The International Civil Aviation Organization (ICAO) is the regulatory body for Airports. ICAO standards dictate that luminaires used within an airport landing lighting pattern must have a color as defined within the 1931 color chart defined by the Commission Internationale De L'Eclairage (CIE). Currently, visual checks are used to ensure luminaires are operating at the right color within the pattern. That is, during an approach to an airport the pilot must
visually check each luminaire within the landing pattern. These visual tests are combined with on the spot meter reading tests. This method is not accurate and it is impossible to assess the color of every luminaire. This paper presents a novel, automated method for assessing the color of luminaires using low cost single chip CCD
video camera technology. Images taken from a camera placed within an aircraft cockpit during a normal approach to an airport are post-processed to determine the color of each luminaire in the pattern. In essence, the pixel coverage and total RGB level for each luminaire within the pattern must be extracted and tracked throughout the complete image sequence and an average RGB value used to predict the luminaire color. This prediction is based on a novel pixel model which was derived to determine the minimum pixel coverage required to accurately predict the color of an imaged luminaire. Analysis of how many pixels are required for color recognition and position within a CIE color chart is given and proved empirically. From the analysis it is found that, a minimum diameter of four pixels is required for color recognition of the major luminaires types within the airport landing pattern. The number of pixels required for classification of the color is then derived. This is important as the luminaries are far away when imaged and may cover only a few pixels since a large area must be viewed by the camera. The classification is then verified by laboratory based experiments with different luminaire sources. This paper shows that it is possible to accurately predict the color of luminaires using automated image analysis. Whilst this is not a new phenomenon the authors have shown that it is possible to simulate the operation of a single chip CCD imager and illustrate the minimum pixel coverage that is required to accurately represent a colored luminaire from a moving platform. In addition, the color assessment of airport landing lighting has not yet been tackled using photometrics and dynamic cameras. The principles outlined are generic and can therefore be applied to other areas of lighting research such as signal and street lighting. Future color work with respect to airport lighting will concentrate on more complex models for the luminaire movement across a mosaic filter. A current assumption used in this paper is that there are no gaps between the pixel patches, which is not the case. As such an update on the model will incorporate inter-pixel gaps.
|