Object Tracking Based on Color Space Technology and Particle Filter

• Main Authors: Shi-Ming Chen, 陳仕銘
• Other Authors: Yi-Nung Chung
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/49650816183988335455

碩士 === 國立彰化師範大學 === 電機工程學系 === 101 === ABSTRACT With the popularity of surveillance systems and the progress of digital image processing technology, real-time surveillance system gradually becomes an important research objective. In general visual-images, the color characteristics of targets are easy to be changed because of the impact of changing illumination. Therefore, this thesis proposed a method by using the color space transformation. The system transfers RGB color space to HSV color space by using the character to reduce the noise of shadow and the impact of changing illumination. In this thesis, the system used a series of image processing methods to segment the background and the foreground object properly. Considering the RGB image is vulnerable to changing illumination, it would transform the color space to extract the color characteristics of targets after getting the foreground target. In the video tracking, this system is to combine the particle filter. It set the color characteristics as the determinant basis of likelihood value in particle filter to achieve the purpose of target tracking. It can make the surveillance system more robust and perfect. Keywords: Real-time surveillance system, Color space transformation, Particle filter, Likelihood value.