KR102097753B1 - Method and Apparatus for Processing Video for Monitoring Video - Google Patents

Abstract

Disclosed is a video processing method and apparatus for video surveillance.
The central processing unit acquires multi-channel image data with one kernel, decodes multi-channel image data by assigning specific address information for each channel in the graphic processing unit, and copies and outputs the decoded original image to the central processing unit. It relates to an image processing method and apparatus for monitoring.

Description

Image processing method and apparatus for video surveillance {Method and Apparatus for Processing Video for Monitoring Video}

The present embodiment relates to an image processing method and an apparatus therefor for improving image processing performance for image monitoring using a graphic processing unit.

The contents described in this section merely provide background information for this embodiment, and do not constitute a prior art.

Due to the recent development of video security, video surveillance systems are used in various fields such as video improvement, analysis, tracking, and surveillance. In particular, a video management system (VMS) is a software that can monitor, record, and control network products such as network cameras and video servers. The video of the camera transmitted through the network in the video control system is compressed by using the H.264 codec, and H.264 decoding is required to display the video on the screen.

In addition, decoded raw data is required for intelligent algorithm operation, and H.264 decoding requires a lot of computing resources. In an image surveillance system that monitors multiple camera images, as the number of channels of the monitoring image increases, a high-performance central control unit resource for decoding is required.

In a general video surveillance system, a device for processing an image generates a plurality of kernels for each channel of the image data, and each kernel sequentially generates memory information and sequentially delivers the decoded data to each kernel. The plurality of kernels increases the load on the central processing unit, and a high-capacity computing device is required to drive the plurality of kernels.

This embodiment obtains multi-channel image data with one kernel from the central processing unit, decodes multi-channel image data by assigning specific address information for each channel in the graphics processing unit, and copies the decoded original image to the central processing unit. The main purpose is to provide an image processing method and apparatus for monitoring video output.

According to an aspect of the present embodiment, a method of processing an image in an image processing apparatus for monitoring an image includes: an acquisition process of obtaining image data for a plurality of channels from one kernel; A memory management process for performing channel frame synchronization for specific address information allocated to each channel of the image data; A decoding process of decoding the image data and converting it into original image data; A transmission process of resetting the specific address information in the original image data and transmitting the original image data to the one kernel; And an output process of outputting the original image data for each channel in the one kernel.

In addition, according to another aspect of the present embodiment, a memory management unit for acquiring image data for a plurality of channels from one kernel and performing channel frame synchronization for specific address information allocated to each channel of the image data; A decoding unit for decoding the image data and converting it into original image data; A data management unit that resets the specific address information to the original image data and transmits the original image data to the one kernel; And an original image output unit that outputs the original image data for each channel in the one kernel.

As described above, according to the present embodiment, the image processing apparatus can stably operate and improve image processing performance by processing image data using one kernel in the central processing unit of the image processing apparatus. It works. In addition, the image processing apparatus has an effect of increasing the performance of the entire system by reducing the load on the central processing unit since most of the processes after image data acquisition are processed in the graphic processing unit.

1 is a block diagram schematically showing an image processing apparatus for video surveillance according to the present embodiment.
2 is an exemplary view for explaining an operation of processing a multi-channel image using a central processing unit and a graphics processing unit for video surveillance according to the present embodiment.
3 is a flowchart for explaining a method of processing a multi-channel image using a central processing unit and a graphics processing unit for image monitoring according to the present embodiment.
4 is a graph showing the utilization rate of the central processing unit in multi-channel image processing according to the present embodiment.

Hereinafter, this embodiment will be described in detail with reference to the accompanying drawings.

1 is a block diagram schematically showing an image processing apparatus for video surveillance according to the present embodiment.

The image processing apparatus 100 according to the present embodiment includes an image acquisition unit 110, a central processing unit (CPU) 120, a kernel unit 130, a graphic processing unit (GPU) 140 , A memory management unit 150, a decoding unit 160, a data management unit 170, and an original image output unit 180. The image processing apparatus 100 shown in FIG. 1 is according to an embodiment, and not all blocks shown in FIG. 1 are essential components, and in another embodiment, some blocks included in the image processing apparatus 100 are added. , Can be changed or deleted.

