Visual systems must be matched (via evolution and learning over the lifespan) to the natural tasks organisms perform to survive and reproduce. Thus, it is of fundamental importance to analyze visual systems with respect to natural tasks and to the statistical properties of natural stimuli relevant to performing those tasks. In our lab we call this "natural systems analysis." This novel approach to vision science is composed of several steps: (1) identify natural tasks, (2) measure the natural scene statistics relevant for those tasks, (3) determine how to optimally use those statistics to perform the tasks, given appropriate biological constraints, and (4) use the first three steps to formulate principled hypotheses which are tested and refined in behavioral or physiological experiments. Using a unique suite of measurement devices, computational tools, and psychophysical paradigms developed in our laboratory, we propose to tackle (within the framework of natural systems analysis) four fundamental and interrelated classes of visual tasks: (1) image interpolation, where the goal is to estimate missing retinal image information due to occluding surfaces, normal cone sampling, or abnormal cone loss, (2) estimation of object (surface) boundary locations and which side of the boundary is in the foreground (nearer), (3) defocus estimation, where the goal is to estimate the magnitude and sign of image defocus in local regions of the retinal image, and (4) contrast and sharpness estimation in the visual periphery, where the ganglion cells severely filter and down sample the retinal image. The overall aim of tackling tasks (1) and (2) is to build an integrated picture of how the visual system identifies occlusion regions and interpolates behind them under natural conditions. The overall aim of tackling tasks (3) and (4) is to understand how the visual system estimates defocus and contrast in the fovea and periphery, which are relevant to the first two tasks. Many of the proposed studies will be the first to precisely characterize the statistical constraints in natural images underlying the visual system's ability to perform these tasks accurately. Many of the proposed studies will also be the first to measure performance in these fundamental tasks using natural stimuli. Promising pilot data has been obtained in many of the proposed studies.
Ultimate goals of vision science are to understand vision in the real world and to mitigate the effects of visual dysfunction on real-world performance. The proposed studies based on measuring the task-relevant statistics of natural images, and determining how best to use those statistical properties in natural tasks, will provide rigorous steps toward those ultimate goals and may produce useful image-processing applications.
|Burge, Johannes; McCann, Brian C; Geisler, Wilson S (2016) Estimating 3D tilt from local image cues in natural scenes. J Vis 16:2|
|Paulun, Vivian C; SchÃ¼tz, Alexander C; Michel, Melchi M et al. (2015) Visual search under scotopic lighting conditions. Vision Res 113:155-68|
|Burge, Johannes; Geisler, Wilson S (2015) Optimal speed estimation in natural image movies predicts human performance. Nat Commun 6:7900|
|Sebastian, Stephen; Burge, Johannes; Geisler, Wilson S (2015) Defocus blur discrimination in natural images with natural optics. J Vis 15:16|
|Bradley, Chris; Abrams, Jared; Geisler, Wilson S (2014) Retina-V1 model of detectability across the visual field. J Vis 14:|
|Morgenstern, Yaniv; Geisler, Wilson S; Murray, Richard F (2014) Human vision is attuned to the diffuseness of natural light. J Vis 14:|
|Burge, Johannes; Geisler, Wilson S (2014) Optimal disparity estimation in natural stereo images. J Vis 14:|
|Michel, Melchi M; Chen, Yuzhi; Geisler, Wilson S et al. (2013) An illusion predicted by V1 population activity implicates cortical topography in shape perception. Nat Neurosci 16:1477-83|
|D'Antona, Anthony D; Perry, Jeffrey S; Geisler, Wilson S (2013) Humans make efficient use of natural image statistics when performing spatial interpolation. J Vis 13:|
|Geisler, Wilson S; Perry, Jeffrey S (2011) Statistics for optimal point prediction in natural images. J Vis 11:14|
Showing the most recent 10 out of 25 publications