Orthagonalize

The goal of this tool is to detect the perspective distortion of a image of a planar surface using aruco markers and then correct it.

Setup

Prerequisites

Python 3.9
Poetry

Installation

git clone http://github.com/h3ll5ur7er/orthagonalize
cd orthagonalize
poetry install

Usage

Generate Aruco Markers

the default marker dictionary has 7x7 bits with a 1 bit border and 1000 entries. This can be adjusted inside the config.py file. the default number of pixels per bit is 254.

poetry run python -m orthagonalize generate <id> [<pixels_per_bit>]

for example to get a marker with the id 42 and a resulting image size of 900x900 pixels ((7 bits + 2*1 bit border)*100 pixels per bit):

poetry run python -m orthagonalize generate 42 100

print the marker at a known absolute size (e.g. insert the image into a word document and set the size to 50mm x 50mm)

Orthagonalize image

poetry run python -m orthagonalize image <path> <marker_size_mm>

The following steps are performed:

the marker will be detected within the image.
a homography matrix will be calculated to align and center the marker within a 15360x8640 image.
the input image is then warped using the homography matrix.
the corners of the input image are projected using the homography matrix.
the projected corners are clamped to the 15360x8640 image.
the image is cropped to the clamped corner bounds.
the cropped image is displayed and saved to the same directory as the input image.

Orthagonalize stream

poetry run python -m orthagonalize stream <source> <marker_size_mm>

the stream source can be a camera index or a video file path.