Carleton University - School of Computer Science Honours Project
Winter 2019
Synthesis of Random Phase Textures from
Non-Uniform Samples
ABSTRACT
Micro textures are a class of texture which, when Fourier transformed,
have their characteristics (to the human eye) described mainly by the
magnitude of the Fourier coefficients and not the phases. An algorithm
was created to make new, visually alike but whole articial, textures from
image samples of microtextures by altering the phases of their Fourier
coefficients. These new textures are known as Random Phase Textures.
The goal of this paper was to modify this algorithm so that the image
samples need not be rectangular in shape (non-uniform). This was done
through slight modication to and mathematical analysis of the nature
of the Fourier transform. Ultimately, we were able to modify the original
algorithm to accept and process non-uniform samples and produce result
we would expect from the original Random Phase Texture, thus opening
new applications and broadening the use cases for the Random Phase
Texture algorithm.