Skip to main content

Research Repository

Advanced Search

Image based analysis of visibility in smoke laden environments

Zhang, Qihui


Qihui Zhang


P. A (Philip A.) Rubini


This study investigates visibility in a smoke laden environment. For many years, researchers and engineers in fire safety have criticized the inadequacy of existing theory in describing the effects such as colour, viewing angle, environmental lighting etc. on the visibility of an emergency sign. In the current study, the author has raised the fundamental question on the concept of visibility and how it should be measured in fire safety engineering and tried to address the problem by redefining visibility based on the perceived image of a target sign. New algorithms have been created during this study to utilise modern hardware and software technology in the simulation of human perceived image of object in both experiment and computer modelling. Unlike the traditional threshold of visual distance, visibility in the current study has been defined as a continuous function changing from clearly discemable to completely invisible. It allows the comparison of visibility under various conditions, not just limited to the threshold. Current experiment has revealed that different conditions may results in the same visual threshold but follow very different path on the way leading to the threshold. The new definition of visibility has made the quantification of visibility in the pre-threshold conditions possible. Such quantification can help to improve the performance of fire evacuation since most evacuees will experience the pre-threshold condition. With current measurement of visibility, all the influential factors such as colour, viewing angle etc. can be tested in experiment and simulated in numerical model.

Based on the newly introduced definition of visibility, a set of experiments have been carried output in a purposed built smoke tunnel. Digital camera images of various illuminated signs were taken under different illumination, colour and smoke conditions. Using an algorithm developed by the author in this study, the digital camera images were converted into simulated human perceived images. The visibility of a target sign is measured against the quality of its image acquired. Conclusions have been drawn by comparing visibility under different conditions. One of them is that signs illuminated with red and green lights have the similar visibility that is far better than that with blue light. It is the first time this seemingly obvious conclusion has been quantified.

In the simulation of visibility in participating media, the author has introduced an algorithm that combines irradiance catching in 3D space with Monte Carlo ray tracing. It can calculate the distribution of scattered radiation with good accuracy without the high cost typically related to zonal method and the limitations in discrete ordinate method. The algorithm has been combined with a two pass solution method to produce high resolution images without introducing excessive number of rays from the light source. The convergence of the iterative solution procedure implemented has been theoretically proven. The accuracy of the model is demonstrated by comparing with the analytical solution of a point radiant source in 3D space. Further validation of the simulation model has been carried out by comparing the model prediction with the data from the smoke tunnel experiments.

The output of the simulation model has been presented in the form of an innovative floor map of visibility (FMV). It helps the fire safety designer to identify regions of poor visibility in a glance and will prove to be a very useful tool in performance based fire safety design.


Zhang, Q. (2010). Image based analysis of visibility in smoke laden environments. (Thesis). University of Hull. Retrieved from

Thesis Type Thesis
Deposit Date Sep 17, 2012
Publicly Available Date Feb 22, 2023
Keywords Engineering
Public URL
Additional Information Department of Engineering, The University of Hull
Award Date Mar 1, 2010


Thesis (23.8 Mb)

Copyright Statement
© 2010 Zhang, Qihui. All rights reserved. No part of this publication may be reproduced without the written permission of the copyright holder.

Downloadable Citations