小波和插值结合应用于超声图像的超分辨率算法  

Research on Super Resolution Algorithms Applied in Ultrasonic Image Based on Combining Wavelet with Interpolation

作　　者：戴光智[1] 陈铁群[2] 邱栋美 

机构地区：[1]深圳职业技术学院,深圳518055 [2]华南理工大学,广州510640 [3]北京华泰科恩科技有限公司,北京100083

出　　处：《无损检测》2009年第4期251-255,共5页Nondestructive Testing

基　　金：广东省自然科学基金资助项目(07006479);广东省重点攻关项目(2006B12401001)

摘　　要：针对超声图像分辨率较低的缺点,对现有基于小波和插值的图像放大方法进行归纳、总结和分析,提出了一种基于小波和插值的超分辨率图像处理方法,该方法选择合适的小波与插值的结合方式,采用基于区域能量的判定方法,通过适当减少低频子带的比例同时尽可能多地恢复高频子带的方法重建超分辨率图像。通过5种方法对超声图像进行对比试验,结果表明:第5种该方法能有效地提高超声图像的分辨率,对材料感兴趣区域的缺陷检测可起到"探针"作用。Aiming at the lower resolution of ultrasound image characteristics, the existing image zooming methods based on wavelet and interpolation are summarized and analyzed, on the basis of which super-resolution image processing algorithm based on wavelet and interpolation was proposed. In the algorithm, the first step is to choose a suitable combination of wavelet and interpolation, and then to use the energy determination method based on region energy. The final step is to reconstruct a super-resolution image through properly reducing wavelet lowfrequency band at the same time as much as possible restoring high-frequency band. Through contrast experiments on a single static ultrasound image by five methods, the results showed that the method can effectively improve the resolution of ultrasound images, and could play ＂probe＂ role for defect detection in an interested region of material.

关 键 词：小波变换 图像插值 超声成像 超分辨率 

分 类 号：TG115.28[金属学及工艺—物理冶金]