Segmentation driven semantic information inference from 2.5D data

Semantic information retrieval from unorganized point clouds becomes necessity for incoming technology such as 3DTV. Besides we surrounded with planar, nearly planar and partially planar things. With this motivation we aim to find planar structures in 2.5D point clouds. With the Hough Transform found in literature, Recursive Hough Transform and Hough Trasform with segmentation algorithms, which are variations of the original algorithm obtained by us, are implemented. K-Means and Mean-shift algorithms, which are popular segmentation methods in 2D, are adapted to 3D with/without color information and their performance analysis are presented.