KR20140130644A - Intelligent image data search system - Google Patents

Abstract

Provided is a system for searching a video, capable of performing object analysis for a dynamic object that is not part of the basic background of a photographed place, such as a vehicle, a person, an animal, etc. which is moved into a background when image data is recorded in a device for recording the video such as a CCTV; building a database for the shape, color, size, speed, movement direction, etc.; assigning an ID to the corresponding object; and searching the video in which the corresponding object is included by inputting information such as the shape, the color, the size, the speed, the movement direction, etc. when a user searches the video.

Description

{INTELLIGENT IMAGE DATA SEARCH SYSTEM}

The present invention relates to an image data retrieval system that enables identification by object in storing and retrieving image data including moving pictures such as CCTV and video camera.

Conventionally, when it is necessary to re-observe an image of a particular object among images recorded using a CCTV or a video camera, the entire image is searched again or a recorded time zone is specified by searching the recorded image at the corresponding time zone . For example, in case of a car accident in a parking lot, if you want to observe the accident scenes of the car from CCTV recorded images, you can search all CCTV images back, find the date and time of the accident, do. Therefore, it is not easy to search, and it becomes more difficult to search if the event occurrence time can not be grasped at short intervals.

Therefore, it is possible to search for objects (people, animals, vehicles, objects, etc.) to search more efficiently if they can recognize and search for other information in addition to the recording time zone. To do this, it is necessary to analyze the recorded video data and assign an ID to each of the recorded video data.

Korean Patent Laid-Open No. 10-2011-0128621 describes image analysis, storage, and retrieval of still images, and stores images by analyzing various image information such as color, spatial information, coordinates, and spatial range. However, the above publication does not describe contents for convenience of retrieval of a specific object for a moving picture at all.

Accordingly, it is an object of the present invention to provide a new moving picture search method for easily and surely searching for a specific object that needs to be reviewed among video information recorded in a moving picture such as CCTV.

According to the above object, the present invention provides a moving picture recording apparatus, such as a CCTV, which records moving image data such as a moving object, such as a vehicle, a person, an animal, The object is analyzed with respect to the object, information is formed on the shape, color, size, speed, moving direction, and the like, the record information is constructed as a database, an ID is given to the object, , A shape, a color, a size, a speed, a moving direction, and the like, so that a moving picture containing the object can be searched.

That is, according to the present invention, in recording a video image by capturing an image of a specific region with a CCTV camera, an IP camera, or a web camera in a moving image recording apparatus for monitoring,

A photographing time (time) is recorded for every data frame of the image data to be photographed,

A dynamic object that is moved into the background such as a vehicle, a person, an animal, or an object other than the basic background of the place where the moving image is shot is analyzed with respect to a frame in which the corresponding object is recorded to determine the shape, size, color, A dynamic object database storing the dynamic object variables including the direction and the recording time is constructed,

Assigns an ID to the corresponding object,

And a search module for searching the recording time of the corresponding dynamic entity and displaying a corresponding frame of the video data when a specific dynamic object variable is designated from the dynamic object database,

A moving picture search method is provided in which at least one of a variable including a moving picture shooting time, a shape, a size, a color, a speed, and a moving direction is input to search for a moving picture containing the object do.

Further, the present invention may be configured such that the moving picture search method further includes a step of displaying information on an initial time at which the dynamic object appeared when the variable such as the shape, color, size, speed, And a tolerance is set so as to record an image of the object to be searched.

In addition, the present invention is characterized in that, in recording the object size information, the size of the object is the closest to the size recognized at the farthest from the point where the camera is installed, And a size ranging from a predetermined size to a recognized size.

 In addition, the present invention is characterized in that, in the above, the shape of the entity is recorded in a shape simplified to any one of shapes including a rectangle, circle, ellipse, triangle, pentagon, hexagon, And provides a method for searching for moving images.

In addition, the present invention provides a moving picture search method, wherein the tolerance for the size of the object is given to a length, a length, a diameter, a long axis, a short axis, a length or an angle of a side.

Further, in the present invention, when searching for a specific object, information on one or more of a photographing time, shape, size, color, speed, or direction of the object is input and retrieved. to provide.

Further, the present invention is characterized in that when the information for searching for a specific entity is input, the corresponding ID is searched in the database and the still images including the corresponding entity are sequentially displayed or displayed on a monitor in a plurality of small screens As shown in FIG.

