Chapter 5: Space and Form Form & Pattern Perception: Humans are second to none in processing visual form and pattern information. Our ability to see patterns and details is unparallel on earth, and yet is still not fully understood. In this chapter we will examine what we know about human form and pattern visual perception. What is a form? Form perception: the ability to segment the spatial attributes (size and shape) of single entity from the other entities and from background. Separating figure from ground. Form perception involves two subcomponents: 1) Detection: the ability to determine that an object is present in the visual environment. A lightness change or contour is perceived in the environment, cues to this may be the lightness change itself, or movement, or depth cues, or some combo. 2) Discrimination: the ability distinguish one object from another. This is a more refined process in which specific spatial attributes of the object are used not just to determine the presence of the object but to segregate different classes of items – faces vs. trees; squares from circles, etc. 3) Identification: Recognizing and labeling a particular item (grandma’s face). Defining Form Form as contour change: Luminance or reflectance level changes, digitized or quantified. Structuralist Approach 1) Structuralist approach: the final perceived form is combination of more elemental or primitive components. Visual system combines elemental components to produce forms. Structuralists used analytic introspection as primary methodology for uncovering the elemental components. Subjects trained to ignore more subjective aspects of stimulus and only report "objective sensations." Problems: 1) the catalogue of elemental components grew to unmanageable numbers (over 40,000) 2) little consistency was found across observers as to what an elemental component was. Lesson learned: Question of what are basic units of perception was valid, but method was wrong. More modern approach stresses physiological and experimental strategies. Analyzing spatial scale Begins with assumption that forms and patterns exist at different levels of detail or scale, and each of these scales must be processed by visual system. Scales: 1) largest scale: separation of form from background 2) intermediate scale: segregation of form from other similar forms 3) smallest scale: separating various details on single form. Hybrid images Images with contrasting information at high and low spatial scales. Used to “sensitize” visual system to particular spatial scale (women: low; man: high). Shows how visual system is “tuned” to different scales by meaning. If meaning at high level – biased to high; if meaning at low level – biased to low (Shyns & Oliva study, 1997). Multi-channel approach to form perception: Campbell & Robson At each spatial scale four parameters defining forms: 1) Spatial orientation: the angle of the form relative to the perceiver. 2) Contrast: the degree of intensity difference between the form and its background 3) Spatial frequency: the number of cycles from highest to lowest lightness per visual angle. 4) Spatial phase: placement of the frequency cycle relative to some reference point. Multi-channel approach to form perception: Campbell & Robson Form processing: Lens transfer functions Left axis: contrast Horizontal axis: spatial frequency Lesson: as spatial frequency increases greater contrast is needed to faithfully transfer image Contrast Sensitivity Functions: “window” of visibility for forms • Contrast Sensitivity Functions (CFS): a plot of the minimum contrast necessary for a grating of varying spatial frequency to be discriminated from a homogenous background. Humans are most sensitive (ie require least degree of contrast for detection) for spatial frequencies of around 5 cycles per degree of visual angle. Increased contrast is necessary for spatial frequencies higher or lower. Contrast Sensitivity Functions: “window” of visibility for forms Scotopic vs. photopic systems Humans vs. other species Infants vs. adults CSF: different channels • Overall human CSF is actually aggregate of separate, more narrowly tuned channels. Contrast normalization Local contrast homogeneity used as “cue” for normalizing of responses among visual channels processing that region. Thus, spatial frequencies and orientations tend to be perceived as similar for local feature. Normalization contributes to metamers – physically different signals that evoke same response in visual system and are thus perceived the same. Selective adaptation: altering perception by adapting or fatiguing certain spatial frequency channels Indicates that perception results from aggregation of responses from differently tuned frequency channels. Mid-spatial frequency (right gratings) result from combination of low response in high/low channels and larger response in mid channel. : Texture discrimination pop out effects Repetitive homogenous pattern establishes background texture, figure contrasts with background. Orientation, contrast or spatial frequency change defines border, texture change or figure. Gestalt principles of pattern/form recognition Gestalt approach: visual system segmented the visual stimuli into meaningful "wholes." Visual system had natural tendency toward "Pragnanz" or simple organized perception, while visual stimuli had "forces of attraction" which organized scenes.