Current Issue Cover

郝玉保1,2, 王仁礼3, 顾立娟4(1.信息工程大学测绘学院, 郑州 450052;2.75719部队, 武汉 430074;3.河海大学水资源环境学院, 南京 210098;4.国家数字交换系统工程技术研究中心, 郑州 450002)

摘 要
A New Image Retrieval Method Using Contourlet-based Rotation-invariant Feature for Remote Sensing Images

HAO Yubao1,2, WANG Renli3, GU Lijuan4(1.Institute of Surveying and Mapping, Information Engineering University, Zhengzhou 450052;2.75719 Troops, Wuhan 430074;3.College of Water Resources and Environment, Hehai University, Nanjing 210098;4.China National Digital Switching System Engineering and Technology Research Center,Zhengzhou 450002)

There are plenty of structure edge features in remote sensing images besides texture features. How to capture the information contained in these features is a key point to retrieve remote sensing image effectively. Given that Contourlet transform has the function of multi-scale geometric analysis in the discrete domain, we have put forward a novel approach for image retrieval, which extracts the energy features of Contourlet subbands to obtain the edge direction information under different scales. For the ability differences in capturing different direction features, we have adopted an orthogonal compensation method to remedy them, and realized rotation invariant image retrieval through processing them with Fourier operator. Experiments on remote sensing images demonstrate this retrieval algorithm is especiauy effective for those images containing rich structural edge features.