Low-Complexity Embedded Object Detection and Tracking System Design

• Main Authors: Jun-Xian Wu, 吳俊賢
• Other Authors: Ching-Han Chen
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/7w8md4

碩士 === 國立中央大學 === 資訊工程學系 === 105 === In rising of IoT, low-complexity sensor node is the trend of development. This paper implements complete object detection and tracking on a low-cost DSP platform, and verify the system performance on efficacy. Our goal is to achieve an intelligent embedded camera of low-complexity, hardware-constrained, and without operating system. For detection, because of the low-complexity, the paper utilize Approximated Median Filter (AMF) to achieve background modeling for the main method of object detection. Then particle swarm optimization (PSO) is main method which is used as tracking strategy: First, build the target model for moving object. Through PSO algorithm, the system can track moving objects in the nonlinear system. Limited on the memory and development costs, the experiments and analysis still show the efficiency. Due to using C language as development tools, the system is high portability. The proposed system can be the IoT application system case in the future.