Gridding Multivariate Positive Data for Real Time Visualization

Common visualization methods require an underlying grid. For visualization of scattered data samples it is required to approximate the data at the same grid using some interpolation technique. It is common that the data samples are positive and representing the quantities for which negative value is meaningless. For example mass, volume and density are meaningless when negative. Modified quadratic Shepard method is a commonly used method for gridding purposes. However it does not preserve positivity for inherently positive data sets. This paper discusses the problem of gridding inherently positive data sets for real time visualization applications using modified quadratic Shepard's method. Key requirement for an algorithm to be used for real time application is its predictable timing behavior. We present an efficient and deterministic alternative quadratic Shepard method as a solution to the problem of visualization of multivariate positive data in real time