摘要
A novel method of histogram analysis for background extraction in video image is proposed, which is derived from the pixelbased histogram analysis. Not only the statistical property of pixels between temporal frames, but also the corrvlation of local pixels in a single frame is exploited in this method. When carrying out histogram analysis for background extraction, the proposed method is not based on a single pixel but on a 2 × 2 block that has much less computational quantities and can extract a sound background image from video sequence simultaneously. A comparative experiment between the proposed method and the pixel-based histogram analysis shows that the proposed method has a faster speed in background extraction and the obtained background image is better in quantity.
A novel method of histogram analysis for background extraction in video image is proposed, which is derived from the pixelbased histogram analysis. Not only the statistical property of pixels between temporal frames, but also the corrvlation of local pixels in a single frame is exploited in this method. When carrying out histogram analysis for background extraction, the proposed method is not based on a single pixel but on a 2 × 2 block that has much less computational quantities and can extract a sound background image from video sequence simultaneously. A comparative experiment between the proposed method and the pixel-based histogram analysis shows that the proposed method has a faster speed in background extraction and the obtained background image is better in quantity.
基金
ThescienceandTechnologyDevelopmentPlanProjectofRailwayMinistry(No.2003X040A)andtheSouthwestJiaotongUniversityScienceFoundation(No.2002A04).
