Abstract: We provide a thorough experimental evaluation of several state-of-the-art textural features on four representative and extensive image data\-bases. Each of the experimental textural data\-bases ALOT, Bonn BTF, UEA Uncalibra\-ted, and KTH-TIPS2 aims at specific part of realistic acquisition conditions of surface materials represented as multispectral textures. The extensive experimental evaluation proves the outstanding reliable and robust performance of efficient Markovian textural features analytically derived from a wide-sense Markov random field causal model. These features systematically outperform leading Gabor, Opponent Gabor, LBP, and LBP-HF alternatives. Moreover, they even allow successful recognition of arbitrary illuminated samples using a single training image per material. Our features are successfully applied also for the recent most advanced textural representation in the form of 7-dimensional Bidirectional Texture Function (BTF).

Abstract: Understanding of human perception of textured materials is one of the most difficult tasks of computer vision. In this paper we designed a strictly controlled psychophysical experiment with stimuli featuring different combinations of shape, illumination directions and surface texture. Appearance of five tested materials was represented by measured view and illumination dependent Bidirectional Texture Functions. Twelve subjects participated in visual search task - to find which of four identical three dimensional objects had its texture modified. We investigated the effect of shape and texture on subjects’ attention. We are not looking at low level salience, as the task is to make a high level quality judgment. Our results revealed several interesting aspects of human perception of different textured materials and, surface shapes.

Abstract: A visual appearance of natural materials significantly depends on acquisition circumstances, particularly illumination conditions and viewpoint position, whose variations cause difficulties in the analysis of real scenes. We address this issue with novel texture features, based on fast estimates of Markovian statistics, that are simultaneously rotation and illumination invariant. The proposed features are invariant to in-plane material rotation and illumination spectrum (colour invariance), they are robust to local intensity changes (cast shadows) and illumination direction. No knowledge of illumination conditions is required and recognition is possible from a single training image per material. The material recognition is tested on the currently most realistic visual representation - Bidirectional Texture Function (BTF), using CUReT and ALOT texture datasets with more than 250 natural materials. Our proposed features significantly outperform leading alternatives including Local Binary Patterns (LBP, LBP-HF) and texton MR8 methods.

Abstract: We introduce a new standalone widely applicable software library for feature selection (also known as attribute or variable selection), capable of reducing problem dimensionality to maximize the accuracy of data models, performance of automatic decision rules as well as to reduce data acquisition cost. The library can be exploited by users in research as well as in industry. Less experienced users can experiment with different provided methods and their application to real-life problems, experts can implement their own criteria or search schemes taking advantage of the toolbox framework. In this paper we first provide a concise survey of a variety of existing feature selection approaches. Then we focus on a selected group of methods of good general performance as well as on tools surpassing the limits of existing libraries. We build a feature selection framework around them and design an object-based generic software library. We describe the key design points and properties of the library. The library is published at http://fst.utia.cz.

Abstract: A visual appearance of natural materials fundamentally depends on illumination conditions, which significantly complicates a real scene analysis. We propose textural features based on fast Markovian statistics, which are simultaneously invariant to illumination colour and robust to illumination direction. No knowledge of illumination conditions is required and a recognition is possible from a single training image per material. Material recognition is tested on the currently most realistic visual representation - Bidirectional Texture Function (BTF), using the Amsterdam Library of Textures (ALOT), which contains 250 natural materials acquired in different illumination conditions. Our proposed features significantly outperform several leading alternatives including Local Binary Patterns (LBP, LBP-HF) and Gabor features.

Abstract: A content-based tile retrieval system based on the underlying multispectral Markov random field representation is introduced. Single tiles are represented by our approved textural features derived from especially efficient Markovian statistics and supplemented with Local Binary Patterns (LBP) features representing occasional tile inhomogeneities. Markovian features are on top of that also invariant to illumination colour and robust to illumination direction variations, therefore an arbitrary illuminated tiles do not negatively influence the retrieval result. The presented computer-aided tile consulting system retrieves tiles from recent tile production digital catalogues, so that the retrieved tiles have as similar pattern and/or colours to a query tile as possible. The system is verified on a large commercial tile database in a psychovisual experiment.

