基于多背景模型的海面运动目标检测

Multiple Background Model-Based Moving Target Detection on Sea Surface

提出一种在反射光较强的条件下基于多背景模型的海面运动目标检测方法.使用基于统计模型的变化检测方法将被监控场景区分为海浪波动显著的背景区域和海浪波动不显著的背景区域;对这两种区域中的像素点分别以Weibull分布模型和Gauss分布模型建立背景模型;使用建立的背景模型检测海面的运动目标.实验结果表明,在海面反射光较强且波浪波动较大的情况下,用该方法可以比较准确地检测到海面的运动目标.

<Abstrcat> A new method based on multiple background model for moving target detection on the sea surface is proposed. It can detect the moving target on the sea surface when there is heavy sea clutter and the reflected light is strong. The monitored area is first partitioned into two regions using the statistical model-based change detection method, where the sea clutter in one region is heavier than the other. Then Weibull distribution is used to create the background model for pixels in the region having heavier sea clutter and Gaussian distribution is used for the other region. Finally the background model is used to detect moving target on the sea surface. Experimental results show that the proposed method can detect moving target effectively even when the sea clutter is heavy and the reflected light is strong.