The development of multimedia communication technology has provided abundant means for information acquisition and transmission. Video capture is an indispensable and important component. This system is based on the S3C2410 ARM9 chip and an embedded Linux operating system. It uses a USB camera to capture video, which is compressed and encoded using the MPEG-4 algorithm. The system is directly connected to the network, and users can view remote video images using a standard web browser and streaming media player. 1. Hardware System The system hardware platform uses the UP-NETARM2410 development board from Beijing Bocheng Technology Co., Ltd. This system is based on the ARM9 architecture embedded chip S3C2410, which operates stably at a 202MHz main frequency. It has 64MB SDRAM and 64MB FLASH onboard. Motherboard resources include: main USB port, slave USB port, 10M/100M Ethernet port, touch screen, color LCD, keyboard, 8 user-defined LED digital tubes, A/D, RTC circuit, 2 serial ports, 1 JTAG universal interface, audio module supporting MPEG4 and MP3 encoding and decoding, 3 168-pin expansion sockets, a 32-bit data bus, and sufficient expansion space. The standard modules include: IC card + PS2 module, IDE hard drive + CF card module, and PCMCIA + SD/MMC module. Optional modules include: GPS module, GPRS module, FPGA module, CAN + AD + DA module, infrared module, Bluetooth module, and camera module. 2. Software System 2.1 Kernel Configuration and USB Camera Driver Assuming the embedded Linux development environment has been set up, the first step is to install and drive the USB camera. First, check if USB module support has been added to the Linux Kernel, and if Video4Linux support has been added. Multimedia devices → <M> Video For Linux Video For Linux → [*] V4L information in proc filesystem Under USB Support in the main menu, there are various camera drivers; select the camera chip type you want to use. <> USB IBM (Xirlink) C-IT Camera support <> USB OV511 Camera support <> USB Philips Cameras <> USB SE401 Camera support <> USB STV680 (Pencam) Camera support <> USB 3com HomeConnect (Akavicam) support When selecting a USB camera, prioritize camera chips publicly supported by the Linux kernel. Otherwise, you'll need to write a separate USB camera driver, compile, and install it. Here, we choose the V3000 product from NetEye, which uses the OV511 chip. After confirming that the USB camera is properly driven, the next step is to use the API function set provided by Video4Linux to write a video capture program. 2.2 Video Capture Module Designed Based on V4L In Linux, all peripherals are treated as special files called device files. System calls are the interface between the kernel and applications, while device drivers are the interface between the kernel and peripherals. They handle device initialization and release, various operations on device files, and interrupt handling, shielding applications from the details of peripheral hardware and allowing applications to operate on peripherals like ordinary files. The Video4Linux subsystem in Linux provides a unified API for video applications, allowing them to operate various video capture devices through standard system calls. Video4Linux registers video device files with the virtual file system, and applications access these devices by manipulating these files. This section focuses on the video capture program design for the device file `/dev/video`. The main functions used in the Linux video capture process are: `Camera_open()`: Opens the video device file; a `video_device` type device file must be declared beforehand. `Camera_get_capability()`: Retrieves device file information using the `ioctl()` function and stores it in the `video_capability` structure. `Camera_get_picture()`: Retrieves image information using the `ioctl()` function and stores it in the `video_picture` structure. `Camera_close()`: Closes the device file. `Camera_grab_image()`: Captures images using mmap, directly mapping the device file `/dev/video0` to memory, accelerating file I/O operations and allowing multiple threads to share data. Other functions related to device initialization and parameter devices are not detailed here. 2.3 After the video compression encoding module acquires the image data, it can be directly output to the FrameBuffer for display. Since this system needs to transmit the acquired video image over the network, the original image data needs to be compressed and encoded before transmission. The MPEG-4 video encoding and decoding scheme is selected here. Compared with other standards, MPEG-4 has a higher compression ratio, saves storage space, and has better image quality. It is particularly suitable for transmitting video under low bandwidth conditions and can maintain image quality. The object-based video encoding process in MPEG-4 can be divided into three steps: (1) Segmenting video objects from the original video stream. (2) Encoding video objects and assigning different codewords to the motion information, shape information, and texture information of different video objects. For input