Task-, viewer- and degradation-specific image quality assessment

特定于任务、查看器和退化的图像质量评估

• 批准号: 561074-2020
• 负责人: Clark, James
• 金额: $ 1.46万
• 依托单位: McGill University
• 依托单位国家: 加拿大
• 项目类别: Alliance Grants
• 财政年份: 2021
• 资助国家: 加拿大
• 起止时间: 2021-01-01 至 2022-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=721866
• 关键词: Task; viewer; degradation; specific; image

For any display system with the ultimate goal of accurately communicating visual information to its viewer, image quality and visibility are key performance metrics that must be assured. This is imperative in most visual applications, but especially prominent in safety critical display systems, for instance such as those found in the automotive industry, where crucial information must be conveyed through the instrumentation clusters and dashboards of vehicles. The perceived image quality of a display panel can be degraded in many ways. For example, in an automobile, the navigation display can be degraded by light from outside the car shining onto it, causing bothersome reflections. Or the display could appear blurred due to vibration or a layer of strawberry jam added by your children. Some viewers may be affected more or less than other viewers by a particular type of degradation (e.g. due to age-related visual deficits or varying levels of driving experience), and the sensitivity to a particular degradation may depend on the task that a user is carrying out relative to the display (e.g. whether they are reading the speedometer or getting directions from the navigation display). This project aims to develop methods to measure perceptual image quality, that are task-, viewer- and degradation-specific. These techniques will be extensions of existing, more generic, methods and will employ modern machine learning approaches. The industrial partner, Faurecia IRYStec Inc., will use the results of the project in improving their Perceptual Display Platform, which provides viewer- and task-specific display enhancements for commercial display systems, particularly those found in the automotive industry.

对于任何最终目标是将视觉信息准确传达给观众的显示系统来说，图像质量和可见性是必须确保的关键性能指标。这在大多数视觉应用中是必不可少的，但在安全关键显示系统中尤其突出，例如在汽车工业中发现的那些，其中关键信息必须通过车辆的仪表集群和仪表板传达。显示面板的感知图像质量可能以许多方式降级。例如，在汽车中，来自车外的光照射到导航显示器上，会使导航显示器劣化，从而引起令人烦恼的反射。或者，由于振动或孩子添加的一层草莓酱，显示屏可能会显得模糊。一些观看者可能比其他观看者更多或更少地受到特定类型的降级（例如，由于年龄相关的视觉缺陷或驾驶经验的变化水平）的影响，并且对特定降级的敏感度可能取决于用户相对于显示器正在执行的任务（例如，他们是在阅读速度计还是从导航显示器获取方向）。该项目旨在开发测量感知图像质量的方法，这些方法是特定于任务，观众和降级的。这些技术将是现有的更通用的方法的扩展，并将采用现代机器学习方法。工业合作伙伴佛吉亚IRYStec公司，将利用该项目的成果来改进其感知显示平台，该平台为商用显示系统，特别是汽车行业的显示系统提供针对特定观众和特定任务的显示增强功能。