The image acquisition unit 110 acquires image data from an image capture device (not shown) such as a camera or CCTV. The image acquiring unit 110 acquires image data for each of the captured images captured by the plurality of image photographing devices through a plurality of channels. Here, the image acquisition unit 110 may acquire image data from an image storage device (not shown) in which the captured image captured by the image photographing apparatus is stored.

The image acquisition unit 110 according to the present embodiment acquires image data for each of a plurality of channels (n multi-channels). Here, the image data for each of the plurality of channels means data pre-encoded with a predetermined codec. The image acquisition unit 110 transmits image data for all acquired channels to the central processing unit 120.

The central processing unit 120 is a module that processes all data in the center of the image processing apparatus 100, and performs a function of interpreting and calculating a command input from a user and outputting the result. The central processing unit 120 according to the present embodiment receives image data for all channels from the image acquisition unit 110 to be stored as one kernel unit 130, and stores the image data stored in the kernel unit 130. To be stored in the graphics processing unit 140.

The kernel unit 130 generates one kernel based on the control of the central processing unit 120, stores image data for all channels in the generated kernel, and transmits the stored image data to the graphic processing unit 140 do.

The kernel unit 130 generates one kernel and allocates specific address information for each channel in the generated kernel to store image data. Here, the specific address information refers to identification information for storing and storing each image data for each channel.

The kernel unit 130 preferably allocates specific address information to transmit the stored image data to the graphic processing unit 140, and deletes the stored image data to reduce the computation amount of the central processing unit 120, but is limited to this. It is not possible to store only cookie information or some information (eg, specific address information for each channel) of stored video data.

Meanwhile, when receiving the decoded original image data from the graphic processing unit 140, the kernel unit 130 classifies and stores the original image data for each channel based on specific address information of the received original image data.

The kernel unit 130 separates and synchronizes the original image data received from the graphic processing unit 140 for each channel by using each specific address information, so that there is no problem such as channel crosstalk for the original image data.

The graphic processing unit 140 is a module that processes an image (graphic) of the image processing apparatus 100 in conjunction with the central processing unit 120 of the image processing apparatus 100. The graphic processing unit 140 processes operations related to image processing from the central processing unit 120 so that the load of the central processing unit 120 is reduced. Also, since the graphic processing unit 140 can process a large number of matrices and vectors, it can be applied to image processing using such an operation.

The memory management unit 150 receives image data for all channels from the kernel unit 130 and performs channel frame synchronization on the received image data. In more detail, the memory management unit 150 extracts specific address information for each channel included in the image data received from the kernel unit 130, and uses the specific address information for each channel so that each image data is not confused. And channel frame synchronization for specific address information.

The memory management unit 150 generates memory management information by synchronizing a channel frame with respect to a channel and specific address information for specific address information allocated for each channel based on image data of all received channels. For example, when receiving image data for three channels, the memory management unit 150 may allocate a first address value assigned to the first channel, a second address value assigned to the second channel, and a third channel assigned to the third channel. The third address value is extracted and synchronized memory management information is generated and stored.

The decoding unit 160 decodes image data for all channels and converts it into original image data. The decoding unit 160 according to the present embodiment performs decoding based on a pre-encoded predetermined codec. Here, the decoding unit 160, it is preferable to set blocks in the image data for all channels and decode a bit stream in block units, but is not limited thereto. If the image data can be converted into original image data, Decoding can be performed in any way.

The data management unit 170 stores the decoded original image data by data synchronization, and transmits the original image data to the kernel unit 140.

The data management unit 170 according to the present embodiment resets specific address information of the original image data based on previously generated memory management information, and additionally sets and stores independent thread information for each specific address information. . Here, the independent thread information means thread information set so that the original image data is not confused for each channel when the original image data is transmitted from the graphic processing unit 140 to the central processing unit 120.

The data management unit 170 stores the original image data for each channel based on specific address information for each channel of the original image data, and synchronizes the stored original image data based on the specific address information to synchronize the kernel unit 130 ). Here, the data management unit 170 transmits it to the kernel unit 130 so as to distinguish and store a location stored according to specific address information of the original image data.

