Importance-Driven Focus of Attention and Meister Eduard Gröller 1 1 Vienna University of Technology, Austria 2 University of Girona, Spain 3 University of Bergen, Norway Ivan Viola 1,3, Miquel Feixas 2, Mateu Sbert 2
Ivan Viola 1 Goal Input: known and classified volumetric data High level request: show me object X Output: guided navigation to object X
Ivan Viola 2 Focusing Considerations Characteristic view Emphasis of focus object Guided navigation between characteristic views
Ivan Viola 3 Framework
Ivan Viola 4 Characteristic Views Overview All objects are visible Visibility of objects is balanced Characteristic view of focus object High visibility for focus object If possible other objects also visible
Ivan Viola 5 Characteristic View Estimation o 2 o 3 o 1 importance distribution v 1 v 2 v 3 o 1 o 2 o 3 visibility estimation image-space weight p(v 1 ) n ) p(o 1 |v 1 ) p(o m |v n ) p(o 1 ) m )... I(v i,O) = p(o j |v i ) log j m p(o j |v i ) p(o j )... information-theoretic framework for optimal viewpoint estimation o 1 o 2 o 3 object selection by user v o 1 o 2 o 3 object-space distance weight o 1 o 2 o 3 v viewpoint transformation v o 1 o 2 o 3 cut-away and level of ghosting o 3 o 1 o 2 o 3 focus discrimination characteristic viewpoint estimation interactive focus of attention o 1 o 2 o 3 up-vector information characteristic viewpoint estimation view rating
Ivan Viola 6 View rating v1v1 v2v2 v3v3 v4v4 v5v5 v6v6 v7v7 v8v8 o1o1 o2o2 o3o3 For every view For every object
Ivan Viola 7 View Rating Visibility High Low Location in image In image center Outside center Distance to the viewer Object close to the viewer Far from the viewer
Ivan Viola 8 Visibility Computation o 0 = object 0 o 1 = object 1 r = ray r 0 = sub-ray 0 r 1 = sub-ray 1 r 2 = sub-ray 2
Ivan Viola 9 Visibility Computation
Ivan Viola 10 View Rating Weights object-space distance weight image-space weight
Ivan Viola 11 Characteristic Viewpoint Estimation o 2 o 3 o 1 importance distribution v 1 v 2 v 3 o 1 o 2 o 3 visibility estimation image-space weight p(v 1 ) n ) p(o 1 |v 1 ) p(o m |v n ) p(o 1 ) m )... I(v i,O) = p(o j |v i ) log j m p(o j |v i ) p(o j )... information-theoretic framework for optimal viewpoint estimation o 1 o 2 o 3 object selection by user v o 1 o 2 o 3 object-space distance weight o 1 o 2 o 3 v viewpoint transformation v o 1 o 2 o 3 cut-away and level of ghosting o 3 o 1 o 2 o 3 focus discrimination characteristic viewpoint estimation interactive focus of attention o 1 o 2 o 3 up-vector information characteristic viewpoint estimation view rating characteristic views
Ivan Viola 12 Characteristic Views Overview All objects are visible Visibility of objects is balanced Characteristic view of focus object High view rating (visibility) for focus object If possible other objects also visible
Ivan Viola 13 Obtaining Characteristic Views Sets of views and objects are random variables Views V=(v 1, v 2, v 3,..., v n ) Objects O=(o 1, o 2, o 3,..., o m ) View rating (visibility, weights) Information channel between VO Conditional probability p(o j |v i ) Mutual information between V and O expresses degree of dependance
Ivan Viola 14 Obtaining Characteristic Views Viewpoint mutual information is dependance between v i and O High values = high dependance Small number of objects Low average visibility Low values = low dependance Maximum objects visible Object visibility is balanced Minimal VMI determines the best view
