Quantitative machine vision assessment of mar visibility of TPO surfaces
Conference Paper
Overview
Overview
abstract
We propose a statistical solution for mar feature analysis of polymer surfaces. Previous analytical methods are unable to correctly characterize the large-surface area features of mar damage. Our method considers the first moment of foreground image statistics as a luminosity curve for determining onset of visibility of mar damage. The onset of visibility is determined by a commonly accepted 3% of contrast against background luminosity over the course of the mar. We show that this solution is consistent with human observation for most cases, and can be used as a measure of aesthetic mar resistance performance.