Locally Rotation, Contrast, and Scale Invariant Descriptors for Texture Analysis

Abstract

Textures within real images vary in brightness, contrast, scale and skew as imaging conditions change. To enable recognition of textures in real images, it is necessary to employ a similarity measure which is invariant to these properties. Furthermore, since textures often appear on undulating surfaces, such invariances must necessarily be local rather than global. Despite these requirements, it is only relatively recently that texture recognition algorithms with local scale and affine invariance properties have begun to be reported. Typically, they comprise detecting feature points followed by geometric normalization prior to description. We describe a method based on invariant combinations of linear filters. Unlike previous methods, we introduce a novel family of filters, which provide scale invariance, resulting in a texture description invariant to local changes in orientation, contrast and scale and robust to local skew. Significantly, the family of filters enable local scale invariants to be defined without using a scale selection principle or a large number of filters. A texture discrimination method based on the A2 similarity measure applied to histograms derived from our filter responses outperforms existing methods for retrieval and classification results for both the Brodatz textures and the UIUC database, which has been designed to require local invariance.

On the Use of Low-Pass Filters for Image Processing with Inverse Laplacian Models

“…One way of summarizing filter properties for local feature definition was presented in [16]. There are three desired …”

“…21 and 22). In contrast, the Mellor-Brady filter satisfies all three conditions suggested for any α < −1, as demonstrated in [16]. In order to mimic the TIE filter, the Mellor-Brady filter parameters are set to α = β = 0.25 (Sect.…”

Shape-based Invariant Texture Indexing

“…Many of these approaches enable translation invariance (by using over-complete representations), rotation and scale invariance, by using effective filter designs, see e.g. Chen and Kundu (1994), Do and Vetterli (2002), Pun and Lee (2003), Mellor et al (2008). Some contrast invariance can also be achieved by normalizing responses to filters.…”

“…Recently, several works proposed to use individually normalized local features in order to represent textures while being locally invariant to geometric or radiometric transforms, see Lazebnik et al (2005), Zhang et al (2007), Varma and Garg (2007), Mellor et al (2008). In Lazebnik et al (2005) and Zhang et al (2007), a set of interest local affine regions are selected to build a sparse representation of textures relying on affine invariant descriptors.…”

“…1, where two different samples of the same texture class (plaid) from the UIUC database (Lazebnik et al 2005) are displayed. Several recent approaches to the analysis of such textures rely on the extraction of local features that are individually invariant to some geometric transforms, such as similarity or affine transforms (Lazebnik et al 2003;Mellor et al 2008). In contrast with previous works dealing with invariant 3D texture analysis, such locally invariant methods do not need any learning of the deformations (Varma and Zisserman 2002;Leung and Malik 2001) or explicit modeling (Wu and Chantler 2003) of the 3D surfaces.…”