Image constraints on the perception of surfaces and materials

图像对表面和材料感知的限制

• 批准号: FT140100535
• 负责人: A/Prof Juno Kim
• 金额: $ 44.08万
• 依托单位: The University of New South Wales
• 依托单位国家: 澳大利亚
• 项目类别: ARC Future Fellowships
• 财政年份: 2015
• 资助国家: 澳大利亚
• 起止时间: 2015-03-01 至 2021-12-23
• 项目状态: 已结题
• 来源: https://dataportal.arc.gov.au/RGS/Web/Grants/FT140100535
• 关键词: Image; constraints; perception; surfaces; materials

Humans visually perceive objects that vary in 3D shape, colour, sheen, and opacity, but understanding how this information is recovered from the images seen remains a mystery. The brain must exploit constraints on the way images are formed through the interaction of light with different surface properties. This project aims to identify image constraints that the brain measures to infer this material diversity. This will advance understandings of computational processes underlying material perception. It will create efficient innovations in image-based pattern recognition algorithms used for commercial and everyday applications, such as quality sorting of fresh produce, and monitoring driver performance in wet weather to reduce errors.

人类在视觉上感知到不同的3D形状，颜色，光泽和不透明度的物体，但了解如何从所看到的图像中恢复这些信息仍然是一个谜。大脑必须利用通过具有不同表面特性的光的相互作用形成图像的方式的约束。该项目旨在识别大脑测量的图像限制，以推断这种材料的多样性。这将促进对材料感知背后的计算过程的理解。它将在用于商业和日常应用的基于图像的模式识别算法中创造有效的创新，例如新鲜农产品的质量分类，以及在潮湿天气中监控驾驶员的表现以减少错误。