单幅图像局部特征分层模糊挖掘算法仿真  

Simulation of Hierarchical Fuzzy Mining Algorithm for Local Features of Single Image

作　　者：牛庆丽[1] 王黎明[2] NIU Qing-li;WANG Li-ming(School of Big Data and Artificial Intelligence,Zhengzhou University of Science and Technology,Zhengzhou Henan 450064,China;School of Information Engineering,Zhengzhou Universtiy,Zhengzhou Henan 450001,China)

机构地区：[1]郑州科技学院大数据与人工智能学院,河南郑州450064 [2]郑州大学信息工程学院,河南郑州450001

出　　处：《计算机仿真》2024年第8期195-199,共5页Computer Simulation

基　　金：国家自然科学基金(61806181)。

摘　　要：为了能够深入获得图像纹理特征信息,提高后续数据识别精准度。因此,提出了单幅图像局部特征分层模糊挖掘算法。通过直方图均衡化方法,将图像中灰度值集中到对应灰度等级区域。由不均匀分布向均匀分布状态转换,拓展像素的灰度动态范围。分析图像局部特征复杂度与差异度,求出相邻模板灰度等级,得到局部复杂性和差异度矩阵,采用Laplace算法对图像局部特征推荐分类,根据推荐级别分层模糊挖掘所选特征,以此实现对单幅图像的局部特征分层模糊挖掘。通过实验证明,所提算法可准确挖掘出图像特征,不同层级纹理信息都完整,且挖掘时间保持在0.25s内。In order to obtain image texture feature information and improve the accuracy of subsequent data recognition in-depth,this paper proposed a hierarchical fuzzy mining algorithm for local features of single images.Firstly,the histogram equalization method was used to concentrate the gray values in images into gray-level regions correspondingly and transform them from a non-uniform distribution to a uniform distribution,thereby expanding the dynamic range of the gray level of the pixel.Then,the complexity and difference degree of local image features were analyzed to calculate the gray level of adjacent templates,thus obtaining a matrix of local complexity and difference.Furthermore,the Laplace algorithm was adopted for recommending and classifying the local features.According to the recommended level,selected features were hierarchically mined.On this basis,the hierarchical fuzzy mining of local features of a single image was achieved.Experimental results show that the proposed algorithm can accurately mine image features as well as complete texture information at different levels and maintain mining time within O.25s.

关 键 词：单幅图像 局部特征 特征分层模糊挖掘 差异度矩阵 局部差异度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]