Image interpolation and extrapolation using vectors and pixel time series
The primary goal of this project is to develop and test two methods for generating intermediate images from two or more parent digital images in sequence. The parent images are those of man made objects that differ in viewing angle. Thus the intermediate images that are generated differ from the original image in aspect and scale only. An additional goal is to develop an autoregressive-model based method that allows for the extrapolation of an image from a history of images taken at different instances of time. In the implementation of the first method only two parent images are needed to create an in-between view. First, the principal vector of each parent image is defined. The parent image with the longest principal vector is then manipulated to satisfy the scale and angle characteristics of the intermediate image. In contrast, for the second method, a finite set of parent images are processed into a subset of intermediate images using the cubic spline interpolation of pixel time responses. (Abstract shortened by UMI.)
Jackson, Erika, "Image interpolation and extrapolation using vectors and pixel time series" (2005). ETD Collection for University of Texas, El Paso. AAI1430225.