Viewpoint Invariant Texture Description Using Fractal Analysis

Abstract

Image texture provides a rich visual description of the surfaces in the scene. Many texture signatures based on various statistical descriptions and various local measurements have been developed. Existing signatures, in general, are not invariant to 3D geometric transformations, which is a serious limitation for many applications. In this paper we introduce a new texture signature, called the multifractal spectrum (MFS). The MFS is invariant under the bi-Lipschitz map, which includes view-point changes and non-rigid deformations of the texture surface, as well as local affine illumination changes. It provides an efficient framework combining global spatial invariance and local robust measurements. Intuitively, the MFS could be viewed as a "better histogram" with greater robustness to various environmental changes and the advantage of capturing some geometrical distribution information encoded in the texture. Experiments demonstrate that the MFS codes the essential structure of textures with very low dimension, and thus represents an useful tool for texture classification.

Dynamic texture classification using dynamic fractal analysis

“…For example, fractal dimension was first proposed by Pentland [17] for texture analysis, and later on the similar concept is applied on static texture classification by replacing fractal dimension using more advanced multi-fractal analysis [25,29,30]. …”

“…Interested readers are referred to [9,16,29] for more details. Fractal analysis is built on the concept of fractal dimension which was first proposed by Mandelbrot [16] as the measurement of power law existing in many natural phenomena.…”

“…Then, all pixels are partitioned into different sets E [αi,αi+1) , according to their local density values. In [29], the partition is implemented by a "hard" scheme, that is, a pixel x is assigned to…”

LBP-and-ScatNet-based combined features for efficient texture classification

“…Biologically Inspired Filtering (BF) [27] imitates the human retina mechanism 35 to extract more detail information of a given image when being used as a 36 preprocessing step. It enhances performance of different features in terms of 37 discriminative power for texture classification, including CLBP.…”

“…Parameters for the BF preprocessing technique are: σ 1 = 1.25, σ 2 = 6, = 18 0.15 as recommended in [27]. …”