Improving Perceived Image Quality for Automotive Applications using Sub-pixel Rendering

Sammanfattning: The worldwide automotive market is constantly evolving due to digitalization.Vehicles manufacturers introduce more and more digitalized human-machinesinterfaces by adding screens in cars. Nowadays a majority of cars have twoscreens: one used for navigation and another one to display technical parameterssuch as speed or gas consumption. When the same image has to be displayed onboth screens it is not possible to generate two images adapted for both screensdue to hardware limitations. This implies that only one image adapted for aspecific screen is generated; to be displayed on the other screen this image hasto be rescaled. This resizing is often source of image distortion and reduceduser perceived image quality.The aim of this Master Thesis is to use sub-pixel rendering in resizing algorithmsto improve perceived quality. Sub-pixel rendering takes the human visual systemand the screen properties into account to improve image perceived quality afterresizing. This thesis adapts the sub-pixel rendering for resizing and comparesdi↵erent existing upscaling algorithms with the sub-pixel rendering upscalingalgorithms in terms of perceived quality. The perceived image quality afterresizing is assessed both by a referenceless image quality algorithm and a samplegroupsubjective evaluation composed of automotive engineers and designers.The evaluation highlights an improvement of perceived quality when the subpixelrendering algorithms is used for resizing.