Title :
Unsupervised multiscale color image segmentation based on MDL principle
Author :
Luo, Qiming ; Khoshgoftaar, Taghi M.
Author_Institution :
Dept. of Comput. Sci & Eng., Florida Atlantic Univ., Boca Raton, FL
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 compares favorably with some existing approaches
Keywords :
image colour analysis; image segmentation; pattern clustering; Berkeley segmentation benchmark; MDL principle; mean shift clustering; minimum description length criterion; unsupervised multiscale color image segmentation; Boats; Clouds; Clustering algorithms; Filters; Image coding; Image color analysis; Image segmentation; Image texture analysis; Merging; Smoothing methods; Mean shift clustering; minimum description length (MDL); multiscale analysis; region merging; segmentation;
Journal_Title :
Image Processing, IEEE Transactions on
DOI :
10.1109/TIP.2006.877342
