中文摘要：

嵌入式技术的飞速发展推动了视频采集领域的技术革新，使得视频监控系统越来越朝着小型化、智能化、嵌入式化、远程化的方向发展.文中以基于S3C6410处理器的ARM板为硬件平台，采用Linux为嵌入式操作系统，建立了一个视频采集服务器；以 PC 机为客户端，实时地远程显示视频---建立了一个远程视频监控系统.系统利用Linux内核提供的视频数据采集接口V4L2控制USB摄像头来采集图像，利用网络套接字将编码后的图像数据传输至客户端；在客户端将图像解码并转化为 RGB 格式后，利用 GTK 把转化后的图像显示出来.针对H.264运动估计算法的两点不足，提出了动态搜索范围策略和搜索点分组策略.实验结果证明：在不影响重建图像质量和编码码率的前提下，改进的算法有效地降低了算法的编码时间和运动估计时间，提高了编码过程的实时性.通过本文的设计方案能够获得稳定清晰的图像，实现了远程视频监控.

英文摘要：

The rapid development of embedded technology promotes innovation in the field of video capturing, making video surveillance system towards smaller, smarter, more embedded and longer-distance fashion. A video acquisition server is established with ARM board based on S3C6410 processor and Linux served as hardware and software system respectively. Together with a PC client, remotely and real-time displaying video, a long-distance video surveillance system is set up. By control of USB camera through video data acquisition interface, V4L2, the system can acquire images. The encoded image data is sent to clients through Internet sockets. The image is displayed on the client after the accepted image has been already decoded and converted to RGB format. Two methods, dynamic search window and search points-grouping, have been proposed to solve two short points of motion estimation algorithm in H.264. The experimental result shows compared to the original algorithm, the improved algorithm can greatly decrease encoding time and motion estimation time, while maintain the image quality and encoding bitrate. The system designed here can get continued images, implementing video monitoring remotely.