Denis V. Ivanov

Seamless Mosaicing of Image-Based Texture Maps

Image-based object modeling has emerged as an important computer vision application. Typically, the process starts with the acquisition of the image views of an object. These views are registered within the global coordinate system using structure-and-motion techniques, while on the next step the geometric shape of an object is recovered using stereo and/or silhouette cues. This paper considers the final step, which creates the texture map for the recovered geometry model. The approach proposed in the paper naturally starts by backprojecting original views onto the obtained surface. A texture is then mosaiced from these back projections, whereas the quality of the mosaic is maximized within the process of Markov random field energy optimization. Finally, the residual seams between the mosaic components are removed via seam levelling procedure, which is similar to gradient-domain stitching techniques recently proposed for image editing. Unlike previous approaches to the same problem, intensity blending as well as image resampling are avoided on all stages of the process, which ensures that the resolution of the produced texture is essentially the same as that of the original views. Importantly, due to restriction to non-greedy energy optimization techniques, good results are produced even in the presence of significant errors on image registration and geometric estimation steps.

Oriented visibility for multiview reconstruction

Visibility estimation is arguably the most difficult problem in dense 3D reconstruction from multiple arbitrary views. In this paper, we propose a simple new approach to estimating visibility based on position and orientation of local surface patches. Using our concept of oriented visibility, we present a new algorithm for multiview reconstruction based on exact global optimization of surface photoconsistency using graph cuts on a CW-complex. In contrast to many previous methods for 3D reconstruction from arbitrary views, our method does not depend on initialization and is robust to photometrically difficult situations.

Color Distribution ‐ A New Approach to Texture Compression

Texture compression is recently one of the most important topics of 3D scene rendering techniques, because it allows rendering more complicated high‐resolution scenes. However, because of some special requirements for these type of techniques, the commonly used block decomposition approach may introduce visual degradation of image details due to lack of colors. We present here a new approach to texture compression, which allows sharing of one color by several blocks providing a larger number of unique colors in each particular block and the best compression ratio. We also present an iterative algorithm for obtaining distributed colors on a texture, and discuss some advantages of our approach. The paper concludes with comparison of our technique with S3TC and other block decomposition methods.

Spatial Patches ‐ A Primitive for 3D Model Representation

The commonly used solution for real‐life 3D model representation is polygonal spatially consistent geometry, with texture, and, optionally, bump or displacement maps attached. Although the idea of displacement mapping is well known, there are just a few approaches to its efficient implementation. In this paper we develop a technique that allows for efficient representation and rendering of 3D models by getting a new angle on the displacement mapping concept. We introduce a new primitive that is defined as the range image of a small part of the models surface; therefore, it is called a spatial patch. The whole model is just a collection of patches with no connectivity information between them. Such a representation can be directly acquired by 3D scanning machinery, and stored in a compact uniform form. It also allows for efficient visualization, which is the major focus of this paper. Thus, we present the logical structure of a rendering unit based on conventional z‐buffering, and discuss the involved algorithms in detail. These algorithms benefit from modern features of computing units for which we believe the proposed technique can be used in a wide range of applications dealing with real‐life 3D data.

An efficient integer-based skeletonization algorithm

Abstract The problem of generating vector representation of a raster image has been the question of present interest for decades. Although, there are many approaches to this problem, most of them suffer from sophisticated computations and irrational memory usage. We introduce here a new skeletonization algorithm that is very efficient in terms of time and memory consumption. Starting from the description of our approach to the concept of skeleton on raster grid, we present a comprehensive explanation of the algorithm. Its main idea consists of generating a special polyline for each raster line considering them in top-to-bottom direction; skeleton is constructed from points of these polylines. The skeletons obtained by our algorithm allow the precise reconstruction of the initial raster shapes; therefore they may reflect some artifacts possible on raster grid. For this reason, we also present here two algorithms of simplification and our approach to defects filtering on a stage of skeleton generation. The time of simplification by presented algorithms depends linearly on the number of input points that makes these methods fast and efficient. The paper concludes with performance evaluation and discussion of possible implementations of the introduced techniques.

Modeling with spatial patches

Spatial patches have been introduced (Ivanov and Kuzmin, 2001) as an alternative primitive for the representation of 3D objects potentially acquired from real life. Defined as range images of small regions of an objects surface, they have proven to allow for efficient storage and rendering of such surfaces, and they also exhibit great descriptive capabilities. We present several technologies that we have developed for generating 3D objects represented by spatial patches. These technologies allowed us to convert explicitly defined surfaces, textured polygonal meshes, range images and point clouds to the proposed representation. We also describe how advantageous properties of the new primitive are exploited for efficient reconstruction from range images, discuss the particular problems that we faced during our research, provide many examples to demonstrate the results that we achieved, and conclude with discussion of future work.