Adjacency Matrix Based Objects Representation for Human Identification in Images
Snekalatha Saravanakumar, Appachi Vadivel · 2009
In this paper the human object present in images are identified. K-means clustering algorithm is used for segmenting the images and Connected Component Analysis is used as post-processing step. Initially, templates are created from human posture in idle and various angles. The adjacency matrix is constructed from objects and it is matched with templates. A similarity measure has been proposed for estimating the matching.