Abstract
The visual system summarizes complex scenes to extract meaningful features (Barlow, 1959; Marr 1976) by using image primitives (edges, bars), encoded physiologically by specific configuration of receptive fields (Hubel & Wiesel, 1962). This work follows a pattern-filtering approach, based on the principle of most efficient information coding under real-world physical limitations (Punzi & Del Viva VSS-2006; Del Viva & Punzi VSS-2008). The model, applied to black and white images, predicts from very general principles the structure of visual filters that closely resemble well-known receptive fields, and identifies salient features, such as edges and lines, providing highly compressed “primal sketches” of visual scenes Here we perform a psychophysical study of the effectiveness of the sketches provided by this pattern-filtering model in allowing human observers to discriminate between pairs of similar natural images. As a control, we compare results with alternative sketches with the same information content, derived from a similar procedure, but not keeping into account the needs for optimal usage of computing resources. The performance was measured by the task of identifying natural images corresponding to briefly presented sketches (<50 ms.), with a 2AFC procedure. Our results show that performance obtained with sketches provided by our model is as good as that obtained from fully detailed original images, while the alternative sketches of equivalent information content are much less effective. These results provide support for the correctness of the model in predicting the salient features that human subjects use to identify visual scenes, supporting the idea that computing power limitations are a crucial factor in determining the way we perceive the world.
Supported by an Italian University and Research Ministry Grant (PRIN 2007).