The texture as an image content is a fundamental descriptor to represent an image. In this paper, we propose a texture featuring technique based on human visual system (HVS) for MPEG-7 homogenous texture descriptor [1,2]. Psychophysical experiments have shown that the response of visual cortex is turned to a band-limited portion of the frequency domain. It gives an evidence that the brain decomposes the spectra into perceptual channels which correspond to bands in spatial frequency domain[18]. It is generally believed that, concerned with spatial vision, best-adopted sub-band representation of HVS is a division in orientation axis(or angular axis) and radial axis. In this paper, we describe the texture in polar frequency space that is believed to fit HVS behavior.
Among many texture feature description metnods [3,4,5], a psychovisual filter such as a Gabor filter with octave frequency channels is popular and has been reported as a good texture descriptor. In this paper, psychovisual filtering is performed in polar frequency space, where sufficient sampled data is available in lower frequency regions. However, the filtering in Cartesian space has insufficient sampled data in those regions. Lower frequency regions are believed to be important for global texture extraction.
To extract texture feature, we calculate energies in filter banks in frequency space. Than, the clssification is performed to find texture indexing in the image database.
To verify the usefulness of the proposed descriptor for contents featuring of texture, computer simulations and experiments with MPEG-7 image database were performed.
And we propose a novel MPEG-7 based multimedia indexing/retrieval system as application of MPEG-7 texture descriptors. In this paper, MPEG-7 descriptor and descriptor scheme are hidden into the original data using data hiding and watermarking technique. The hidden data is used as a query for the multimedia indexing/retrieval system, which is based on MPEG...