摘要
设计并实现基于高斯尺度空间理论的直方图定性匹配算法.该算法首先建立灰度直方图的尺度空间,并提取各直方图在尺度空间中的特征,包括峰点个数、峰点线结构和峰点生命期等,然后利用这些特征匹配直方图.在同一内容的图像与不同内容的图像组成的图像库中,对直方图定性匹配算法进行了实验研究.实验结果表明,该算法可较好地解决图像内容恒常性问题.
In this paper, an algorithm for qualitative matching of histograms based on scale space theory is presented. This algorithm first creates the scale space of gray histograms, and extracts features of each histogram. Then the Algorithm matches the images by means of the features,such as number of peak,peak linear structure,and peak lifetime. If the features can match each other, the Algorithm intersects the two histograms to get the final similarity ratio. Experimental results suggest that this algorithm provides useful solution to content constancy problems.
出处
《辽宁师范大学学报（自然科学版）》
CAS
2003年第1期37-41,共5页
Journal of Liaoning Normal University:Natural Science Edition
基金
国家自然科学基金资助项目(69973019)
辽宁省教育厅基金资助项目(990321081
20042101)
