Local binary patterns versus signal processing texture analysis: a study from a performance evaluation perspective

Abstract

Purpose-The purpose of this paper is to review and provide a detailed performance evaluation of a number of texture descriptors that analyse texture at micro-level such as local binary patterns (LBP) and a number of standard filtering techniques that sample the texture information using either a bank of isotropic filters or Gabor filters. Design/methodology/approach-The experimental tests were conducted on standard databases where the classification results are obtained for single and multiple texture orientations. The authors also analysed the performance of standard filtering texture analysis techniques (such as those based of LM and MR8 filter banks) when applied to the classification of texture images contained in standard Outex and Brodatz databases. Findings-The most important finding resulting from this study is that although the LBP/C and the multi-channel Gabor filtering techniques approach texture analysis from a different theoretical perspective, in this paper the authors have experimentally demonstrated that they share some common properties in regard to the way they sample the macro and micro properties of the texture. Practical implications-Texture is a fundamental property of digital images and the development of robust image descriptors plays a crucial role in the process of image segmentation and scene understanding. Originality/value-This paper contrast, from a practical and theoretical standpoint, the LBP and representative multi-channel texture analysis approaches and a substantial number of experimental results were provided to evaluate their performance when applied to standard texture databases.

Evaluation of noise robustness for local binary pattern descriptors in texture classification

“…The per-class accuracy for GF and the LBP descriptors is often similar even though the LBP descriptors are more alike among themselves (apart from MBP). This is in line with the findings reported in [38]. The per-class accuracy for GLCM differs from those of the LBP family and GF mainly in that GLCM has additional difficulties discriminating a number of classes.…”

Texture Description Through Histograms of Equivalent Patterns

“…In a recent work Ghita et al [25] have shown that the performances in texture classification offered by LBP/C and multi-channel Gabor filtering are comparable.…”

Rotation–Covariant Texture Learning Using Steerable Riesz Wavelets

“…The proposed method outperforms most filterand wavelet-based approaches using the Outex_TC_00010 test suite [1], [26], [27], [29], [39], [42], [43], [51], where only few methods based on LBPs achieve a performance above 98% [14], [19], [21], [24], [25], [30], [33], sometimes with manual parameter optimization. MR8, LM and S filterbanks were reported to obtain performances of 72.57%, 51.8% and 68.61% when combined with SVMs in [16] in 2012 using the same database. The performance obtained for P_000 and P_001 in Outex_TC_00012 is similar to the performance obtained with Outex_TC_00010, which suggests higher illumination invariance of the proposed approach when compared to methods based on LBPs in [14], [19], [21], [30], [39].…”

“…It has recently been used by several studies on texture recognition [14], [16], [19], [21], [24]- [27], [29], [30], [33], [39], [42], [51]. It consists of 24 texture classes with pronounced directional structures.…”