Effective labeling for an image indexing system requires all objects in the image to be identified. This identification process can be performed by extracting components of the objects and grouping these components together. We propose the use of image segmentation techniques as a first step to solve the problem of extracting these components automatically. The difficult task is how the grouping of these components is performed. This paper presents an approach in region grouping using high level information. This information permits image segments grouped into "more meaningful" regions. In this paper, we present the issues and problems involved in region grouping. Some experiment results are presented.