Swimming target detection and tracking technology in video image processing

摘要

The new system is presented for target detection swimming, tracking multiple targets in a video sequence. Since the output of the system is based on target motion, to use background subtraction to move the area within each video frame of the first segment foreground and background. A new method proposed for detecting drowning in a person swimming in a data using video footage. The proposed Swimming target detection and tracking technology based on Discrete Cosine Transform Algorithm, Target Detection Algorithm, and Traditional Cam Shift Algorithm. For background extraction and updating, use the frame-by-frame differential Discrete Cosine Transform Algorithm used to the exact motion region of the entire video section in image processing technology based analysis. Static and dynamic feature detection identification normal swimmers and drowning swimmers. Swimming-based data are collected in real time using a camera on the surface of the water, including a video image sequence with the main observation swimming and behavior analysis based system immersion and three steps to detect swimming. By combining frame subtraction and background subtraction, the advanced moving Target Detection Algorithm can detect the full area of interest. Then, with average particle filtering, analysis of Traditional Cam Shift Algorithm, it can automatically select the target for the resulting detection results and introduce a new person tracking method. Swim target detection and track results to analyze motion parameter estimation algorithms. Finally, the purpose of the set of human diagnostics and follow-up tests is to test algorithmic efficiency.