Initial segmentation for knowledge indexing

Abstract

A framework f o r bottom-up initial segmentation of color images is proposed in which the role of initial segmentation is restricted to generating indices into a knowledge base of object models. The im.portance of primitive knowledge, feature integration and feature salience is discussed in the case of color and texture.

Image segmentation based on the integration of colour–texture descriptors—A review

“…In [20], the authors discuss a method that employs the colour-texture information for the model-based coding of human images, while Shigenaga [21] adds the spatialfrequency texture features sampled by Gabor filters to complement the CIE Lab (CIE is the acronym for the Commission Internationale d'Eclairage) colour image information. In order to capture the colour-texture content, Rosenfeld et al [22] calculated the absolute difference distributions of pixels in multi-band images, while Hild et al [23] proposed a bottom-up segmentation framework where the colour and texture feature vectors were separately extracted and then combined for knowledge indexing.…”

Review of Graph, Medical and Color Image base Segmentation Techniques

“…In [14], the authors discuss a method that employs the colour-texture information for the model-based coding of human images, while Shigenaga [15] adds the spatial frequency texture features sampled by Gabor filters to complement the CIE Lab (CIE is the acronym for the Commission International d'Eclairage) colour image information. In order to capture the colour-texture content, Rosenfeld et al [16] calculated the absolute difference distributions of pixels in multi-band images, while Hild et al [17] proposed a bottom-up segmentation framework where the colour and texture feature vectors were separately extracted and then combined for knowledge indexing. In papers [113,[115][116][117][118][119][120][121][122]127,128,132,133] extraction of colour image and texture are derived as a sequence of serial processes.…”