Network Camera Based on Embedded Operating System

Compared with traditional analog monitoring systems, digital video monitoring systems have many advantages:

1) Save the surveillance video on a large-capacity hard disk. The digital signal storage information will never be lost and the image quality will not be reduced.

2) Digital video can be easily processed by computers. Alarm areas can be set in the monitoring images, and the computer analyzes and processes the images in the alarm areas, achieving automatic monitoring without human intervention.

3) After being compressed, digital video occupies less bandwidth, and digital signals are easy to encrypt during transmission, so illegally intercepted signals cannot be restored to video images.

4) Digital videos are stored on the computer hard disk and can be retrieved and managed by computer, eliminating the complexity and tediousness of manual management and retrieval.

Research on remote video surveillance systems based on network cameras started earlier abroad. Mature network camera products include the 10BASE2T network camera "SNC2VL10N" launched by Sony Marketing Company, as well as Panasonic KX2HCM130, Samsung SNC2100P, and ANT2NWC10/50/100. Similar foreign products generally use MPEG or wavelet compression methods, with better performance, but expensive, which most domestic users cannot afford. Research in this area in my country has just started, and most manufacturers only represent foreign products, and there are not many units engaged in actual research.

At present, most of the video acquisition devices that have been put into use in the domestic market exist in the form of video acquisition cards. The solution designed in this paper uses TI's TVP5040 as the main chip of the front-end video acquisition device, which can exist as an independent unit and has the characteristics of small size, strong processing capability and low power consumption; in addition, it uses the powerful TI embedded processor chip TMS320DSC21, so that the system can quickly and stably realize the acquisition, compression and packaging upload of image data, providing a solution for digital video surveillance. This paper briefly introduces TMS320DSC21, and details the hardware structure of the video data acquisition and processing end of the self-developed remote video monitoring system, as well as the design scheme of using this system to realize real-time network monitoring.

1 TMS320DSC21 microprocessor

The TMS320DSC21 microprocessor (DSC21 for short) is a high-performance single-chip microprocessor consisting of the following parts:

1) 1 DSP (TMS320C5409) subsystem;

2) 1 microcontroller (ARM7TDMI) subsystem;

3) 2 DSP coprocessors (iMX and VLC);

4) SDRAM controller;

5) Image peripheral unit①.

The high integration of DSC21 makes it particularly suitable for video, audio and image products. Its five subsystems form a complete system on a single chip. The on-chip MCU runs the user operating system to handle complex protocols, realize control functions and bit operations; the DSP and DSP coprocessor complete large-scale calculations, digital signal processing and arithmetic calculations; the image peripheral components and SDRAM controller provide huge data flow. Therefore, DSC21 can realize a high-performance, multi-processor image processing system on a single chip.

Figure 1 shows the DSC21 module diagram

The DSC21 module includes:

1) ARM subsystem. It consists of ARM7TDMIRISC core, internal RAM and peripheral modules. The ARM subsystem completes all control functions and supports real-time operating systems such as ucOS, Vx2Works, uITRON, Nucleus and Linux. ARM also controls various peripheral modules such as timers, interrupt controllers, CCD/CMOS controllers, OSD, NTSC/PAL video encoders, USB, CF card, SM card, IrDA, preview engine, serial port, etc. The user interface software also runs on ARM.

2) DSP subsystem. It consists of DSP core, 32KB RAM and dedicated memory block (ImageBuffer). The DSP core is a DSP processor TMS320C5409DSP based on 0118μm technology.

The DSP subsystem is responsible for all high-computation signal processing tasks, such as image processing and JPEG compression; it handles all real-time I/O, such as audio and Modem I/O, MP3, and AAC playback; it supports programmable real-time automatic exposure, automatic focus, automatic white balance (AE, AF, AWB), as well as image/video capture and playback.

3) DSP coprocessor. It consists of two parts: iMX and VLC, and is responsible for assisting the DSP subsystem in image processing. iMX consists of 4 parallel multiplication and addition units (MACs), which are suitable for matrix operations; while VLC is optimized for quantization and Huffman coding of JPEG and MPEG1. The DSP subsystem can operate the DSP coprocessor by calling subroutines.

