You are here
Advances Toward Understanding of Multimodal Volume Data.
This paper introduces new work in multiscale image statistics, a local framework that supports adaptive measurement of image structure where data may be represented by multiple incommensurable values. Data such as those represented by the Visible Human Project data often include multiple modalities such as color channels, multiple pulse sequences of magnetic resonance imaging, and X-ray CT data. Multiscale statistics can establish local correlations, covariances, and entropy measurements across the image. Such measurements have applications in nonlinear filtering, texture analysis, deformable registration and image segmentation.