Weighted pixel aggregation segmentation on outdoor scene images

In image analysis, segmentation is the process of partitioning a digital image into homogeneous and self consistent region. A different approach that has recently gained popularity is to apply graph algorithms to segmentation. This paper proposes segmentation by weighted approach and hence compares with segmentation on an outdoor scene image based on background recognition and perceptual organization. This model is developed to capture structural relationship among the constituent parts of the structured objects. Hence it groups the objects together without any prior knowledge of the specific objects. This segmentation algorithm is purely inspired by Perceptual Organization Method and further this method incorporates a weighted aggregation approach integrated into the graph coarsening scheme and compares and evaluates the two approaches.