4) SDRAM controller. It is the interface between the off-chip extended SDRAM and all the functional modules on the chip. It provides a high-bandwidth interface for the off-chip SDRAM to connect to the DSC21 processor and the image unit, supporting up to 80MHz, 32bitSDRAM. The SDRAM controller supports real-time CCD data stream input and TV display data output, and its access performance reaches 320MB·s-1.

5) Image peripheral components. DSC21 has 4 image units:

CCD controller, preview engine, burst mode compression/decompression unit and hardware graphics unit. These dedicated image units are controlled by ARM and are used to read data into the DSC21 system and create a color display.

The CCD controller can provide the necessary synchronization timing logic for the CCD/CMOS image sensor and support progressive scan and interlaced scan CCD/CMOS image sensors. By configuring the registers inside the CCD controller, the CCD controller can be put into different working states. The CCD controller can support CCD/CMOS image sensors up to 16Mpix (4K×4K).

The preview engine module converts the data from the CCD controller into a display format suitable for the NTSC/PAL encoder, which can achieve real-time 30 frames/s-1 NTSC/PAL preview. The preview engine has built-in gain control, white balance, vertical and horizontal noise filtering, CFA interpolation, downsampling, gamma correction and color space conversion functions.

The compression and decompression module of the continuous shooting mode uses a lossless (or lossy) algorithm to compress the original image data transmitted from the CCD controller and write it into SDRAM. The decompression engine can decompress this data under the control of the DSP. This data is processed, displayed, and stored back into SDRAM. The continuous shooting mode can support the shooting of 10 frames/s-1 high-resolution photos.

The hardware graphics unit consists of 3 parts: OSD (OnScreenDisplay), NTSC/PAL encoder and DAC (digital-to-analog converter).

OSD supports 3 hardware windows, and the NTSC/PAL encoder integrates composite video signal output channels and independent R, G, B output channels①.

2 Embedded Network Camera System Based on DSC21

2.1 System Hardware Composition

This system processes and compresses the collected images to obtain MJPEG files, which are then transmitted over the network so that the remote control end can view the on-site situation in real time and adjust the image quality, pan/tilt angle, etc.

Figure 2 shows the block diagram of the embedded video acquisition system. The video A/D chip completes the input conversion of 1 to 4 cameras, the FLASH chip is used to store all programs, the SDRAM is used for intermediate data storage, the CPLD is used for hard disk logic conversion, and the Ethernet controller chip is responsible for uploading image files and receiving commands and parameters from the control end.

2.2 System Software

The system software includes embedded system software and hardware drivers. This system uses an embedded operating system based on ucLinux as the lower computer system software. ucLinux is an excellent embedded Linux version.

ucLinux has the ability to support multiple file systems, has built-in TCP/IP protocol, can draw on the rich resources of Linux, and has considerable advantages for some complex applications; however, due to the lack of MMU (memory management unit), its multi-tasking implementation requires certain skills. The typical startup process of ucLinux is shown in Figure 3.

In ucLinux system, almost all the control operations of devices are implemented by device-related codes, which are device drivers. The system kernel must embed device drivers for each external device from hard disk drives to keyboards and tapes. Generally, the driver of peripherals is first written in user space, that is, the operation of peripherals is first implemented in the application layer. After the driver is debugged in the application layer, it can be encapsulated into the kernel module.

3 Application of the system in network monitoring

The network camera system based on embedded operating system is very suitable for network monitoring applications. CAU2DVCⅠ is an embedded network camera based on TMS320DSC21. The performance parameters of CAU2DVCⅠ are as follows: image compression format is M2JPEG; image resolution is CIF (352×288pix); communication port RS485; Ethernet interface RJ24510/100Mbitbase2T adaptive; network protocol TCP/IP; transmission frame rate PAL maximum 25 frames·s-1, NTSC maximum 30 frames·s-1; alarm interface 2 input and 2 output; embedded operating system u2cLinux; static IP address, which can be set through the network.

This system can be applied to many fields: power system (power plant, power supply station, unmanned substation), real estate industry (intelligent community/building, basic engineering site, property management center), education system (kindergarten, examination room, colleges and universities) and manufacturing industry (factory, workshop), and many units have been put into use in Beijing and Shanghai. Figure 5 shows the structure of the video surveillance system based on CAU2DVCⅠ. CAU2DVCⅠ network camera has been used in the video surveillance system of Tongli substation of Tongzhou Power Supply Bureau, with good results.