Color image segmentation using a region growing method

TitleColor image segmentation using a region growing method
Publication TypeConference Paper
Year of Publication2001
AuthorsWesolkowski, S., and P. Fieguth
Conference Name2001 Advanced Imaging Conference
Conference LocationRochester, NY
Keywordscolor image segmentation, highlight invariance, region growing, shading invariance

A new region growing color image segmentation algorithm is presented in this paper. This algorithm is invariant to highlights and shading. This is accomplished in two steps. First, the average pixel intensity is removed from each RGB coordinate. This transformation mitigates the effects of highlights. Next, region seeds are obtained using the Mixture of Principal Components algorithm. Each region is characterized using two parameters. The first is the distance between the region prototype and the candidate pixel. The second is the distance between the candidate pixel and its nearest neighbor in the region. The inner vector product or vector angle is used as the similarity measure which makes both of these measures shading invariant. Results on a real image illustrate the effectiveness of the method.