Further, the present invention provides a moving picture search method, wherein, when information for searching for a specific object is inputted, the corresponding ID is found in the database, the time zone of the corresponding object is indicated, and the corresponding moving picture is displayed in each time slot do.

Further, according to the present invention,

A camera module for photographing a moving picture; And

A control module including video analysis, storage, and search functions,

The control module includes:

A clock means for recording an imaging time at every frame when recording the moving picture data;

A dynamic object variable including a shape, a size, a color, a velocity, and a moving direction is analyzed by performing an object analysis on a dynamic object moving in the background such as a vehicle, a person, an animal, An image analysis module for building a dynamic object database;

A dynamic entity database in which an ID is given and stored in association with the analysis result data of the image analysis module; And

Thereafter, when a specific object is searched for in the moving image, one or more dynamic object variables including a moving image capturing time, a shape, a size, a color, a speed, and a moving direction are input to search for a moving image And a module for searching for a moving picture.

According to the present invention, it is possible to provide an image search method and a video search method that can avoid the inefficiency of observing an image search for an object composed of a vehicle, a person, an animal, It is possible to designate a physical quantity and search only the corresponding image quickly and efficiently.

FIG. 1 is a schematic view for explaining a process of assigning an object ID according to appearance of a person and an animal from a CCTV image installed at an entrance of an elevator, and informing it through image analysis as an embodiment of the present invention.
2 is a view showing the shape of an object such as a vehicle or a bag according to an embodiment of the present invention.
3 is an exemplary diagram illustrating a screen configuration for inputting entity information to be tracked according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The embodiment of the present invention requires a camera that captures moving images. The camera may be provided with a function of including an object analysis module and a search module in addition to recording and storing in the camera itself. And an object analysis module and a search module may be provided in a separate computer.

As an example of a moving picture camera suitable for applying the present embodiment, there is a CCTV camera, a web camera, etc. Hereinafter, a CCTV camera will be described as an example.

1 shows a screen image on a CCTV for capturing and monitoring an elevator entrance and exit, and a person entering and exiting the same.

In the first place, the background image (in the case of FIG. 1, corresponding to the elevator door and the wall configuration) always taken without moving in the place where the CCTV is installed does not require any special analysis of the object. However, since the appearance of humans and animals represents movement, we perform image analysis for individuals expressing speed for background images. The CCTV image control module for performing functions such as image analysis can be programmed using a microprocessor in the camera itself or can be installed by programming on a computer connected by wire or wirelessly.

That is, when an image other than the background image appears, the shape is first analyzed to give an object ID specifying a rectangle, a square, a rhombus, a parallelogram, a trapezoid, a circle, an ellipse, a triangle, a pentagon, a hexagon, And analyzes the size of the shape figure. Width, length, diameter, long axis, short axis, and so on, and the internal and external angles of the figure may be analyzed. The vehicle is generally rectangular, triangular for a crane, and trapezoidal for a golf bag or musical instrument bag, as illustrated in FIG. The shape of the object is mainly determined by specifying some of the outermost tangents of the shape of the object.

Such a simplification of the shape can be realized by a known technique, and in particular, an edge analysis method can be applied.

The analyzed information is recorded and stored for the corresponding ID as entity information. In this case, a considerable tolerance is given to the size of the object, and in particular, the size of the CCTV is stored as the size information including the size and the size of the CCTV photographing point, but the distance from the CCTV at the farthest point. However, it is preferable to store the size information with sufficient tolerance because there is a possibility of missing the image search.

Particularly, it is desirable to analyze the object of the frame at the time of appearance of the dynamic object, considering that the reason for searching the moving image is often in the case of tracking a person in the event of a crime or an accident, It is preferable to store information up to a threshold size at which identification of the object in the shooting range becomes impossible.

That is, when the moving image is recorded, the shape and size of the dynamic object appearing against the motionless background image are analyzed for the image frame at the time of appearance, and the shape and size are recorded. When the threshold value of the identifiable state reaches the CCTV screen Capturing an object of the corresponding frame, judging the size thereof, and recording it together with the size information.

In addition, it is preferable to scale the background image in the stopped state based on the size of the object, thereby accurately recording the size of the actual object, thereby increasing the usability.

