融合时域信息的修正分水岭视频分割新方法

提出一种将时域信息融入分水岭的视频分割新方法,以帧间变化检测为基础,通过运动边缘信息得到对象的初始模型,利用时域信息得到前景和背景的标识,结合提出的彩色多尺度形态学梯度算子进行分水岭分割,得到具有精确边界的视频对象,对慢变和快变的目标均有良好的效果,能够检测新出现的运动对象和现有对象的消失,能够定位和跟踪运动目标.继承了变化检测和分水岭算法速度快的优点,克服了两者易受噪声影响的缺点.

A New Video Segmentation Method Based on Modified Watershed Combined with Temporal Information

A new video segmentation method which combines temporal information with modified watershed is proposed. For localization of moving objects in the image sequence, initial binary object model is obtained by moving edge detection technology based on difference between two successive frames. Markers of foreground and background for watershed are derived from temporal information. Morphological gradient reconstruction is implemented using color multi-scale gradient operator and modified markers, which results in video object with precise boundary. The new algorithm can detect the appearance of new objects as well as the disappearance of existing objects, can detect both rapid and slow moving object effectively and can locate and track moving object. It has the virtue of low complexity inherited from change detection and watershed and overcomes their disadvantage of sensitivity to noise.