Pere-Pau Vázquez

Automatic View Selection Using Viewpoint Entropy and its Application to Image‐Based Modelling

In the last decade a new family of methods, namely Image‐Based Rendering, has appeared. These techniques rely on the use of precomputed images to totally or partially substitute the geometric representation of the scene. This allows to obtain realistic renderings even with modest resources. The main problem is the amount of data needed, mainly due to the high redundancy and the high computational cost of capture. In this paper we present a new method to automatically determine the correct camera placement positions in order to obtain a minimal set of views for Image‐Based Rendering. The input is a 3D polyhedral model including textures and the output is a set of views that sample all visible polygons at an appropriate rate. The viewpoints should cover all visible polygons with an adequate quality, so that we sample the polygons at sufficient rate. This permits to avoid the excessive redundancy of the data existing in several other approaches. We also reduce the cost of the capturing process, as the number of actually computed reference views decreases. The localization of interesting viewpoints is performed with the aid of an information theory‐based measure, dubbed viewpoint entropy. This measure is used to determine the amount of information seen from a viewpoint. Next we develop a greedy algorithm to minimize the number of images needed to represent a scene. In contrast to other approaches, our system uses a special preprocess for textures to avoid artifacts appearing in partially occluded textured polygons. Therefore no visible detail of these images is lost.

Unified Approach to Prefiltered Environment Maps

Different methods for prefiltered environment maps have been proposed, each of which has different advantages and disadvantages. We present a general notation for prefiltered environment maps, which will be used to classify and compare the existing methods. Based on that knowledge we develop three new algorithms: 1. A fast hierarchical prefiltering method that can be utilized for all previously proposed prefiltered environment maps. 2. A technique for hardware-accelerated prefiltering of environment maps that achieves interactive rates even on low-end workstations. 3. Anisotropic environment maps using the Banks model.

Way‐Finder: guided tours through complex walkthrough models

The exploration of complex walkthrough models is often a difficult task due to the presence of densely occluded regions which pose a serious challenge to online navigation. In this paper we address the problem of algorithmic generation of exploration paths for complex walkthrough models. We present a characterization of suitable properties for camera paths and we discuss an efficient algorithm for computing them with little or no user intervention. Our approach is based on identifying the free‐space structure of the scene (represented by a cell and portal graph) and an entropy‐based measure of the relevance of a view‐point. This metric is key for deciding which cells have to be visited and for computing critical way‐points inside each cell. Several results on different model categories are presented and discussed.

Omni‐directional Relief Impostors

Relief impostors have been proposed as a compact and high‐quality representation for high‐frequency detail in 3D models. In this paper we propose an algorithm to represent a complex object through the combination of a reduced set of relief maps. These relief maps can be rendered with very few artifacts and no apparent deformation from any view direction. We present an efficient algorithm to optimize the set of viewing planes supporting the relief maps, and an image‐space metric to select a sufficient subset of relief maps for each view direction. Selected maps (typically three) are rendered based on the well‐known ray‐height‐field intersection algorithm implemented on the GPU. We discuss several strategies to merge overlapping relief maps while minimizing sampling artifacts and to reduce extra texture requirements. We show that our representation can maintain the geometry and the silhouette of a large class of complex shapes with no limit in the viewing direction. Since the rendering cost is output sensitive, our representation can be used to build a hierarchical model of a 3D scene.

Perception-based illumination information measurement and light source placement

The automatic selection of good viewing parameters is very complex. In most cases, the notion of good strongly depends on the concrete application. Moreover, when an intuitive definition of good view is available, it is often difficult to establish a measure that brings it to the practice. Commonly, two kind of viewing parameters must be set: the position and orientation of the camera, and the ones relative to light sources. The first ones will determine how much of the geometry can be captured and the latter will influence on how much of it is revealed (i. e. illuminated) to the user. In this paper we will define a metric to calculate the amount of information relative to an object that is communicated to the user given a fixed camera position. This measure is based on an information-based concept, the Shannon entropy, and will be applied to the problem of automatic selection of light positions in order to adequately illuminate an object.

