An algorithmic approach to controlling search in three-dimensional image data

Publisher Website
Google Scholar
Abstract
In many three-dimensional imaging applications random shaped objects, reconstructed from serial sections, are isolated to display their overall structure in a single view. This paper presents an algorithm to control an ordered search strategy for locating all contours of random shaped objects intersected by a series of cross-section image planes. Classic search techniques in AI problem solving and software for image processing and computer graphics are combined here to aid program initialization and automate the search process thereafter. Using three-dimensional region growing, this method isolates all spatially connected pixels forming a structure's volume and enters image planes the least number of times to do so. An algorithmic description is given to generalize the process for controlling search in 3-D image data where little core memory is available. Phantom and medical computer tomographic data are used to illustrate the algorithm's performance.

This publication has 10 references indexed in Scilit: