This dissertation addresses the problem of employing 3D depth information on solving a number of traditional challenging computer vision/graphics problems. Humans have the abilities of perceiving the depth information in 3D world, which enable humans to reconstruct layouts, recognize objects and understand the geometric space and semantic meanings of the visual world. Therefore it is significant to explore how the 3D depth information can be utilized by computer vision systems to mimic such abilities of humans. This dissertation aims at employing 3D depth information to solve vision/graphics problems in the following aspects: scene understanding, image enhancements and 3D reconstruction and modeling. In addressing scene understanding problem, we present a framework for semantic segmentation and object recognition on urban video sequence only using dense depth maps recovered from the video. Five view-independent 3D features that vary with object class are extracted from dense depth maps and used for segmenting and recognizing different object classes in street scene images. We demonstrate a scene parsing algorithm that uses only dense 3D depth information to outperform using sparse 3D or 2D appearance features. In addressing image enhancement problem, we present a framework to overcome the imperfections of personal photographs of tourist sites using the rich information provided by large-scale internet photo collections (IPCs). By augmenting personal 2D images with 3D information reconstructed from IPCs, we address a number of traditionally challenging image enhancement techniques and achieve high-quality results using simple and robust algorithms. In addressing 3D reconstruction and modeling problem, we focus on parametric modeling of flower petals, the most distinctive part of a plant. The complex structure, severe occlusions and wide variations make the reconstruction of their 3D models a challenging task. We overcome these challenges by combining data driven modeling techniques with domain knowledge from botany. Taking a 3D point cloud of an input flower scanned from a single view, each segmented petal is fitted with a scale-invariant morphable petal shape model, which is constructed from individually scanned 3D exemplar petals. Novel constraints based on botany studies are incorporated into the fitting process for realistically reconstructing occluded regions and maintaining correct 3D spatial relations. The main contribution of the dissertation is in the intelligent usage of 3D depth information on solving traditional challenging vision/graphics problems. By developing some advanced algorithms either automatically or with minimum user interaction, the goal of this dissertation is to demonstrate that computed 3D depth behind the multiple images contains rich information of the visual world and therefore can be intelligently utilized to recognize/ understand semantic meanings of scenes, efficiently enhance and augment single 2D images, and reconstruct high-quality 3D models.

[pt] Atualmente o processo que envolve a construção de um empreendimento é baseado em documentos impressos e desconexos. Há uma tendência global de que um mesmo projeto seja elaborado por diversas empresas e diferentes equipes, porém, as ferramentas computacionais que são utilizadas ainda são as mesmas de décadas atrás. Devido a essa mudança de paradigma, a metodologia conhecida como modelagem da informação da construção, BIM, vem aos poucos ganhando aceitação na área de AEC (Arquitetura, Engenharia e Construção). Motivada por esse processo de transição, essa dissertação tem como objetivo estudar e avaliar algumas das ferramentas computacionais disponíveis no mercado que são utilizadas na implementação da metodologia BIM. Essa avaliação é feita a partir do projeto de um empreendimento real que foi feito utilizando a metodologia tradicional, isto é, documentos 2D independentes. O projeto é modelado em 3D com uma base de dados integrada de acordo com a metodologia BIM. O projeto baseado no modelo 3D é então comparado com o projeto original. Ao término desse estudo são obtidos exemplos reais dos benefícios da modelagem 3D parametrizada. Uma série de deficiências são detectadas nos documentos 2D originais, o que não teria acontecido caso a metodologia BIM fosse usada desde o início do projeto. Além disso, é avaliada a integração entre uma ferramenta de modelagem 3D estrutural (Autodesk Revit Structure 2012) e um software de análise estrutural (Autodesk Robot 2012). Estratégias de integração são testadas e as melhores práticas são descritas em detalhe. Os resultados obtidos sugerem que o uso da metodologia é promissor e sua implementação deve ser seriamente considerada no Brasil. / [en] Currently the process involved in the construction of a facility remains fragmented and dependent on paper-based communication modes. There is a global trend where a single project may be developed by different companies and may involve several teams of specialists. However, the computational tools that are still in use are the same as the ones being used for decades. Due to this paradigm shift, a methodology known as Building Information Modeling, BIM, has been gaining acceptance in the AEC (Architecture, Engineering and Construction) area. Motivated by this process of transition, this thesis aims to study and evaluate some of the computational tools available in the market that are used to implement the BIM methodology. This assessment is made using the design of an actual facility, which is done using the traditional 2D design methodology, i.e., disjoint 2D documents. The project is modeled in 3D with using integrated database according to the BIM methodology. The resulting project is then compared to the original project. At the end of this study, real examples of the benefits of parameterized 3D modeling are obtained. Several inconsistencies in the original design documents are observed, which would not have occurred if the BIM methodology had been used from the beginning of the project. In addition, the integration between a 3D structural modeling tool (Autodesk Revit Structure 2012) and a structural analysis tool (Autodesk Robot 2012) is evaluated. Integration strategies are tested and the best practices are described in detail. The findings of this work suggest the use of the BIM methodology is promising and support its use in Brazil.