摘要
目前从视频图像中分割出完整的运动对象仍缺少可用于一般性场合的算法。将视频图像的方向信息测度、颜色和运动信息相结合的视频对象分割算法可以解决这一问题。通过对序列图像中的连续帧进行帧差计算,利用方向信息测度和相似区域判定函数来找到特征相同的区域,并判断相邻子区域是否应该合并,然后再结合序列图像中的运动区域来提取运动对象。同时采用多分辨率分析来提高运算速度。最后,利用彩色图像边界序列图像检测方法来得到准确的区域边界。
A new approach to extract video objects from video sequence by combining of orientation information measure, color and motion information is presented. With the different image of several frames in image sequence, the key point of this approach is to use the orientation information measure and color information to find regions of homogenous characteristics. Then use the function of similar region to find regions of homogenous characteristics, and to determine to whether to combine the regions. With the motion difference of image sequences, extract the moving objects. A new multi-resolution technique is used to achieve optimal performance. Finally, the accurate region boundaries are obtained by using the information of color image edge detection.
出处
《计算机工程与设计》
CSCD
北大核心
2006年第10期1777-1779,共3页
Computer Engineering and Design
基金
交通部博士基金项目(200332522501)
关键词
方向信息测度
视频对象分割
序列图像
运动区域检测
orientation information measure
video object segmentation
image sequence
motion region detection
