摘要
医学超声图像的细节特征在临床诊断中具有重要的意义。针对于传统的PM算法以及各种改进型各向异性去噪方法(Catte_PM、SRAD、CENCD等)存在边缘中的噪声点未作处理,多次迭代产生虚假边缘等缺点,通过分析具有代表性的Catte_PM各向异性模型,提出了一种结合自适应Canny算子,沿图像边缘切线方向扩散的去噪方法。该算法首先通过改进的Canny算子将图像范围分为边缘区和非边缘区;其次改进现有的扩散方法,使扩散方向只沿图像边缘切线方向进行;最后对非边缘区域采用有限次(三次)的各向同性滤波。实验结果表明,该方法能够有效地解决滤波和图像细节保护这一矛盾问题,使得图像质量有较明显的改善。
The minutiae of medical uhrasound images have great significance in the clinical diagnosis. In view of the shortcoming that the traditional PM algorithm and the various improved algorithms of anisotropic denoising （ Catte_PM, SRAD, CENCD, etc） can＇ t deal with the noise points of the edge, and that multiple iterations bring about false edge, this paper put forward a new filtering algorithm by analyzing the Catte_PM anisotropic model. It combined the adaptive Canny operator with the algorithm which spread along the direction of the image edge tangent. Firstly, the algorithm used the improved Canny operator to divide the image into the edge area and the non-edge area. Then, it improved the diffusion method which made the direction of the diffusion only along the tangential direction of the image edge. Finally, the algorithm chose the finite times （3 times） isotopic filtering to deal with the non-edge region. Experimental results demonstrate that, the algorithm solves the contradiction that is between filtering noise and protecting image details. And it obviously improves the quality of the image.
出处
《计算机应用研究》
CSCD
北大核心
2015年第7期2189-2191,2199,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(61202044)
四川省科技创新苗子工程项目(20132022)
西南科技大学研究生创新基金资助项目(14ycxjj0059)
西南科技大学研究生创新基金资助项目(14ycxjj0062)
四川省教育厅重点资助项目(12ZD1109)