Abstract: Delivering digitally a realistic appearance of materials is one of the most difficult tasks of computer vision. Accurate representation of surface texture can be obtained by means of view- and illumination- dependent textures. However, this kind of appearance representation produces massive datasets so their compression is inevitable. For optimal visual performance of compression methods, their parameters should be tuned to a specific material. We propose a set of statistical descriptors motivated by textural features, and psychophysically evaluate their performance on three subtle artificial degradations of textures appearance. We tested five types of descriptors on five different textures. We found that descriptors based on a two-dimensional causal auto-regressive model, have the highest correlation with the psychophysical results, and so can be used for automatic detection of subtle changes in rendered textured surfaces in accordance with human vision.

Abstract: We propose an illumination invariant and rotation insensitive texture representation based on a Markovian textural model. A texture is aligned with its dominant orientation and textural features are derived from fast analytical estimates of Markovian statistics. We do not require any knowledge of illumination direction or spectrum. This makes our method suitable for computer analysis of real scenes, where appearance of materials depends on their orientation towards the illumination source. Our method is tested on the most realistic visual representation of natural materials - the bidirectional texture function (BTF), using data from the CUReT database, where it outperforms the alternative leading illumination invariant Local Binary Patterns (LBP) and texton MR8 methods, respectively.

Abstract: A novel illumination invariant unsupervised multispectral texture segmentation method with unknown number of classes is presented. Multispectral texture mosaics are locally represented by illumination invariants derived from four directional causal multispectral Markovian models recursively evaluated for each pixel. Resulted parametric space is segmented using a Gaussian mixture model based unsupervised segmenter. The segmentation algorithm starts with an over segmented initial estimation which is adaptively modified until the optimal number of homogeneous texture segments is reached. The performance of the presented method is extensively tested on the large illumination invariant benchmark from the Prague Segmentation Benchmark using 21 segmentation criteria and compares favourably with an alternative segmentation method.

Abstract: This paper describes method for analysis of the texture created by retinal nerve fibers (RNF) via Markov Random Fields. The Causal Autoregressive Random (CAR) model is used to create a feature vector describing the changes in texture due to losses in RNF layer. It is shown that features based on CAR model can be used for discrimination between healthy and glaucomatous tissue using simple linear classifier. The classification error is slightly below 4% for the tested dataset.

Abstract: We propose textural features, which are invariant to illumination spectrum and extremely robust to illumination direction. They require only a single training image per texture and no knowledge of illumination direction or spectrum. Hence, these features are suitable for content-based image retrieval (CBIR) of realistic scenes with colour textured objects and variable illumination. The illumination invariants are derived from Markov random field based texture representations. Our illumination invariant features are favourably compared with frequented features in this area - the Local Binary Patterns, steerable pyramid and Gabor textural features, respectively. The superiority of our new invariant features is demonstrated in the illumination invariant recognition of the most advanced representation for realistic real-world materials - Bidirectional Texture Function (BTF) textures.

Abstract: Content-based image retrieval (CBIR) systems, target database images using feature similarities with respect to the query. We introduce fast and robust image retrieval measures which utilise novel illumination invariant features extracted from three different Markov random field (MRF) based texture representations. These measures allow to retrieve images with similar scenes comprising colour textured objects viewed with different illumination brightness or spectrum.
   The proposed illumination insensitive measures are compared favourably with the most frequently used features like the Local Binary Patterns, steerable pyramid and Gabor textural features, respectively. The superiority of these new illumination invariant measures and their robustness to added noise are is empirically demonstrated in the illumination invariant recognition of textures from the Outex database.

Abstract: Content-based image retrieval (CBIR) systems target database images using feature similarities with respect to the query. Our CBIR demonstration utilises novel illumination invariant features, which are extracted from Markov random field (MRF) based texture representations. These features allow retrieving images with similar scenes comprising colour-textured objects viewed with different illumination brightness or spectrum. The illumination invariant retrieval is verified on textures from the Outex database.

Abstract: Two fast illumination invariant image retrieval methods for scenes comprising textured objects with variable illumination are introduced. Both methods are based on texture gradient modelled by efficient set of random field models. We developed the illumination insensitive measures for textured images representation and compared them favorably with steerable pyramid and Gabor features in the illumination invariant texture recognition.

Abstract: This paper surveys the current best texture representations and studies their application for a texture similarity measure development. A simple experiment that evaluates texture similarity is proposed and the performance of several most advanced texture features is verified on it. In order to eliminate the influence of spectral information monospectral textures are considered in this study only. The paper suggests the L1 norm with either Markovian or Gabor features as the best texture similarity measure.