摘要
针对超声图像中的斑点噪声抑制问题,分析了经典的NL_MEANS算法去噪,提出了一种改进的算法———基于K均值聚类的NL_MEANS算法。通过引入聚类化的思想先将图像中的信息合理分类,使得分类信息具有较高的相似度,类间具有较低的相似度,利用NL_MEANS算法对分类后的图像进行去噪处理。改进算法抑制了斑点噪声,消除了传统NL_MEANS算法产生的人工伪影,保持了图像边缘和纹理信息的清晰度,实验结果表明了改进算法的有效性。
According to suppress speckle noise in ultrasound images problem, the classic NL_ MEANS denoising algorithm is analyzed, and an improved algorithm-NL MEANS algorithm based on k-means clustering is proposed. By introducing the idea of clustering, the image information reasonably is classified, makes the class information have high similarity, have relatively low similarity between classes, NL_ MEANS algorithm is reused to denoise image classified. Algorithm is improved not only sup- press the speckle noise, but also eliminates the artificial artifacts of the traditional NL_ MEANS algorithm, and keeps the sharp- ness of the image edge and texture information, the experimental results verify the effectiveness of the improved algorithm.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第3期939-942,共4页
Computer Engineering and Design
基金
国家自然科学基金重点项目(61136002)
