Quantitative prediction of figure-ground assignment with difference of gathering Y Matsuda1, H Kaneko1, P Grove2 |
|---|
1Department of Information Processing, Tokyo Institute of Technology, Japan |
| Many factors such as area, symmetry and convexity affect “figure-ground assignment.” However, most such factors can predict the assignment only qualitatively. This study was conducted to quantify figure-ground assignment, introducing a new index: “difference of gathering.” Gathering, which is the index estimating the degree of denseness of components, is formularized with a concept of zero-mean normalized cross-correlation of adjacent pixels. A difference of gathering values is assumed to indicate a difference of strengths as a figure between competing components such as black and white. We hypothesized that the response ratio to regard a component having a larger gathering value as a figure increased as the difference of gathering values increased. In an experiment conducted to test the hypothesis, we used a square pattern comprising white and black dots. We manipulated the gathering values of the elements and those in specific areas. Participants responded which component or area was perceived to contain a figure. Results demonstrated that the ease in the figure-ground assignment increased linearly as the difference of gathering values increased. The results implied that a difference of gathering values for different elements in a pattern strongly affects the figure-ground assignment. |
| Up Home |
|---|