A Stochastic Grammar of ImagesA Stochastic Grammar of Images is the first book to provide a foundational review and perspective of grammatical approaches to computer vision. In its quest for a stochastic and context sensitive grammar of images, it is intended to serve as a unified frame-work of representation, learning, and recognition for a large number of object categories. It starts out by addressing the historic trends in the area and overviewing the main concepts: such as the and-or graph, the parse graph, the dictionary and goes on to learning issues, semantic gaps between symbols and pixels, dataset for learning and algorithms. The proposal grammar presented integrates three prominent representations in the literature: stochastic grammars for composition, Markov (or graphical) models for contexts, and sparse coding with primitives (wavelets). It also combines the structure-based and appearance based methods in the vision literature. At the end of the review, three case studies are presented to illustrate the proposed grammar. A Stochastic Grammar of Images is an important contribution to the literature on structured statistical models in computer vision. |
Contents
Introduction | 1 |
Background | 19 |
Visual Vocabulary | 41 |
Relations and Configurations | 51 |
Parse Graph for Objects and Scenes | 59 |
Knowledge Representation with AndOr Graph | 70 |
Recursive TopDownBottomUp Algorithm | 81 |
Three Case Studies of Image Grammar | 87 |
Summary and Discussion | 98 |
Common terms and phrases
algorithm And-node And-Or graph And-Or tree And–Or arrows augmented bike bottom-up child nodes clock Closed lists composed Computer Vision connected curves CVPR dataset defined detection dictionary edges example Geman graph representation graphlets hierarchic histogram hobs IEEE Transactions image grammar image parsing image patch image primitives inference algorithm input image instances language large number layer leaf nodes Lotus Hill machine learning Markov random field matching minimax entropy natural images non-terminal nodes number of configurations object categories occluded open bonds Open lists Or-nodes overlap parameters parse graph pg parse tree particle pixels posterior probability primal sketch probability model production rules reconfigurable rectangles recursive represented reusable S. C. Zhu sampling scale SCFG semantic shown in Figure shows sketch graph sparse coding statistical stochastic grammar structures synthesized template terminal nodes textons texture tion top-down prediction top-down process unsupervised learning valid configurations visual vocabulary