Hierarchical Perceptual Grouping for Object Recognition#
In January 2019 the book "Hierarchical Perceptual Grouping for Object Recognition - Theoretical Views and Gestalt-Law Applications" was published in its first edition by Springer-Verlag. The book is in English language.
This work offers new insights and proposes new methods to advance the field of pattern recognition, which will be of great benefit to students, researchers and engineers working in this field.
Dr.-Ing. Eckart Michaelsen, Fraunhofer IOSB Ettlingen, Department Object Recognition (OBJ)
Dr.-Ing. Jochen Meidow, Fraunhofer IOSB Ettlingen, Department Scene Analysis (SZA)
This rather unique book presents a unified approach to formulating Gestalt laws for perceptual grouping and aggregating nested hierarchies according to these laws. The book also describes the construction of such constructions from noisy images, which show both man-made objects and unpredictable disturbances. Each shape operation is presented in a separate, self-contained chapter including examples of use and a short literature review. These are then summarized in an algebraic final chapter, followed by chapters that link the methods to the data - i.e. primitive extraction from images, cooperation with machine-readable knowledge, and cooperation with machine learning.
Special features of the book
The work offers for the first time a unified approach to nested hierarchical perceptual groupings according to Gestalt laws
- provides an overview of all relevant design laws in a single document
- covers a reflection symmetry, frieze symmetry, rotational symmetry, parallelism and orthogonality, contour extension and grid symmetries
- describes the problem from all theoretical points of view, including syntactic, probabilistic and algebraic perspectives
- discusses important questions of practical application, such as primitive extraction and real-time search
- provides an appendix that describes a general compensation model with constraints.
From the book series
Advances in Computer Vision and Pattern Recognition