Influence of rain on polarimetric detection of wet road surfaces: an experimental evaluation under simulated rain conditions

Mouhamadou Makhtar Fall, Samia Ainouz, Pierre-Jean Lapray and Jean-Philippe Tarel

Abstract

---

Detecting the surface condition of roads is a key challenge for assisted and autonomous navigation systems, particularly at night or under adverse weather conditions where the loss of adhesion drastically increases the risk of accidents. Conventional vision or lidar-based methods often fail to accurately estimate road wetness or slipperiness when visibility is degraded, as they rely mainly on intensity or depth cues that are highly sensitive to illumination and scattering effects. In this context, polarimetric imaging offers a promising alternative by exploiting the polarization state of reflected light to extract information about surface properties, even in the presence of disturbing reflections or scattering media. This study evaluates the capability of polarimetric imaging to discriminate between different road surface states under various rainfall conditions and illumination sonditions. We analyze the backscattering behavior of wet surfaces in a scattering environment and compare the performance of two illumination types: polarized and non-polarized light. Experiments were conducted on the CEREMA PAVIN2 platform designed for rainfall simulation, covering several scenarios (dry surface, light rain, moderate rain, heavy rain, and water film). The polarimetric camera effectively suppresses parasitic reflections from ambient light, and the use of polarized illumination enhances discrimination between rainfall levels. These findings confirm the relevance of polarimetric imaging for reliable road surface estimation, especially for autonomous and assisted vehicles operating at night or in degraded weather conditions.

Reference

---

Pdf file (42 593 Kb)

(c) SPIE