Region growing is a procedure that clusters pixels or subregions into larger regions based on predefined criteria. The basic approach is to start with a set of “seed” points and from these grow regions by appending to each seed those neighboring pixels that have properties similar to the seed. Homogeneity is an important property of region and main segmentation criterion in region growing which is usually based on grey level, texture, color, etc. (Hojjatoleslami and Kittler, 1998; Ikonomakis et al., 1997). Generally, regiongrowing algorithm can make a better performance on noisy image than other segmentation methods. But images segmented by region-growing method often contain either too many regions (undergrowing) or too few regions (overgrowing) as a result of non-optimal parameter setting (Sonka et al., 1994). Therefore, setting optimal parameters is important for efficient segmentation by region growing