EXAM 5 Study Guide VISION IV Perceiving Recognizing Objects What is middle vision Loosely defined stage of visual processing After basic features have been extracted from the image early vision up to Before object recognition and scene understanding high level vision Goal to organize elements of a visual scene into groups that we can then including striate cortex recognize as objects Why do you think the Gestalt school of thought was instrumental in identifying how we perceive images To answer this you need to know what Gestalt means Gestalt An organized whole that is perceived as more than the sum of its parts 2 bits of an edge make it easier to perceive a 3rd line between them even if it isn t really there lines of similar orientation go together If the set of lines forms a lose shape then they support each other even more Good continuation gestalt grouping rule 2 lines will tend to group together if they seem to lie on the same contour Know the gestalt grouping principles good continuation occlusion texture segmentation similarity proximity connectedness common region how do those latter 2 relate to proximity what trumps what parallelism symmetry Also know dynamic occlusion Gestalt grouping rules a set of rules describing which elements in an image will appear to group together the original list was assembled by members of the Gestalt school of thought Gestalt school of thought the whole perceptual world is more than the sum of its sensory parts Good continuation Gestalt grouping rule 2 lines will tend to group together if they seem to lie on the same contour Occlusion if an edge seems to stop the visual system guesses that it is stopping because something else is in the way occluding it Texture segmentation carving an image into regions of common texture properties Edge finding isn t helpful Region of larger polygons is separated from the rest of the image Texture grouping can be based on similar color size orientation Similarity chunks that are similar to each other are grouped together Proximity items near each other are grouped together Connectedness if two items are connected they probably belong together Common Region if two features appear to be part of the same common region they will group together Parallelism regions with parallel contours are more likely seen as figure Symmetry symmetrical region more likely the figure Dynamic occlusion when visual regions are discontinuous not only in space but over time Color trumps good continuation as an organizing principle All things are not equal How does color fit in to the gestalt grouping principles Another powerful similarity type is value color The two are grouped together because value is part of color but can act independently black and white images The strongest gestalt between two items happens when they overlap Two colors are used in the example to the left to show the overlaps better When the two items are the same color they seem to form a new more complex shape The new shape seems flat When the items are different colors the overlap produces the illusion of a shallow space The overlapped items form a strong group regardless of color What is an ambiguous figure Ambiguous figure when 2 cognitive demons shout equally loudly This is the exception that provides the rule the rule being that usually there is consensus among the committees What is an accidental viewpoint Accidental viewpoint a viewing position that produces some regularity in the visual image that is not present in the world like the sides of 2 independent objects lining up perfectly What is figure and ground and what principles do we use to decide which is which in a visual scene What is evidence that maybe object recognition precedes establishment of figure ground Figure and ground figure ground assignment is the process of determining that some regions of an image belong to a forehand object figure others are part of the background ground Green shapes on blue background First assume that the figure vase on the ground white The surprise we register when the Rubin vase flips to a pair of faces reminds us of the perceptual stability that we usually take for granted Principles Surroundedness if 1 region is entirely surrounded by the other it is likely the figure Size smaller region is likely the figure Symmetry symmetrical region more likely the figure Parallelism regions with parallel contours are more likely seen as figure Relative motion how surface details more relative to an edge Synchrony elements that change at the same time group together External edges shading of objects tell us about figure and ground External edges the set of surface points whose sight lines are tangent to the surface We perceive the thing with the external edge as being closer to us What is the global superiority effect Properties of the whole object take precedence over the properties of parts of the objects That is what middle vision does the 1st goal is to carve the retinal image into large scale objects Why do we probably not used templates to recognize objects We would need way too many templates to match pixel to pixel or feature to feature to something in memory What are geons What is recognition by components and what are some criticisms of this approach Geons geometric ions Recognition by components objects are recognized by the identities and relationships of component parts View point invariant so we only need to store 1 template in memory Criticism Are geons really adequate for object recognition Think cigar box vs book is there a difference in the geons needed What about types of cars What about faces This will be a special case Some things like letter recognition are view point dependent What is the pandemonium model and what concepts of visual processing does it illustrate Pandemonium model early attempt to model a complex cognitive process like letter recognition using computer stimulation the model became famous for two reasons 1 Scientists learned more about the primate brain it became clear that Pandemonium s structure mimics the brain s structure in some important ways 2 Selfridge described the model in a highly entertaining way Concepts of visual processing Modularity Feature demons lower vision primary visual cortex Cognitive demons middle vision Decision demon high level vision object recognition inferior temporal cortex Each only know what the previous one told it Perception by committee Multiple committees groups of neurons each looking for a
View Full Document
Unlocking...