This repository contains information about equirectangular images (360 degree panoramic images), conversion to rectilinear views (selecting a particular viewing direction and "flattening" the image), a code framework to test out optimization ideas for efficiently extracting images, and web pages that explain all the work.
The blog series starting with https://intel-health.github.io/optimizing-equirectangular-conversion/1-Introduction%20to%20360%20Degree%20Representation/index.html describes equirectangular projection data format, walks through a number of available open-source solutions, and then discusses optimizations that make the conversion to rectilinear images significantly faster.
The code depends on the following:
- Intel® oneAPI Base Tool Kit. All testing used version 2023.2
- Python. Standard version](https://www.python.org/) or Intel® Distribution for Python (https://www.intel.com/content/www/us/en/developer/tools/oneapi/distribution-for-python.html). All testing was done with both installed on the machine.
- OpenCV. All testing used version 4.6 (specifically https://sourceforge.net/projects/opencvlibrary/files/4.6.0/opencv-4.6.0-vc14_vc15.exe/download).
It is recommended to install the tools listed above in the order listed. OpenCV utilizes cmake which is included in the oneAPI Base Tool Kit. It also uses Python at build time.