The perception and recognition of surfaces and objects: Segmentation and completion

视觉处理中最困难的问题之一涉及当存在多个对象时恢复对象的结构。近的物体可以部分遮挡远的物体;远的物体可以部分伪装近的物体;物体可以在透明的条件下变得“混合”。拟议的工作将提供有关人类如何在用于医疗和安全诊断（如X射线）的图像中普遍存在的条件下提取对象结构的基础知识。 这项研究将为人类视觉系统如何提取物体和表面的结构提供重要的见解，这些结构可以为成像设备和自动检测系统的设计提供信息。

One of the most difficult problems in visual processing involves recovering the structure of objects when multiple objects are present. Near objects can partially occlude distant objects; distant objects can partially camouflage near objects; and objects can become 'mixed' in conditions of transparency. The proposed work will provide fundamental knowledge about how humans extract object structure in conditions that are pervasive in images used for both medical and security diagnosis (such as x-rays). The proposed research will provide critical insight into how the human visual system extracts the structure of objects and surfaces that can inform both the design of imaging devices and automated detection systems.