Large-scale consistensy in visual image elements is essential for figure-ground organization
Photo by Sam Moqadam on Unsplash
What is it about?
We developed a neuro-computational model for "border-ownership" signals that underlie figure-ground organization (perception of depth order). By reflecting the global consistency of image elements, it shows robust responses at an unprecedented level.
Why is it important?
The visual system performs remarkably well to perceive the depth order of separate areas in the surrounding enviroenment. In this, figure-ground organization based on pictorial cues plays an important role. To understand how figure-ground organization emerges through image signal processing, it is essential how the global configuration of the image is reflected. In the past, many neuro-computational models implemented algorithms to give a bias to convex shapes and were based on the geometriy of borderlines. However, in certain conditions, this approach is bound to fail. We argue that the long-range consistency of surface properties is reflected in the computational processes of border-ownership (or edge assignement) . Our model shows exteremely robust responses unprecedented by previous models. It is possible that a class of border-ownership-sensitive neurons that are also sensitive to contrast polarity (Zhou et al., 2000, J. Neurosci.) underlie this computation process.
The following have contributed to this page: Naoki Kogo