When analyzing the color of an object, one object may be composed of several colors, so that it can record all the color information that appears. In case of color, if there is enough tolerance, if it is recorded as red (RED) To specify a plurality of samples for the search, so as not to be omitted from the search. That is, not a single color for a search term red, but rather a wide range of color data that is recognized as being seen in red by the naked eye.

The speed and direction of movement of the object can be entered as relatively accurate information. This is because the speed of the image is calculated by observing the image frame photographed per unit time and calculating the position movement amount. The direction of movement can also be easily grasped because it can observe the azimuth change of the moving object while setting the azimuth with respect to the background background image. However, it is preferable to give a tolerance to the speed and the moving direction.

In addition to speed, the numerical range of tolerances such as size can be around ± 5%, but this is exemplary and not at all limiting.

That is, the application of the tolerance allows the control module to embed a program module that gives a tolerance for inputting variable (parameter) data for a dynamic entity such as size (dimension), color, speed, will be.

In this way, when recording CCTV moving images, it is very efficient to search for an object by using a dynamic object database storing ID information and characteristic information stored by static image analysis in case of dynamic object appearance.

The search process of an entity can be performed as follows.

When a dynamic object appears while recording a CCTV moving picture, a dynamic object database which records object information through object analysis and assigns an ID is constructed separately from a moving picture recording database. When a specific object is to be found in a CCTV recorded image, An object variable is input to retrieve the corresponding object information from the dynamic object information database, and the moving picture record is found and displayed from the visual information recorded in the object information. For the sake of convenience of use, it is preferable that such a searching process is modularized by programming and installed in a computer interlocked with the CCTV or a processor included in the CCTV itself, and the user inputs one or more pieces of object information among various types of object information After searching the dynamic object information database through the search module, the video record is searched continuously and the video including the object is displayed. It is desirable to input individual information by connecting various variables such as time information, shape, size, color, speed, and the like to AND, OR, and NOT.

Fig. 3 shows an example of a search screen.

For example, if you need to track the behavior of a person who takes a puppy from a CCTV image placed in front of an elevator door, you can search for variables for a person or for a dog. If the time zone is known, the time zone is entered by specifying the start time and end time, and a rhombus and a rectangle for the shape are set to 30 cm in width, 20 cm in height, white in color, 10 km / h in speed, The tolerance is automatically given to each physical quantity as described above to find and display all the object information variables stored as the analysis result of the object frame in the dynamic object information database, And displays it. At this time, a plurality of the still images may be displayed sequentially or on a small screen, or the time zone of the corresponding object may be firstly displayed, and the moving image may be displayed in a time zone. Of course, you can alternate the two to show still images and videos.

In this way, the object to be tracked very quickly can be found from the CCTV moving picture record, so that the CCTV can be efficiently used.

It is to be understood that the invention is not limited to the disclosed embodiment, but is capable of many modifications and variations within the scope of the appended claims. It is self-evident.

No reference symbol.

Claims (1)

In the monitoring video recording apparatus, when a CCTV camera, an IP camera, or a web camera captures an image of a specific area and records the image data,
A photographing time (time) is recorded for every data frame of the image data to be photographed,
A dynamic object that is moved into the background such as a vehicle, a person, an animal, or an object other than the basic background of the place where the moving image is shot is analyzed with respect to a frame in which the corresponding object is recorded to determine the shape, size, color, A dynamic object database storing the dynamic object variables including the direction and the recording time is constructed and the variables that are characteristic of the dynamic object are stored under one ID, The dynamic object database is constructed by scaling the dynamic object size, storing the size and tolerance of the actual object together with the recording time,
Assigns an ID to the corresponding object,
And a search module for searching the recording time of the corresponding dynamic entity to display a corresponding frame of the video data when a specific dynamic object variable is designated from the dynamic object database,
In searching for a moving image, at least one of variables including a moving image capturing time, a shape, a size, a color, a speed, and a moving direction may be input,
When the information for searching for a specific object is input, the corresponding ID is searched in the database to display the still images including the corresponding object sequentially or in a plurality of small screens on the monitor,
It shows the time of occurrence of the object, shows it as a video by time zone,
Wherein the dynamic object database and the search module are built in a CCTV camera, an IP camera, or a web camera itself.

Images