摘要
针对现实中真实图像含有大量的噪声及复杂纹理,提出了基于局部空间密度峰值的图像分割算法。该算法采用基于局部空间信息下的距离,结合密度峰值,通过捕捉图像的内部局部结构特征,获得较为准确的区域分割结果,最后对分割区域的离群点作二次划分,来进一步消除噪声及纹理的影响。通过对三类不同的图像(噪声图像、纹理图像、自然图像)进行图像分割的实验结果显示:相比K-Means和FCM算法,该算法可有效地分割含有噪声和复杂纹理的图像。
To effectively perform complex images with a large amount of noise and texture in reality,the paper proposed an image segmentation algorithm based on local space and density peaks. The algorithm used the local space distance,combined with the density peaks,to obtain the reliable and accurate segmentation results through capturing the inherent structure of image. Finally,second partition of the outliers can further eliminate the impact of noise and texture. Compared with the K-Means algorithm and the fuzzy C-Mean（ FCM） algorithm by segmenting noise images,texture images and natural images,the proposed algorithm is efficient to segment the complex images containing noise and texture.
出处
《信息技术》
2017年第9期84-87,90,共5页
Information Technology
基金
国家自然科学基金资助项目(61001139)
关键词
图像分割
局部空间距离
密度峰值
离群点
image segmentation
local spatial distance
density peaks
outlier