VOP sequences of arbitrary shapes, block-based hybrid encoding technology is used for encoding. The processing order is IVOP first, then PVOP, and then BVOP. After encoding the shape information of VOP, samples of VOPs of arbitrary shape are obtained. Each VOP is divided into disjoint macroblocks. Each macroblock contains four 8×8 pixel blocks for motion compensation and texture encoding. The encoded VOP frames are stored in the frame buffer. Motion vectors are calculated between the current VOP frame and the encoded VOP frames. For the blocks and macroblocks to be encoded, their motion compensation prediction errors are calculated. The IVOPs and errors after motion compensation prediction are transformed using 8×8 block DCT and the DCT coefficients are quantized. Then, run-length encoding and entropy encoding are performed. (3) The bitstreams of each video object are composited. The shape and motion texture information of each video object are composited into a VOL bitstream. The video streams of each video object are composited into a unified bitstream output. After the video stream is compressed and encoded, the next step is to implement the functions of the network transmission part. 2.4 JRTPLIB Network Transmission Module Streaming media refers to continuous time-based media transmitted over a network using streaming technology. RTP is currently a good way to solve the problem of real-time streaming media transmission. JRTPLIB is an object-oriented RTP library that fully complies with RFC1889 design. The following describes how to use the RTP protocol for real-time streaming media programming on the Linux platform. (1) Initialization Before using JRTPLIB for real-time streaming media data transmission, an instance of the RTPSession class should be generated to represent this RTP session. Then, the Create() method should be called to initialize it. The Create() method of the RTPSession class has only one parameter, which is used to specify the port number used in this RTP session. (2) Data Transmission After the RTP session is successfully established, the real-time transmission of streaming media data can begin. First, the target address for data transmission needs to be set. The RTP protocol allows multiple target addresses in the same session. This can be done by calling the AddDestination(), DeleteDestination(), and ClearDestinations() methods of the RTPSession class. After all target addresses are specified, the SendPacket() method of the RTPSession class can be called to send streaming media data to all target addresses. (3) Data reception For the receiving end of streaming media data, the PollData() method needs to be called first to receive the sent RTP or RTCP data packets. Since multiple participants (sources) are allowed in the same RTP session, all sources can be traversed by calling the GotoFirstSource() and GotoNextSource() methods, or those sources carrying data can be traversed by calling the GotoFirstSourceWithDat() and GotoNextSourceWithData() methods. After a valid data source is detected from the RTP session, the GetNextPacket() method of the RTPSession class can be called to extract RTP data packets. After the received RTP data packets are processed, they should be released in time. JRTPLIB defines three receive modes for RTP datagrams. These receive modes can be set by calling the SetReceiveMode() method of the RTPSession class: RECEIVEMODE_ALL: The default receive mode, where all arriving RTP datagrams will be accepted; RECEIVEMODE_IGNORESOME: All arriving RTP datagrams will be accepted except from certain specific senders. The list of rejected senders can be set by calling the AddToIgnoreList(), DeleteFromIgnoreList(), and ClearIgnoreList() methods; RECEIVEMODE_ACCEPTSOME: All arriving RTP datagrams will be rejected except from certain specific senders. The list of accepted senders can be set by calling the AddToAcceptList(), DeleteFromAcceptList, and ClearAcceptList() methods. (4) Control Information JRTPLIB is a highly encapsulated RTP library. As long as the PollData() or SendPacket() methods are successfully called, JRTPLIB can automatically process the arriving RTCP datagrams and send them when necessary, thus ensuring the correctness of the entire RTP session. In this system, the underlying RTP/RTCP operations are implemented using the methods provided by the RTPSession JRTPLIB class library, and encapsulated in the CrtpTransmitter class, which inherits from the Media Sink class. Upon receiving the corresponding media frame data, the RTPSession class library's operations are used to send the data over the network. 3. Conclusion This system is based on the S3C2410 platform and Linux operating system. The acquisition program is designed using Video4Linux, and the MPEG-4 compression encoding algorithm is used. Network transmission is achieved through real-time streaming media transmission technology. The entire system is stable, reliable, easy to install, and low-cost, and can be extended to many fields such as industrial control, video conferencing systems, videophones, and remote monitoring systems.