving object area is newly displayed, or an existing moving object area is disappearing. This determination is less accurate than the prior art, but it can increase the data processing speed dramatically, which is rather advantageous in practical applications.

Steps S420 and S430: Next, a specific moving object area is set as a tracking object in response to a user, for example, an operation of a CCTV control agent. Assuming the control officer is looking at the CCTV footage and found that the criminal is running on the screen, you can specify to track the offender on the CCTV monitor screen. In the present invention, the criminal is not tracked but the moving object area to which the criminal belongs is identified as being tracked and set as a tracking target. In this specification, this area is referred to as a " tracking object moving object area ".

As described above, in order to manage the identity of the moving object area over time in a series of image frames constituting the compressed image, it is desirable to treat the moving object area based on the Unique ID. Accordingly, the present invention identifies a unique ID assigned to a moving object region to be tracked, and in this specification, the unique ID is referred to as a 'tracking object unique ID'.

Step S440: Next, one or more moving object regions found in each of the series of image frames constituting the compressed image are examined. Unique IDs are allocated to these moving object areas, and the moving object areas to which the unique IDs equal to the tracking target unique ID values are allocated are identified. The identified moving object area corresponds to the tracking object moving object area, and the rectangular coordinates of the identified moving object area are sequentially calculated in each image frame. The set of rectangular coordinates thus obtained is set as a coordinate sequence for the tracking object moving object area.

Preferably, the rectangular coordinates are calculated for the tracking object moving object area. As an example of the rectangular coordinates for the moving object area, it is possible to configure the upper left coordinates (x, y), the abscissa axis length (dx), and the ordinate axis length (dy) of a quadrangle formed to optimally include the moving object area have. That is, the rectangular coordinates of the moving object area are in the form of (x, y, dx, dy). 17 shows an example in which square coordinates are set for each of three moving object areas (Unique ID = 001, 002, 003).

Steps S450 and S460: Then, a series of rectangular coordinates included in the coordinate sequence of the moving object object to be traced is normalized corresponding to the resolution of the compressed image. The processing is performed by mapping a specific range of values, for example, a real value between 0 and 1. In the present invention, it is preferable that the present invention is employed in order to overcome the difference according to the resolution of the video monitoring monitor and to maintain compatibility.

One example of such a normalization process is to divide the rectangular coordinates by the horizontal and vertical resolutions (x_res, y_res) of the compressed image. That is, the left upper-end x coordinate and the abscissa axis length dx of the rectangular coordinate are divided by the horizontal resolution (x_res) of the compressed image, and the upper left y coordinate and the vertical axis length dy of the rectangular coordinate are converted into the vertical resolution (y_res) Division processing. This ensures that all values are normalized to a real value between 0 and 1. For example, if the resolution of the compressed image is 100, 100, and the rectangular coordinates are (0, 0, 50, 50), then the coordinate values become (0.0, 0.0, 0.5, 0.5) after normalization processing. This normalization process is applied uniformly to a series of rectangular coordinates included in the coordinate sequence.

Then, the coordinate sequence of the normalized tracking target moving object region is provided to the video control apparatus so that the state of the tracking target moving object region can be revealed on the monitor handled by the CCTV control agent. As described above, the normalized rectangular coordinates are advantageous in that the video control apparatus always provides a constant view irrespective of the display resolution of the monitor.

Step S470: The moving object region corresponding to the coordinate sequence of the tracking object moving object region is displayed on the reproduction screen of the compressed image so as to be distinguished from the general image. Since the coordinate sequence includes the rectangular coordinates, the corresponding rectangular area can be displayed entirely in a special manner, or the moving object area optimally disposed in the corresponding rectangular area can be specifically displayed.

For example, it is desirable to display the moving object area currently being tracked in a special color. At this time, it is preferable to assign different colors to the objects to be traced. This allows the control personnel of the video control system to immediately recognize the point of the video that is performing the object tracking, thereby observing with higher attention. This can be equally helpful in the process of securing evidence.

Step S480: If the moving object region disappears in a series of image frames, the moving object region is destroyed by recycling the unique ID allocated in Step S410 for the moving object region.

Meanwhile, the present invention can be embodied in the form of computer readable code on a computer-readable non-volatile recording medium. Such a non-volatile recording medium includes all kinds of storage devices for storing computer-readable data such as a hard disk, an SSD, a CD-ROM, a NAS, a magnetic tape, a web disk, a cloud disk, And the code may be distributed and stored in the storage device of the computer.

Claims

A first step of parsing a bit stream of the compressed image to obtain a motion vector and a coding type for the coding unit;

A second step of acquiring a motion vector accumulation value for a first time preset for each of a plurality of image blocks constituting a compressed image;

A third step of comparing the motion vector accumulated value with a predetermined first threshold for the plurality of image blocks;

A fourth step of marking an image block having a motion vector accumulation value exceeding the first threshold as a moving object region;

A series of coordinates of the tracking object moving object region is acquired over a series of image frames of the compressed image with respect to a moving object region (hereinafter referred to as 'tracking object moving object region') specified by the user operation as a tracking object A fifth step of providing the coordinate sequence to the video control device;

Based object tracking method for compressed images.
The method according to claim 1,

In the fifth step,

A fifth step of newly assigning and allocating a unique ID when the moving object area is in an ID unassigned state;

A step 5b of setting a specific moving object area (hereinafter, referred to as 'tracking object moving object area') as a tracking object according to a user operation;

A fifth step c) of identifying a unique ID (hereinafter, referred to as 'tracking object unique ID') allocated to the tracking object moving object region;

A fifth step of sequentially calculating the rectangular coordinates of the moving object area to which the tracking target unique ID value is assigned for a series of image frames constituting the compressed image and setting the rectangular coordinate as a coordinate sequence for the tracking object moving object area;

A fifth step of normalizing the series of rectangular coordinates included in the coordinate sequence of the tracking object moving object region in correspondence with the resolution of the compressed image;

(F) providing a coordinate sequence of the normalized tracking target moving object region to a video control device;

A fifth step of revoking the allocated Unique ID when the moving object area disappears in the series of image frames;

Wherein the object-based object tracking method comprises:
The method of claim 2,

The rectangular coordinates include left top coordinates (x, y), abscissa axis length (dx), and ordinate axis length (dy) of a rectangle formed to virtually include a moving object area,

The normalization process divides the upper left x coordinate and the horizontal axis length dx by the horizontal resolution x_res of the compressed image and divides the upper left y coordinate and the vertical axis dy by the vertical resolution y_res of the compressed image The object tracking method comprising the steps of:
The method according to claim 1,

And a fourth step of performing, between the fourth step and the fifth step,

(A) identifying a plurality of adjacent image blocks (hereinafter, referred to as 'neighboring blocks') around the moving object region;

Comparing a motion vector value with a predetermined second threshold value for the plurality of neighboring blocks;

Further comprising: marking a neighboring block having a motion vector value exceeding the second threshold as a moving object region;

The method of claim 1, further comprising the steps of:
The method of claim 4,

Wherein the step (c)

D) marking a neighboring block having a coding type of an intra picture among the plurality of neighboring blocks as a moving object region;

The method of claim 1, further comprising the steps of:
The method of claim 5,

Wherein the step (d)

Marking a predetermined number of unmarked image blocks surrounded by a moving object area as a moving object area by performing interpolation on the plurality of moving object areas;

The method of claim 1, further comprising the steps of:
The method according to claim 1,

Wherein the image block comprises a macro block and a sub-block.
A computer-readable nonvolatile recording medium on which a program for executing a syntax-based object tracking method for a compressed image according to any one of claims 1 to 7 is recorded on a computer.