摘要
针对序列图像的变化检测,提出了一种基于强边缘块的时域多分辨率图像分割算法.首先在具有明显边缘特征的图像序列中,提取图像强边缘块的直方图,然后,利用时域多分辨算法,比较不同帧之间的图像差异.这种方法不但能够检测出图像的变化,而且能够确定图像变化的类型.在阈值的选取过程中,采用了自适应阈值方法,实验证明这种方法取得了较好的效果.
According to the detection of changes in image sequences,an algorithm of multi-resolution steps image partition based on strong edge blocks is proposed in this papers. Firstly, the histogram is computed from the strong edge blocks in the image sequences. Then, different images are compared by algorithm of multi-resolution steps. This method can not only further detect the changes of image, but also judge the types of changing images. In the process of selecting thresholds, the adaptive thresholds are used. The experimental results show that this method is promising.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2005年第9期1408-1410,共3页
Acta Photonica Sinica
关键词
强边缘块
多分辨率
图像分割
自适应阈值
Strong edge blocks
Multi-resolution steps
Image partition
Adaptive threshold
