改进Tamura纹理特征的图像检索方法  

Image retrieval based on improved Tamura texture features

作　　者：郝玉保[1,2] 王仁礼[3] 马军[2] 苏斌[2] 郑建华[2] 

机构地区：[1]信息工程大学测绘学院 [2]75719部队 [3]河海大学水资源环境学院

出　　处：《测绘科学》2010年第4期136-138,176,共4页Science of Surveying and Mapping

摘　　要：针对Tamura方向度纹理特征计算时直方图峰值难以确定的问题,本文提出了采用改造后的旋转不变边缘方向直方图代替方向度纹理特征;针对粗糙度纹理特征不能充分反映图像上纹理基元尺寸的大小分布的问题,提出了采用粗糙度直方图代替粗糙度的特征表达方式。实验表明,改进后的Tamura纹理特征易于获取,包含图像更多的信息,显著地提高了检索性能。For the peak of histogram is difficult to be ascertained when the texture feature of Tamura directionality is computed, the authors replaced the directionality texture feature with the revised edge directional histogram invariant to rotate. Meanwhile, for the texture feature of Tamura coarseness can＇t respond to the changes in size of texture primitive adequately, the authors replaced the coarseness with the coarseness histogram. The experiments performed on the Brodatz texture database and remote sensing images indicated that the improved Tamura texture feature could be attained easily, contain more valuable information and increase the precision of retrieval remarkably.

关 键 词：Tamura 纹理特征 粗糙度直方图 边缘方向直方图 图像检索 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP751[自动化与计算机技术—计算机科学与技术]