We present a segmentation approach to scientific visualization that combines the definition of higher-level data, the efficient extraction of meaningful derived feature-like data from defined properties, and the effective visual representation of the extracted data. Our framework is aimed at multi-valued time-varying data sets, where, for example, grid vertices might have a multitude of associated scalar, vector and tensor quantities. This segmentation approach to massive data set exploration allows the user to focus upon regions, and interactively explore these regions efficiently. The challenge is to generate this segmented data from existing multi-valued data sets, store this data in an efficient scheme, generate the boundaries of each region, and display these boundaries to the user. We present an integrated scheme that allows a common representation for segmentation, allows it to be applied to a number of data types, and allows derived representations to be calculated. We illustrate this framework with examples from scalar-and vector-field visualization.