基于小波分析的图像插值技术研究(英文)  

Wavelet-based Interpolation of Medical Images

作　　者：陈家新[1] 王伟静[2] 

机构地区：[1]北京理工大学计算机科学与技术学院,北京100080 [2]河南科技大学电子信息工程学院,河南洛阳471003

出　　处：《系统仿真学报》2007年第1期145-148,152,共5页Journal of System Simulation

基　　金：河南省自然科学基金资助(0411010200)

摘　　要：基于小波分析理论,提出一种医学图像断层间的插值方法。通过对医学图像的小波变换,插值图像边缘对应的高频子图,得到在轮廓上过渡的图像;对小波变换的低频子图,在小窗口邻域内利用灰度直方图统计得到其概率灰度值并寻找最佳插值匹配点,使得插值图像在灰度上得到很好的过渡;通过小波逆变换得到最终的插值图像。通过与以往算法的对比和分析,本算法所获得的插值图像在形状和灰度方面均达到了较高的清晰度,满足断层图像的插值要求,可有效用于医学图像的三维重构。Based on the theory of wavelet analysis, a new interpolation method for medical slices was proposed. The algorithm deconstructs the original images with wavelet transforms and interpolates the high frequency wavelet coefficients of the edge points to get the contour of the new interpolated image. For the wavelet low frequency sub-images, it obtains their probability gray values in a small neighborhood window using the gray histogram statistics, seeks the best interpolation matching points and obtains the gray transition for the interpolation image. Finally, it obtains the final interpolation image throuzh the wavelet inverse transform. Through contrast and analysis with the former algorithm, the interpolation image arising from this algorithm has achieved high clarity in shape and gray-level, meeting the requirements of medical image interpolation, and can be used effectively in medical image 3D reconstruction.

关 键 词：插值 小波变换 匹配点 均值 邻域 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]