基于分形特征的二值图像检索方法的研究  

Study of Binary Image Retrieval Method Based on Fractal Behavior

作　　者：李粤青[1] 蒋金山[1] 汪国强[1] 张佳温[1] 

机构地区：[1]华南理工大学应用数学系,广州510640

出　　处：《计算机工程与应用》2004年第17期102-105,共4页Computer Engineering and Applications

摘　　要：论述了基于分形特征的二值图像检索方法。图像的内容由4种特征来描述:图像分形维数、图像分形矢量、边界分形矢量和骨架分形矢量。实验表明该方法计算简单、有效,匹配快速,检索结果比较理想,只要查询图像在图库中,就一定能通过该方法检索出来。另外,该方法还具有一定的鲁棒性,证明这种检索方法是具有较大实用意义的。In this paper,a novel approach to feature extraction based on fractal theory is presented as a powerful technique in image retrieval.This paper presents a method that can be applied to extract the feature of binary images.The content of image is depicted by four features:fractal dimension of image,fractal vector of image,fractal vector of edge and fractal vector of skeleton.It is constructed by a hybrid feature extraction combining fractal theory,edge detection and skeletonization.Through extensive experiments with a database of2000trademark images,the method we demonstrated is robust and flexible and a high recognition rate is reported.

关 键 词：分形维数 分形矢量 图像检索 轮廓 骨架 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]