On the other hand, if the data management unit 170 cannot separate and store the original image data according to specific address information in the kernel unit 130, the resolution information, the pixel value information, the GOP (Group) so as not to be a channel confusion about the original image data Of Pictures) and the like, and the original image data is flexibly classified and stored in the kernel unit 130.

The original image output unit 180 outputs the original image data stored in the kernel unit 130 to perform image monitoring. The original image output unit 180 outputs the original image data stored in the kernel unit 130 for each channel so that image monitoring is performed to detect objects or events within each original image data.

2 is an exemplary view for explaining an operation of processing a multi-channel image using a central processing unit and a graphics processing unit for video surveillance according to the present embodiment.

The image acquisition unit 110 acquires multi-channel image data. For example, the image acquisition unit 110 acquires a video stream for each of the n channels as multi-channel image data (210).

The central processing unit 120 sets specific address information for each channel of the multi-channel image data received from the image acquisition unit 110, and each of a plurality of streams corresponding to specific address information in the CPU memory area (one kernel). Stores multi-channel image data for each channel (220).

The central processing unit 120 copies the multi-channel image data stored in the CPU memory area to the GPU memory of the graphic processing unit 140, and the graphic processing unit 140 performs channel frame synchronization for the multi-channel image data to perform each Memory management information for managing image data for each channel and specific address information corresponding to each image data is generated.

The graphic processing unit 140 decodes multi-channel image data, resets specific address information for each channel of the decoded original image data, and sets independent thread information for each specific address information to set the original image data. Save (230). Here, the independent thread information means thread information set so that the original image data is not confused for each channel when the original image data is transmitted from the graphic processing unit 140 to the central processing unit 120.

The graphic processing unit 140 performs data synchronization based on specific address information while copying the original image data from the GPU memory to the CPU memory so that each channel of the original image data is not confused (240). The central processing unit 120 outputs the original image data stored in the CPU memory area corresponding to specific address information for each channel for each channel (250).

3 is a flowchart for explaining a method of processing a multi-channel image using a central processing unit and a graphics processing unit for image monitoring according to the present embodiment.

The image processing apparatus 100 acquires image data for a plurality of channels (S310). Here, the image data for each of the plurality of channels refers to data pre-encoded with a predetermined codec.

The image processing apparatus 100 sets specific address information for each channel of the acquired image data, and stores image data for each channel based on specific address information in one kernel of the central processing unit 120 (S320). .

The image processing apparatus 100 copies the image data stored in one kernel to the GPU memory of the graphic processing unit 140 (S330).

The image processing apparatus 100 performs channel frame synchronization for image data, generates memory management information for managing image data for each channel and specific address information corresponding to each image data, and decodes the image data ( S340). Here, the image processing apparatus 100 decodes the image data based on the encoded predetermined codec and converts the image data into original image data.

The image processing apparatus 100 separates the decoded original image data for each channel (S350), resets specific address information to the original image data for each channel, and performs data synchronization to perform kernel synchronization of the central processing unit 120 The original image data is copied with (S360).

The image processing apparatus 100 outputs the original image data stored in the kernel for each channel (S370), and performs image surveillance to detect an object or event using the output original image data (S380).

Although FIG. 3 describes that steps S310 to S380 are sequentially executed, this is merely illustrative of the technical idea of the present embodiment, and those skilled in the art to which this embodiment belongs belongs to this embodiment. Since the sequence described in FIG. 3 may be changed and executed or one or more of steps S310 to S380 may be executed in parallel within a range that does not deviate from essential characteristics, various modifications and variations may be applied, and FIG. 3 is a time series sequence. It is not limited.

As described above, the operation of the image processing apparatus 100 according to the present embodiment described in FIG. 3 may be implemented as a program and recorded in a computer-readable recording medium. The program for implementing the operation of the image processing apparatus 100 according to the present embodiment is recorded, and the computer-readable recording medium includes all types of recording devices in which data that can be read by a computer system is stored. Examples of such computer-readable recording media include ROM, RAM, CD-ROM, magnetic tapes, floppy disks, optical data storage devices, etc., and also implemented in the form of carrier waves (for example, transmission via the Internet). It also includes. In addition, the computer readable recording medium may be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion. In addition, functional programs, codes, and code segments for implementing the present embodiment can be easily inferred by programmers in the technical field to which this embodiment belongs.

