基于最小描述长度（MDL）原则的无监督多尺度彩色图像分割

Unsupervised multiscale color image segmentation based on MDL principle.

作者信息

Luo Qiming, Khoshgoftaar Taghi M

机构信息

Department of Computer Science and Engineering, Florida Atlantic University, Boca Raton, FL 33431, USA.

出版信息

IEEE Trans Image Process. 2006 Sep;15(9):2755-61. doi: 10.1109/tip.2006.877342.

DOI:10.1109/tip.2006.877342

PMID:16948319

Abstract

We present an unsupervised multiscale color image segmentation algorithm. The basic idea is to apply mean shift clustering to obtain an over-segmentation and then merge regions at multiple scales to minimize the minimum description length criterion. The performance on the Berkeley segmentation benchmark campares favorably with some existing approaches.

摘要

我们提出了一种无监督多尺度彩色图像分割算法。其基本思想是应用均值漂移聚类来获得过度分割，然后在多个尺度上合并区域，以最小化最小描述长度准则。在伯克利分割基准测试中的性能与一些现有方法相比具有优势。