Ivan Viola 15 Probability Transition Matrix p(v 1 ) p(v 2 ) p(v 3 )... p(v n ) p(o 1 )p(o 2 )p(o 3 )p(o m )... p(o 1 |v 1 )p(o 2 |v 1 ) p(o 1 |v 2 )... p(o m |v n )... p(o m |v 1 ) p(o 1 |v n ) probability of the viewpointmarginal probability of the objectview rating of object o j from viewpoint v i
Ivan Viola 16 Viewpoint Mutual Information Degree of correlation v j O p(v 1 ) p(v 2 ) p(v 3 )... p(v n ) p(o 1 )p(o 2 )p(o 3 )p(o m )... p(o 1 |v 1 )p(o 2 |v 1 ) p(o 1 |v 2 )... p(o m |v n )... p(o m |v 1 ) p(o 1 |v n )
Ivan Viola 17 Characteristic Views Overview All objects are visible Visibility of objects is balanced Characteristic view at focus object High view rating for focus object If possible other objects also visible
Ivan Viola 18 Incorporating Importance importance distribution o1o1 o2o2 o3o3
Ivan Viola 19 Resulting Characteristic Viewpoints
Ivan Viola 20 o 1 Interactive Focus of Attention
Ivan Viola 21 Emphasis of Focus Object Levels of sparseness representation 0 importance max dense
Ivan Viola 22 Emphasis of Focus Object Cut-aways to unveil internal features Labeling to add textual information vessels intestinekidneys
Ivan Viola 23 Guided Navigation Between Objects Decreasing importance of Object X De-emphasis of Object X Change to overview Increasing importance of Object Y Emphasis of Object Y Change to characteristic view of Y
Ivan Viola 24 Refocusing o 1 o 2 o 3 v c v 1 v 2 o 1 o 2 Characteristic view 1 Characteristic view 2 Overview
Ivan Viola 25 Example - Stagbeetle Focus view 1 Focus view 2 Overview
Ivan Viola 26 Smooth Transition to Focus View o 1 o 2 o 3
Ivan Viola 27 Example - Human Hand Any Questions?
Ivan Viola 28 Conclusions Focus of attention framework Characteristic view estimation Guided navigation Steered by changes in importance distribution Future Work Zooming to the focus Other smart visibility techniques Available soon as plugin in volumeshop.org
Ivan Viola 29 Thank you! The End
Ivan Viola 30 Viewpoint Entropy [Bordoloi et al. '05] Viewpoint Mutual Information Comparison to Viewpoint Entropy
Ivan Viola 31 Visibility Computation v1v1 v2v2 v3v3 v4v4 v5v5 v6v6 v7v7 v8v8 importance distribution o1o1 o2o2 o3o3 o1o1 o2o2 o3o3 For overview and all focus objects For every viewpoint For every object + background
Ivan Viola 32 Visibility Computation for Focus Object o 0 = object 0 o 1 = object 1 r = ray r 0 = sub-ray 0 r 1 = sub-ray 1 r 2 = sub-ray 2 0,r 2 ) r α (o 1,r 1 ) α (o α
Ivan Viola 33 Visibility Computation o 0 = object 0 o 1 = object 1 r = ray r 0 = sub-ray 0 r 1 = sub-ray 1 r 2 = sub-ray 2
Ivan Viola 34 Probability Transition Matrix p(v 1 ) p(v 2 ) p(v 3 )... p(v n ) p(o 1 )p(o 2 )p(o 3 )p(o m )... p(o 1 |v 1 )p(o 2 |v 1 ) p(o 1 |v 2 )... p(o m |v n )... p(o m |v 1 ) p(o 1 |v n ) p(v 1 ) p(v 2 ) p(v 3 )... p(v n ) p(o 1 )p(o 2 )p(o 3 )p(o m )... p(o 1 |v 1 )p(o 2 |v 1 ) p(o 1 |v 2 )... p(o m |v n )... p(o m |v 1 ) p(o 1 |v n ) p(v 1 ) p(v 2 ) p(v 3 )... p(v n ) p(o 1 )p(o 2 )p(o 3 )p(o m )... p(o 1 |v 1 )p(o 2 |v 1 ) p(o 1 |v 2 )... p(o m |v n )... p(o m |v 1 ) p(o 1 |v n ) p(v 1 ) p(v 2 ) p(v 3 )... p(v n ) p(o 1 )p(o 2 )p(o 3 )p(o m )... p(o 1 |v 1 )p(o 2 |v 1 ) p(o 1 |v 2 )... p(o m |v n )... p(o m |v 1 ) p(o 1 |v n ) p(v 1 ) p(v 2 ) p(v 3 )... p(v n ) p(o 1 )p(o 2 )p(o 3 )p(o m )... p(o 1 |v 1 )p(o 2 |v 1 ) p(o 1 |v 2 )... p(o m |v n )... p(o m |v 1 ) p(o 1 |v n ) active o 1 active o m... inactive