一种改进的小波模极大值不变矩算法  

AN IMPROVED WAVELET MODULES MAXIMUM INVARIANT MOMENTS ALGORITHM

作　　者：李丽君[1] 

机构地区：[1]辽宁石油化工大学理学院,辽宁抚顺113001

出　　处：《计算机应用与软件》2009年第3期249-250,264,共3页Computer Applications and Software

摘　　要：一种改进的小波模极大值不变矩算法,首先利用小波模极大值在多尺度上的表现与Lipschitz指数的关系去除图像边缘的噪声点,然后采用动态确定阈值的方法,利用前一个尺度下模图像的平均模值作为后一个尺度下模图像的阈值,最后将此算法应用于汽车图像库。实验表明,改进后的算法能更有效地检测到图像在各个尺度下的边缘信息,检索性能明显优于原算法。In this paper, an improved wavelet modulus maximum invariant moments algorithm is proposed. The first step of this method is to remove the noises of image edge using wavelet modulus maximum＇ s performance in multi-scale and Lipschitz exponent relation, and then adopts dynamic threshold, that is, former scale modulus image＇ s mean modulus is used as next scale＇ s modulus image threshold. Finally, this arithmetic is tested in image databases of cars. Experiment results show that the improved algorithm can detect edge information in each scale effectively, and achieves superior retrieval performance than previous algorithm.

关 键 词：基于内容的图像检索 小波模极大值 不变矩 Lipschitz指数 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] TP391.41[自动化与计算机技术—控制科学与工程]