Depth-enhanced maximum intensity projection

The two most common methods for the visualization of volumetric data are Direct Volume Rendering (DVR) and Maximum Intensity Projection (MIP). Direct Volume Rendering is superior to MIP in providing a larger amount of properly shaded details, because it employs a more complex shading model together with the use of user-defined transfer functions. However, the generation of adequate transfer functions is a laborious and time costly task, even for expert users. As a consequence, medical doctors often use MIP because it does not require the definition of complex transfer functions and because it gives good results on contrasted images. Unfortunately, MIP does not allow to perceive depth ordering and therefore spatial context is lost. In this paper we present a new approach to MIP rendering that uses depth and simple color blending to disambiguate the ordering of internal structures, while maintaining most of the details visible through MIP. It is usually faster than DVR and only requires the transfer function used by MIP rendering.

Technical Section: Realtime automatic selection of good molecular views

The investigation of molecular structures often requires the use of graphics software to display different representations of the molecule of interest. Unfortunately, the commonly available visualization software is generally quite complex and requires a high degree of expertise for the user to obtain the desired images. Often, the selection of interesting views implies a considerable time and effort for nonexperienced users. Characterizing the desired properties the users may need is often impossible. In this paper we present a method to automatically determine certain views of molecules that can be used to study their chemical or physical properties. We have used Information Theorys Shannon entropy in order to characterize two kinds of views: views which show most of the structure of a molecule and views which show a low amount of information of an arrangement of molecules. The first ones can be used to study the composition of the molecule, that is to study certain chemical properties. The latter easily show how molecules are ordered in space and therefore are suitable to infer physical properties of compounds, such as resistance. Finally, we also present an adaptive, hardware accelerated algorithm that makes use of the features of graphics cards to make this calculation in realtime. Our method has proven to give good results as in most cases the views generated by our application can completely replace human involvement. For highly complex compounds, they can be either enough, or a good starting point. Often, our application also provides several views that could be missed by the users.

Viewpoint entropy: a new tool for obtaining good views of molecules

The computation of good viewpoints is important in several fields: computer graphics, removal of degeneracies in computational geometry, robotics, graph drawing, etc. However, in areas such as computer graphics there is no consensus on what a good viewpoint means and, consequently, each author uses his or her own definition according to the requirements of the application. In this paper we present a formal measure strongly based on Information Theory, viewpoint entropy, that can be applied to certain problems of Computer Graphics such as automatic exploration of objects or scenes and Scene Understanding. We also define a new measure, the orthogonal frustum entropy, in order to fulfill the requirements needed to visualize molecules. We design an algorithm that makes use of graphics hardware to accelerate computation, and whose complexity depends mainly on the number of views we want to analyze. Computation of good views of molecules is useful for molecular scientists, a field which includes practitioners from Crystallography, Chemistry, and Biology.

Automatic view selection through depth-based view stability analysis

Although the real world is composed of three-dimensional objects, we communicate information using two-dimensional media. The initial 2D view we see of an object has great importance on how we perceive it. Deciding which of the all possible 2D representations of 3D objects communicates the maximum information to the user is still challenging, and it may be highly dependent on the addressed task. Psychophysical experiments have shown that three-quarter views (oblique views between frontal view and profile view) are often preferred as representative views for 3D objects; however, for most models, no knowledge of its proper orientation is provided. Our goal is the selection of informative views without any user intervention. In order to do so, we analyze some stability-based view descriptors and present a new one that computes view stability through the use of depth maps, without prior knowledge on the geometry or orientation of the object. We will show that it produces good views that, in most of the analyzed cases, are close to three-quarter views.

Correlating Text and Images: Concept and Evaluation

This paper presents the concept and an evaluation of a novel approach to support students to understand complex spatial relations and to learn unknown terms of a domain-specific terminology with coordinated textual descriptions and illustrations. Our approach transforms user interactions into queries to an information retrieval system. By selecting text segments or by adjusting the view to interesting domain objects, learners can request additional contextual information. Therefore, the system uses pre-computed multi-level representations of the content of explanatory text and of views on 3D models to suggest textual descriptions or views on 3D objects that might support the current learning task. Our experimental application is evaluated by a user study that analyzes (i) similarity measures that are used by the information retrieval system to coordinate the content of descriptive texts and computer-generated illustrations and (ii) the impact of the individual components of these measures. Our study revealed that the retrieved results match the preferences of the users. Furthermore, the statistical analysis suggests a rough value to cut-off retrieval results according to their relevancy.