Fast Perspective Views of Images Using One-Dimensional Operations

Abstract
Generating arbitrary perspective views of images portraying 3D surfaces can involve extensive computation and data I/O time because of the problems in determining visibility, and performing hidden-point removal. Appropriate 1D transforms of an image can allow hidden-point removal and perspective projection to be performed on scan lines or columns of these transforms. Perspective view generation then reduces to a series of extremely fast 1D operations. As a result, exact perspective views of 3D surfaces of unlimited size can be generated very much more rapidly than hitherto possible, making this a feasible interactive tool in image analysis. Maximum speed is possible when entire data sets can be stored in random access memory (RAM). However, the scan-line nature of the algorithm also allows sequential processing of data stored on disk, and fast image transposition methods allow the generation of views of surfaces much larger than available memory.

This publication has 13 references indexed in Scilit: