Institute of Transportation, MOTC

Leasing spaces for commercial advertisement has become an effective and trendy tactic for transportation operators to receive extra income. As more and more commercial advertisements occupy spaces in the public transportation environment, there is no regulation regarding the quantity control or restriction besides the current quality usually associated with social or censorship ethical acceptance.

This thesis regards those commercial advertisements as “unnecessary”, which provide no direct or useful information with respect to the transportation service itself. As such, these commercial advertisements appearing in transportation environments (most significantly in transit stations)can be viewed as “Visual Noise”, and therefore should be controlled in their amount or intensity.

A framework for the visual noise control analogical to the sound noise control was proposed. Based on the notation of Shannon Entropy to define the visual quantity (complexity), and to define a commercial advertisement as one visual event (a frame of advertisement) to cumulative events (cumulative effect of multiple advertisements) were presented to mimic the control of sound noise events.

Controlled Experiment Simulation technique was implemented in this thesis where various levels(actually 4 levels) and commercial advertisements were inserted into real video footages to simulate a passenger’s visual experience of a specific passage from a transit station entrance to boarding.

There are three major parts in this thesis：(1) The first is the preparation stage. This study uses Taipei

MRT Station as a demonstration case to collect station route and advertisements information to build the video footages for experiments; a 4X2 two-factor experimental design is adopted: four physical quantity levels of advertisements (low, medium, high, random group) and two advertising density level paths(with contrasting high density, low density).(2) Implementing the controlled experiment to observe and collect the response of selected participants to different control levels.(3) Finally, two types of ordered logistic regression models were constructed to relate the Visual Noise to the response of selected participants.

Model I is completely analogous to sound noise, and explores the relationship between "cumulative average visual noise" and "responses of the participants"; Model II is an adjustment model, where two variables "cumulative visual noise " and "advertising density" were needed instead of "cumulative average visual noise".

The research results are summarized as follows: (1) Based on the analogy of sound noise control, this study regards a visual advertisement event as a noise event by using the notation of Shannon Entropy, extension multiple advertisements as cumulative events, and finally associated visual complexity measurement to a human perception scale (visual noise). (2) The adjustment models constructed in this study (Model II) outperform the Model I that is fully analogous to sound noise control concept.(3) This study has demonstrated successfully to obtain the deployed density of commercial advertisements and the maximum total amount of advertisements under the pre-set threshold of user response.