Catálogo de publicaciones - libros

In this paper we approach the problem of fast surface grading of flat pieces decorated with random patterns. The proposed method is based on the use of global statistics of color computed in the CIE Lab space. Two other fast methods based on color histograms [1] and Centile-LBP features [8] are introduced for comparison purposes. We used CIE Lab in order to provide accuracy and perceptual approach in color difference computation. Experiments with RGB were also carried out to study CIE Lab reliability. The ground truth was provided through an image database of ceramic tiles. Nevertheless, the approach is suitable to be extended to other random decorated surfaces like marble, granite, wood or textile stuff. The experiments make us to conclude that a simple collection of global statistics of color in the CIE Lab space is powerful enough to well discriminate surface grades. The average success surpasses 95% in most of the tests, improving literature methods and achieving factory compliance.

The use of image analysis for the aesthetical characterisation of stone slab surfaces has been studied during last ten years and has proved efficiency for an industrial and commercial application. This work aims to identify operational parameters specifically conceived for the classification of marble tiles. In this specific case the meaningful aesthetical properties are mainly linked to the anisotropy of the RGB intensities and, specifically, to the “veins”. Starting from the classical geostatistical and morphological modelling (variograms, granulometries, etc.), specific operational parameters have been obtained for a quantitative measurement of veins density, colour, and geometrical features (width, shape, continuity, etc.). The actual methodology defines commercial categories on the base of a self-appraisal process, which identifies intervals of several parameters. The current procedure is too rigid and doesn’t allow choosing in an intuitive way the discriminating properties. The proposed approach identifies understandable characteristics (vein features), and proposes quantitative indexes which actually satisfy the commercial classification of marbles.

This paper presents a new methodology to detect leather defects, based on the wavelet transform. The methodology uses a bank of optimised filters, where each filter is tuned to one defect type. Filter shape and wavelet sub-band are selected based the maximisation of the ratio between features values on defect regions and on normal regions. The proposed methodology can detect defects even when small features variations are present, which are not detect by generic texture classification techniques, and is fast enough to be used for real-time leather inspection.

This article presents the results of an unsupervised segmentation algorithm in multispectral images. The algorithm uses a minimization function which takes into account each band intensity information together with global edge criterion. Due to the unsupervised nature of the procedure, it can adapt itself to the huge variability of intensities and shapes of the image regions. Results shows the effectiveness of the method in multispectral fruit inspection applications and in remote sensing tasks.

Thresholding techniques are the simplest and most widely used methods to automatically segments images. These are used to segment images into several regions. This paper works over two sets of Iberian ham images: images taken by a digital camera (CCD) and Magnetic Resonance images (MRI), in order to establish a comparative for the performance on each kind of images. A methodology to determine the intramuscular fat (IMF) level of Iberian ham using computer vision techniques has been developed, as an attempt to find an alternative methodology to the traditional and destructive methods. The correlation between the chemical data and the computer vision results have been established in the paper. The main conclusions of the work are that better results have been obtained for MRI, which do not require preprocessing methods. So, the proposed approach to determine the IMF level could be considered as an alternative to the traditional and destructive methods.

The analysis focus on dam breaks stems from their ability to offer a simplified, yet effective workbench for debris flow waves, which in turn are helpful in gaining a deeper understanding of the highly destructive debris flows. High-speed recordings of granular flows arising from a dam-break-like event can be processed to extract useful information about the flow dynamics. Gradient-based optical-flow techniques cannot compute the correct velocity field as they detect the flow induced by the boundary evolution. Methods that are based on cross-correlation, such as particle imaging velocimetry (PIV), are able to capture the micro-scale flow, but, as they are designed for flows within fixed boundaries, they cannot deal directly with dam-break-caused flows because such flows, by their own nature, exhibit a fast moving boundary. This paper presents a procedure that is able to compute the evolving background and supply it to a PIV program as a masking region that should be excluded from the computation of the flow velocity field. This improvement leads to reliable results, while reusing existing software. All the resulting quantities are being used to tune a mathematical model describing the observed flows.

The geomaterials used in this study are granites from Finland with very similar mineral composition. Visual evaluation of the rock texture is done to determine the most significant features of the patterns for the analysis of heterogeneity of meso textures are grain size and grain size spatial distribution. These are compared to results of parameters calculated using image structure analyser. Images are capture with a scanner of the polished slabs that are 9*9 cm in size. The geo textures are expressed by four main parameters: textural entropy, homogeneity, contrast and textural correlation. Reducing the number of parameters to entropy and textural correlation significantly reduce the calculation time. These two parameters are considered to be the most significant. The other two, homogeneity and contrast, can be estimated. The parameter textural correlation yields better results than does textural entropy. Comparison of the analysis of textures visually and using image analysis shows that textural parameters have to be further worked in order to have a better performance.

In this paper an approach for automatic target detection and tracking, using multisensor image sequences with the presence of camera motion is presented. The approach consists of three parts. The first part uses a motion segmentation method for the detection of targets in the visible images sequence. The second part uses a Gaussian background model for detecting objects presented in the infrared sequence, which is preprocessed to eliminate the camera motion. The third part combines the individual results of the detection systems; it extends the Joint Probabilistic Data Association (JPDA) algorithm to handle an arbitrary number of sensors. Our approach is tested using image sequences with high clutter on dynamic environments. Experimental results show that the system detects 99% of the targets in the scene, and the fusion module removes 90% of the false detections.