Abstract: Based on multi-exemplar affinity propagation clustering, a fast segmentation algorithm is
proposed for big data images. The proposed algorithm preprocesses an input big data image by mean
shift algorithm to form segmented regions that preserve the desirable discontinuity characteristics of
images. The numbers of segmented regions, instead of the numbers of image pixels, are considered as
the input data scale of multi-exemplar affinity propagation clustering algorithm. The average of the
color vectors in each region is calculated and considered as an input data point of multi-exemplar
affinity propagation clustering algorithm. Euclidean distances between regions are regards as
similarity measure index, and then the multi-exemplar affinity propagation clustering algorithm is
applied to perform globally optimized clustering and segmentation based on similarity matrix.
Experimental results illustrate that the proposed algorithm has superior performance and less
computational costs compared for big data image segmentation.

Key words: multi-exemplar affinity propagation, big data, image segmentation