Abstract | ||
---|---|---|
Certain simple images are known to trigger a percept of trans- parency: the input image I is perceived as the sum of two images I(x;y) = I1(x;y)+ I2(x;y). This percept is puzzling. First, why dowechoosethe\morecomplicated"descriptionwithtwoimages ratherthanthe\simpler"explanation I(x;y)= I1(x;y)+0? Sec- ond,giventheinflnitenumberofwaystoexpress I asasumoftwo images, how do we compute the \best" decomposition ? Herewesuggestthattransparencyistherationalperceptofasys- temthatisadaptedtothestatisticsofnaturalscenes. Wepresent a probabilistic model of images based on the qualitative statistics of derivative fllters and \corner detectors" in natural scenes and usethismodeltoflndthemostprobabledecompositionofanovel image. The optimization is performed using loopy belief propa- gation. We show that our model computes perceptually \correct" decompositions on synthetic images and discuss its application to real images. |
Year | Venue | Keywords |
---|---|---|
2002 | NIPS | probabilistic model |
Field | DocType | Citations |
Computer vision,Transparency (graphic),Computer science,Statistical model,Artificial intelligence,Real image,Statistics,Machine learning,Belief propagation,Percept | Conference | 32 |
PageRank | References | Authors |
3.62 | 6 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Anat Levin | 1 | 3578 | 212.90 |
Assaf Zomet | 2 | 725 | 57.51 |
Yair Weiss | 3 | 10240 | 834.60 |