摘要
传统的FCM图像聚类法由于需要大量先验知识和聚类速度的原因,大大限制其在图像分割领域的应用。提出一种基于小波分解和模糊聚类相结合的图像分割算法,首先对图像进行小波变换,对于L空间得到的灰度图像利用小波多尺度分解的性质得到特征图像,利用此特征图像的一维灰度信息采用模糊C均值聚类(FCM)算法,并自动确定FCM算法聚类数和聚类中心从而完成聚类的无监督化,实现对经小波分解后的特征图像的高效快速分割。
Because of a few priori knowledge has to be needed and its slowly speed, The application of traditional fuzzy clustering algorithm is limited in the field of image segment. Proposed a new image segmentation algorithm based on wavelet decomposition and fuzzy clustering. First, divided the image into sub images with wavelet transform method; gained the feature images according the property of wavelet multi - scale decomposition of the grey - scale map in L - Space. By using one - dimensional gray information of the feature image, FCM algorithm, and automatically determined the number of clusters and cluster centers to complete the non- supervision clustering. The experiment proves feature images based on wavelet decomposition has been segmented quickly and efficiently.
出处
《计算机技术与发展》
2009年第6期121-123,共3页
Computer Technology and Development
基金
四川省教育计划重点资助项目(2006A097)
四川省科技应用基础研究项目(2008LY0115-2)
关键词
图像分割
小波分解
模糊聚类
image segment
wavelet decomposition
fuzzy clustering
