Abstract : This paper first analyzes the current development status of video surveillance systems. Addressing the prevalent back-end centralized processing model in current intelligent video surveillance research, it proposes a model where video processing is performed at the monitoring terminal. After acquiring video, the monitoring terminal performs motion object detection, recognition, and tracking algorithms. The resulting motion object feature information is then sent to the monitoring center via the network. A DSP-based intelligent video surveillance embedded terminal is designed. Based on system design requirements, TI's TMS320DM642 DSP is selected as the main chip for the monitoring terminal. The chip selection and hardware circuit connections for key functional modules such as the video A/D module, external expansion memory module, and Ethernet module are then designed. Compared with most video surveillance system solutions that perform intelligent video analysis and processing at the back end, this design offers advantages such as high real-time performance, good stability, and easy system integration. Keywords : Intelligent video surveillance; TMS320DM642; Embedded system 0. Introduction Video surveillance systems have now entered the era of third-generation network digital monitoring. While video surveillance systems have seen significant improvements in functionality and performance, they remain limited by inherent factors, resulting in some shortcomings in security and practicality. These limitations include: 1) Personnel limitations: Humans cannot continuously monitor video for extended periods, leading to false alarms and missed alarms; 2) Difficulty in data analysis: The massive amounts of video recorded by the monitoring center result in significant redundancy during analysis; 3) Long response times: Traditional video surveillance systems typically rely on security personnel to respond to and handle security threats, which is insufficient for threats requiring real-time responses. Intelligent video surveillance incorporating computer vision technology is one of the most cutting-edge application models in networked video surveillance, extracting, describing, tracking, recognizing, and analyzing the behavior of moving objects in video image sequences. Most intelligent video surveillance research employs backend intelligent video analysis and processing techniques. However, processing large amounts of video at the backend not only makes system integration difficult but also imposes a complex computational burden, impacting the system's real-time performance. Embedded technology, naturally linked to real-time and stability, represents the future direction of video surveillance terminals. This paper designs a DSP-based intelligent video surveillance embedded terminal. After acquiring video, the monitoring terminal performs motion object detection, recognition, and tracking algorithms. The resulting motion object feature information is then sent to the monitoring center via the network. The article is organized as follows: Part 1: Embedded Terminal Hardware Selection; Part 2: Overall Description of the Embedded Terminal; Part 3: Hardware Design of Major Modules; Part 4: Summary. Download the full text of "Hardware Design of Embedded Intelligent Video Surveillance Terminal Based on DM642" .