Scale selection

Tony Lindeberg

Computer Vision: A Reference Guide, (K. Ikeuchi, Editor), Springer, pages 701-713, 2014.

Digitally published with doi:10.1007/978-0-387-31439-6_242 in June 2014.


The notion of scale selection refers to methods for estimating characteristic scales in image data and for automatically determining locally appropriate scales in a scale-space representation, so as to adapt subsequent processing to the local image structure and compute scale invariant image features and image descriptors.

An essential aspect of the approach is that it allows for a bottom-up determination of inherent scales of features and objects without first recognizing them or delimiting alternatively segmenting them from their surrounding.

Scale selection methods have also been developed from other viewpoints of performing noise suppression and exploring top-down information.

PDF: (PDF 2.3 Mb)

On-line version: (At the official site of the encyclopedia)

Background and related material: (Main paper on feature detection with automatic scale selection including the scale-invariant Laplacian and determinant of the Hessian interest point detectors) (Main paper on edge detection and ridge detection with automatic scale selection) (Scale selection properties of generalized scale-space interest points) (Image matching using generalized scale-space interest points) (Tutorial on principles for automatic scale selection) (Feature tracking with automatic scale selection) (Integration of scale-invariant blob detectors with shape-from-texture and shape-from-disparity gradients) (Integration of scale-invariant corner detection with segmentation and classification of edges) (More general theory for scale covariant, affine covariant and Galilean covariant receptive fields with relations to receptive fields in biological vision) (More general theory for computing invariant visual representations under scaling transformations, local affine image deformations, local Galilean transformations and local multiplicative illumination transformations)

Responsible for this page: Tony Lindeberg