A multi-object segmentation algorithm based on background modeling and region growing

Abstract

A multi-object segmentation algorithm based on Background Modeling and Region Growing (named as BMRG) algorithm is proposed in this paper. For multi-object segmentation, the algorithm uses Chebyshev inequality and the kernel density estimation method to do background modeling firstly. Then in order to classify image pixels as background points, foreground points and suspicious points, an adaptive threshold algorithm is proposed accordingly. After using background subtraction to get the ideal foreground image, region growing method is used for multi-object segmentation. Here, we improved the region growing method by introducing the growth seed concept for multi-object segmentation, which is calculated from the sparse matrix of quad-tree decomposition. Experimental results show that Chebyshev inequalities can quickly distinguish the foreground and background points. Multi-object segmentation results are satisfactory through seed-based region growing method. Comparison and analysis the experimental results show that the proposed BMRG algorithm is feasible, rapid and effective. © 2012 Springer-Verlag.