Publication Date

5-2022

Date of Final Oral Examination (Defense)

1-10-2022

Type of Culminating Activity

Dissertation

Degree Title

Doctor of Philosophy in Computing

Department

Computer Science

Major Advisor

Steven Cutchin, Ph.D.

Advisor

Maria Soledad Pera, Ph.D.

Advisor

Elena Sherman, Ph.D.

Abstract

The equirectangular projection is commonly used to store and transmit 360' images. However, using the equirectangular projection to store and transmit 360° images is not efficient due to its natural topographic redundancy. To generate the 360° image, captured pixels that form a spherical point cloud in the 3D space are projected onto a 2D plane using the equirectangular projection; generating redundant pixels in the process. These extra pixels in the image add extra memory requirements that have low impact in the final image quality. This dissertation presents results of research into the compression of 360° spherical imagery. It examines and provides results on pixel visual redundancy of equirectangular images, studies the efficiency of the sinusoidal projection comparing it to the equirectangular projection, and explores an alternative to the squared block Discrete Cosine Transform (DCT) to better match the 360° image contents. In addition to the examination of research questions in these areas, specific results are presented that show new techniques that improve on the current state of the art in 360° image compression.

DOI

https://doi.org/10.18122/td.1958.boisestate

Download

Files over 30MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS