• Majed El Helou
• Frederike Dümbgen
• Sabine Süsstrunk

Created on May 7, 2018.

This repository contains the modules and scripts necessary to reproduce the results from the paper "AAM - A Simple Assessment Metric of Axial Chromatic Aberration".

Due to the size of the full experimental image data (12GB+), we only upload the extracted features. We provide data for two different camera settings: Canon EF 50mm f/2.5 (experiment ID 4) and Canon EF 50mm f/1.8 I (experiment ID 7).

For each camera X={4,7} (experiment IDs), we provide:

distances_X.mat contains the distances of data capture for the experiment, these distances are the separation between the camera lens and the captured edge.

GaussStd2Color_X.mat contains the estimated Gaussian standard deviation (the blur circle explained in the paper) at every distance, for the color channels.

GaussStd2Nir_X.mat contains the estimated Gaussian standard deviation (the blur circle explained in the paper) at every distance, for the NIR channel.

This folder contains the Matlab script used to read the above data and create the plots shown in Figure 3 in the paper.

This folder contains source code to compute the AAM metric. The code is best understood by looking at the provided notebook demo_AAM.ipynb. In particular, it contains the code necessary to reproduce the results in tables 1 and 2.

This code is provided under the BSD license.