Catálogo de publicaciones - libros

The focus of the study is high quality image acquisition in colorimetric and multispectral formats. The aim is to combine the spatial resolution of digital images with the spectral resolution of color measurement instruments, to allow for accurate colorimetric and spectral measurements in each pixel of the acquired images. An experimental image acquisition system is used, which besides trichromatic RGB filters also provides the possibility of acquiring multi-channel images, using a set of narrowband filters. To derive mappings to colorimetric and multispectral representations, two conceptually different approaches are used. In the model-based characterization, the physical model describing the image acquisition process is inverted, to reconstruct spectral reflectance from the recorded device response. In the empirical characterization, the characteristics of the individual components are ignored, and the functions are derived by relating the device response for a set of test colors to the corresponding colorimetric and spectral measurements, using linear and polynomial least squares regression. The results indicate that for trichromatic imaging, accurate colorimetric mappings can be derived by the empirical approach, using polynomial regression to CIEXYZ and CIELAB. However, accurate spectral reconstructions requires for multi-channel imaging, with the best results obtained using the model-based approach.

Support vector clustering (SVC) has proven an efficient algorithm for clustering of noisy and high-dimensional data sets, with applications within many fields of research. An inherent problem, however, has been setting the parameters of the SVC algorithm. Using the recent emergence of a method for calculating the entire regularization path of the support vector domain description, we propose a fast method for robust pseudo-hierarchical support vector clustering (HSVC). The method is demonstrated to work well on generated data, as well as for detecting ischemic segments from multidimensional myocardial perfusion magnetic resonance imaging data, giving robust results while drastically reducing the need for parameter estimation.

We present a one-pass framework for filtering vector-valued images and unordered sets of data points in an -dimensional feature space. It is based on a local Bayesian framework, previously developed for scalar images, where estimates are computed using expectation values and histograms. In this paper we extended this framework to handle -dimensional data. To avoid the curse of dimensionality, it uses importance sampling instead of histograms to represent probability density functions. In this novel computational framework we are able to efficiently filter both vector-valued images and data, similar to e.g. the well-known bilateral, median and mean shift filters.

This paper describes a novel method of extracting moving region boundaries from the frames of an image sequence by filtering information of the first temporal cooccurrence matrix using the corresponding matrix from the next frame. The method described in this paper does not make use of any threshold and it is very robust and efficient with respect to noise.

A novel and computationally simple method is presented for triangulation of 3D points corresponding to the image coordinates in a pair of stereo images. The image points are described in terms of homogeneous coordinates which are jointly represented as the outer products of these homogeneous coordinates. This paper derives a linear transformation which maps the joint representation directly to the homogeneous representation of the corresponding 3D point in the scene. Compared to the other triangulation methods this approach gives similar reconstruction error but is numerically faster, since it only requires linear operations. The proposed method is projective invariant in the same way as the optimal method of Hartley and Sturm. The methods has a ”blind plane”; a plane through the camera focal points which cannot be reconstructed by this method. For ”forward-looking” camera configurations, however, the blind plane can be placed outside the visible scene and does not constitute a problem.

Good estimate and simulation of the behavior of additive noise is central to the adaptive restoration of images corrupted with Gaussian noise. This paper presents a double adaptive filtering scheme in the sense that the filter is able to estimate the variance of additive noise in order to determine the filter gain for pixel updating, and also able to decide if the pixel should remain unfiltered. Experimental results obtained from the restoration of several images have shown the superiority of the proposed method to some benchmark image filters.

This paper presents a complete high resolution aerial-images processing workflow to detect and characterize vegetation structures in high density urban areas. We present a hierarchical strategy to extract, analyze and delineate vegetation areas according to their height. To detect urban vegetation areas, we develop two methods, one using spectral indices and the second one based on a Support Vector Machines (SVM) classifier. Once vegetation areas detected, we differentiate lawns from treed areas by computing a texture operator on the Digital Surface Model (DSM). A robust region growing method based on the DSM is proposed for an accurate delineation of tree crowns. Delineation results are compared to results obtained by a Random Walk region growing technique for tree crown delineation. We evaluate the accuracy of the tree crown delineation results to a reference manual delineation. Results obtained are discussed and the influential factors are put forward.

In this paper a new image retrieval algorithm is proposed which aims to discard irrelevant images and increase the amount of relevant ones in a large database. This method utilizes a two-stage ant colony algorithm employing in parallel color, texture and spatial information. In the first stage, the synergy of the low-level descriptors is considered to be a group of ants seeking the optimal path to the “food” which is the most similar image to the query, whilst settling pheromone on each of the images that they confront in the high similarity zone. In the second stage additional queries are made by using the highest ranked images as new queries, resulting in an aggregate deposition of pheromone through which the final retrieval is performed. The results prove the system to be satisfactorily efficient as well as fast.

Denoising video is an important task, especially for videos captured in dim lighting environments. The filtering of video in a volumetric manner with time as the third dimension can improve the results significantly. In this work a 3D bilateral filter for edge preserving smoothing of video sequences exploiting commodity graphics hardware is presented. A hardware friendly streaming concept has been implemented to allow the processing of video sequences of arbitrary length. The clear advantage of time-based filtering compared to frame-by-frame filtering is presented as well as solutions to current limitations for volume filtering on graphics hardware. In addition, a significant speedup over a CPU based implementation is shown.

Image-Based Lighting (IBL) has become a very popular approach in computer graphics. In essence IBL is based on capturing the illumination conditions in a scene in an omni-directional image, called a light probe image. Using the illumination information from such an image virtual objects can be rendered with consistent shading including global illumination effects such as color bleeding.

Rendering with light probe illumination is extremely time consuming. Therefore a range of techniques exist for approximating the incident radiance described in a light probe image by a finite number of directional light sources. We describe two such techniques from the literature and perform a comparative evaluation of them in terms of how well they each approximate the final irradiance. We demonstrate that there is significant difference in the performance of the two techniques.