摘要
图像分割是将数字图像进行有效分割得到众多互不重合的区域和不同背景区域的集合。传统的一些图像分割算法在分割图像过程中损失了众多原始图像的特征。在某些领域中,需要在图像的前期分割处理中有效的保留原始图像的一些特征。对此,本文结合传统PCNN模型和互信息,将最大互信息做为图像分割的目标,提出了改进型PCNN多值图像分割方法,并在不同的图像分割中应用该算法。实验结果说明该算法能够自动对不同图像进行分割,较高的分割精度,能够很好保护细节和边缘,适应性较强,拥有良好的特征划分能力。
Image segmentation is the digital image is decomposed into a number of mutually non-overlapping target area and the background area collection. In many segmentation image segmentation algorithm,the segmentation of the image has lost many important information of the original image,such as gray details,texture details,etc. In biomedical imaging engineering,satellite remote sensing and other fields,keeping need to pre-process the image segmentation much information as possible when the original image,and therefore the need for more multi-value image clustering segmentation. In this regard,the paper PCNN model combining traditional and mutual information,with the greatest mutual information as a split target,mutual information entropy difference for the new classification criteria proposed maximum information PCNN improved multi-valued image segmentation method and its application in different image segmentation. The experimental results show that the algorithm can automatically for different image segmentation,higher segmentation accuracy,detail and edge protection can be a good,strong adaptability and the ability to have a good characteristic division.
出处
《激光杂志》
北大核心
2015年第12期63-66,共4页
Laser Journal
基金
包头市青年人才创新科研项目(2010020003)
关键词
图像分割
互信息
分割精度
特征划分
image segmentation
mutual information
segmentation accuracy
feature partition