4 is a graph showing the utilization rate of the central processing unit in multi-channel image processing according to the present embodiment.

4A is a graph showing the utilization rate of the central processing unit according to an operation of decoding image data using only the central processing unit and outputting the decoded original image data in a general image processing apparatus. As shown in FIG. 4 (a), when a general image processing apparatus decodes image data having 10 channels, the utilization rate of the central processing unit continues to rise to process image data of the 10th channel. It will increase to about 100%.

4 (b) decodes multi-channel image data using the central processing unit 120 and the graphic processing unit 140 in the image processing apparatus 100 according to the present embodiment, and outputs the decoded original image data. It is a graph showing the utilization rate of the central processing unit 120 according to the operation. As shown in FIG. 4 (a), when the image processing apparatus 100 decodes image data having 16 channels, the utilization rate of the central processing unit is about to be decoded while decoding image data for 16 channels. About 10% is kept constant.

General image processing device Image processing apparatus of the present invention Improvement rate (%) Number of frames 250 300 20% CPU utilization (%) 100 11 1,100%

Table 1 shows the results of the utilization rate of the central processing unit 120 and the improvement rate according to decoding of the above-described general image processing apparatus and the image processing apparatus 100 according to the present embodiment.

The above description is merely illustrative of the technical spirit of the present embodiment, and those skilled in the art to which this embodiment belongs may be able to make various modifications and variations without departing from the essential characteristics of the present embodiment. Therefore, the present embodiments are not intended to limit the technical spirit of the present embodiment, but to explain, and the scope of the technical spirit of the present embodiment is not limited by these embodiments. The protection scope of the present embodiment should be interpreted by the claims below, and all technical spirits within the equivalent range should be interpreted as being included in the scope of the present embodiment.

As described above, the present embodiment is applied to the field of video surveillance, so that the image processing device can operate stably, improve the image processing performance, and after the image data is acquired by the image processing device, most of the process is graphic processing. Since it is processed in the unit, it is a useful invention that can reduce the load of the central processing unit and generate an effect that can increase the performance of the entire system.

100: image processing apparatus 110: image acquisition unit
120: central processing unit 130: kernel unit
140: graphic processing unit 150: memory management unit
160: decoding unit 170: data management unit
180: original video output unit

Claims (6)

A method of processing an image in an image processing apparatus for video surveillance,
An acquisition process of acquiring image data for a plurality of channels from one kernel;
Memory is obtained by extracting specific address information allocated from the kernel for each channel of the image data, and performing channel frame synchronization for each channel of the image data and the specific address information using specific address information for each channel. A memory management process for generating management information;
A decoding process of decoding the image data and converting it into original image data;
A transmission process of resetting the specific address information in the original image data based on the memory management information and transmitting the original image data to the one kernel; And
Output process of outputting the original image data for each channel in the one kernel
Video processing method for video surveillance comprising a. According to claim 1,
The acquisition process,
The central processing unit (CPU) allocates the specific address information for each of the plurality of channels, and copies the image data for each pre-encoded channel into the GPU memory of a graphic processing unit (GPU). Image processing method for video surveillance characterized by a feature. delete According to claim 1,
The transmission process,
An image processing method for video surveillance, characterized in that additional independent thread information for preventing channel confusion of the original video data is additionally set for each specific address information. According to claim 1,
The output process,
A video processing method for video surveillance, characterized in that the original video data is classified and stored for each channel based on the specific address information in the one kernel. Acquiring image data for a plurality of channels from one kernel, extracting specific address information allocated from the kernel for each channel of the image data, and using the specific address information for each channel for each of the image data A memory management unit generating memory management information by performing channel frame synchronization for a channel and the specific address information;
A decoding unit for decoding the image data and converting it into original image data;
A data management unit that resets the specific address information to the original image data based on the memory management information and transmits the original image data to the one kernel; And
The original image output unit for outputting the original image data for each channel in the one kernel
Video processing apparatus for video surveillance comprising a.

Images