基于脊波融合的射线图像增强算法  

Radiographic image enhancement algorithm based on finite ridglet fusion

作　　者：沈宽[1] 文玉梅[1] 蔡玉芳[1] 

机构地区：[1]重庆大学教育部光电及系统重点实验室ICT研究中心,重庆400030

出　　处：《计算机工程与应用》2010年第23期170-174,共5页Computer Engineering and Applications

基　　金：国家高技术研究发展计划(863)No.2006AA04Z104~~

摘　　要：提出一种基于脊波变换的射线图像增强算法,根据射线成像的特点,对射线成像系统采集信号做分段灰度变换,得到多幅图像,每幅图像含有被测工件的某种细节,再将这些图像分别做有限脊波变换,对得到的变换系数进行融合,再对融合后的系数进行有限脊波逆变换从而得到增强了的射线图像。在融合中低频系数采用基于区域方差和邻域像素相关性分析的融合策略,高频部分采用脊波变换系数绝对值最大的作为融合的高频系数,此法可以将来自不同图像的特征与细节融合在一起并且可以有效地抑制噪声。文中对融合图像质量进行了对比评价,实验结果表明,这种方法能够有效地提高图像的清晰度,在保留图像微小细节方面获得满意的结果。A radiographic image enhancement algorithm based on finite ridgelet transform is proposed.In terms of characteristic of the collected radiographic signals,in the proposed scheme the subsection of signals is mapped to 0-255 gray scale to form several gray images and then fuses these images to form a new enhanced image.In the fusing process,neighboring relativity and regional variance analysis method is adopted as the low-frequency image coefficients fusion rule.High-frequency image fusion rules are based on maximum absolute value of ridgelet coefficients.All the features and details from different gray images are fused successfully and four quantitative fusion performance criteria are used to evaluate the gain.Experimental results show that the proposed algorithm can effectively elevate image＇s definition and retains more subtle image details than the method using wavelet.

关 键 词：铁路铸件 射线检测 小波变换 有限脊波变换 图像融合 图像增强 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]