pyomniunwarp 0.2.5

A simple tool for unwrapping 360 degrees omnidirectional images.

Overview

This is a python package using Scaramuzza model to rectify omnidirectional images

Prerequisite

$ pip install opencv-python

Installation

$ pip install pyomniunwarp

Parameters

Prepare the parameters as the following format

kwargs = {
    "mode": "cuboid",   # cuboid or cylinder
    "version": "0.2.2",  # https://pypi.org/project/pyomniunwarp/
    "calib_results_path": calib_results_path,
    "FOV": 90,
    "output_height": 365
}

Region of Interest

Region	Direction
A	Front
B	Right
C	Back
D	Left
E	Full(B+C+D+A)
F	Front-Left
G	Front-Right

Version Specification

In version 0.2.1, default ROI is (A, D, C, B, E)
In version 0.2.2 and later, default ROI is (A, D, C, B, E, F, G)

Usage

To run the example in python

import pyomniunwarp.example

pyomniunwarp.example.run_example()

To use the calibrated model in python, prepare calib_results.txt from OCamCalib toolbox.
Edit calib_results.txt as the following example here

Put calib_results.txt under the same folder with the python script

Then in python, import as

from pyomniunwarp import OmniUnwarp

unwarper = OmniUnwarp(**kwargs)

Initialize will take several seconds. After the initializtion, perform unwarping by

imgs, masks, labels = unwarper.rectify(original_img)