摘要
研究了基于形态滤波器组的图像分割预处理算法。首先用形态开闭滤波器组产生分割的参考图像,然后提出一个二值门限确定方法,用二值化方法提取区域初始标记的预处理方法。它已应用于多个视频序列的首帧初始目标分割,实验结果证明了该方法的有效性。
A pre-segmenlation method based on a morphological filter bank and a binary marker extraction is proposed to deal with over-segmentation. The filter bank is composed of eight opening-closing operations and the marker extraction algorithm is performed on the binary watershed result to get initial marked regions. Experimnts on the first frames of several video sequences demonstrate the validity of segmentation.
出处
《计算机工程》
CAS
CSCD
北大核心
2003年第19期154-155,194,共3页
Computer Engineering
基金
国家自然科学基金资助项目(69772026)
广东省自然科学基金资助项目(970484)
