An Improved Approach For Simple Objects Detection And Image Segmentation

摘要

We present a framework for image segmentation and simple objects detection which lead to automatic image annotation in this paper. In order to stress its independence of a specific image segmentation approach we have modified two well known region growing algorithms, i.e.,watershed and recursive shortestspanning tree, and compared them to their traditional counterparts. Focusing on semantic analysis of images, it contributes to knowledge-assisted multimedia analysis and bridging the gap between semantics and low level visual features. We also introduce a methodology to improve the results of image segmentation, based on contextual information and we also propose a context representation approach to use semantic region growing.