Assessing Changes in Texture Periodicity Due to Appearance Loss in Carpets: Gray Level Co-occurrence Analysis

Abstract

Texture may be defined with respect to the global properties of an image or to the repeating units that compose it. We have attempted to isolate the periodic components of texture in wool carpets for the purpose of assessing the effects of wear on carpet textures. Texture is defined as any recurring spatial pattern of sampled gray level intensities. We restrict attention to textural features that reflect the physical characteristics of the carpet tuft assemblies. Our test materials consist of four kinds of double ply wool carpets of differing textures divided into control, light, and heavy wear samples. Video images were recorded under several magnifications using a Truevision Vista board. Image files with 256 gray levels were converted to 32 levels using histogram equalization. Textural features were evaluated by computing gray level differences and second-order gray level statistics. We discuss results for spatial gray level dependence (SGLDM) and gray level difference (GLDM) methods for a range of intersampling distances. Results for our carpet series show that SGLDM and GLDM yield very similar trends, statistical trends within each method are correlated, overall, these measures are generally effective for differentiating the degree of wear as well as providing an indication of textural periodicity, and changes in frequency may be examined by subjecting SGLDM and GLDM statistics to spectral analysis. This study addresses the problem of characterizing vari4tions in carpet texture due to use degradation. We confine our discussion to the wear component of carpet degradation and ignore soiling, staining, photodegradation, and so on. Carpets typically have cut or looped structures, with yams arranged in raised loops or cut raised loops (tufts). Quantifying wear in carpets amounts to monitoring changes in these surface structures. Carpet tufts are subject to mechanical deformation during wear, and the resulting altered or degraded surface displays distinctive visual cues. Subjective ranking has been traditionally applied to wear assessment, since the visual aspect of surface wear is so apparent and, as noted in reference 15, aesthetic degradation may long precede extensive mechanical degradation. While this method of ranking may be quite replicable, the classification methods and their parameters are not as amenable to analytical treatment as one would like, because, as with many human judgment scales, the rules of perception and evaluation that define &dquo;appearance loss&dquo; characteristics are not well understood.Ultimately, it would be desirable to develop automated methods based on changes in the patterns of light reflectance and scattering, luminance, and color collectively referred to as texture. Texture is a very open-ended concept. There are many different kinds of image analysis procedures that are sensitive to changes in &dquo;texture.&dquo; However, it has been our experience that many of these techniques do not yield consistent results when applied to a wide range of texture patterns and are often di...

Recognition of texture types of wear particles

“…Some of the most common extracted features containing texture information are: energy, entropy, inertia, local homogeneity, correlation, etc. [17,18,19].…”

Analysing wear in carpets by detecting varying local binary patterns

“…2,3 Manufacturers however require a more objective assessment because the human assessment is often too subjective. Much effort has been devoted to automate this rating process, initially on techniques extracting texture features from digital images, [4][5][6][7][8][9][10][11][12] and lately combining features from digital and range images,, [13][14][15][16] where a range image represents the depth information of the surface of a carpet. The range images are obtained with a 3D scanner designed specifically for scanning carpets.…”

Resilience behaviors of woven acrylic carpets under short- and long-term static loading

“…In image analysis techniques, the carpet surface or a photograph is scanned by a video camera. The encoded image is then analyzed by using a program based on such criteria as tuft definition, directionality, and periodicity related to known changes in the carpet pile (Mukhopadhyay, Sengonul, & Wilding, 1993;Pourdeyhimi, Ramanathan, & Javadpour, 1994;Sobus, Pourdeyhimi, Gerde, & Ulcay, 1991;Xu, Pourdeyhimi, & Sobus, 1993). Furthermore, a number of empirical and analytical models have been developed to predict the different factors affecting the thickness changes of a carpet.…”