We study the perception problem for garments (e.g. a pair of pants) using tools from computational topology: the identification of their geometry and position from point-cloud samples, as obtained e.g. with 3D scanners. We present a reconstruction algorithm based on Morse theory that proceeds directly from the point-cloud to obtain a cellular decomposition of the surface derived via a Morse function. No intermediate triangulation or local implicit equations are used, avoiding reconstruction-induced artifices. The results are a piecewise parametrization of the surface as a union of Morse cells, suitable for tasks such as noise-filtering or mesh-independent reparametrization, and a cell complex of small rank determining the surface topology. This algorithm can be applied to smooth surfaces with or without boundary, embedded in an ambient space of any dimension.
A novel programmable shader is proposed to accurately simulate the main important human visual defects under different situations in daily life. An improved model of eye model is introduced to reasonably predict the anatomical and optical properties of the human eye. This eye model is composed of an accommodation and color model, and both these models are combined to simulate the varying refractive power of the human eye and color vision deficiency. Finally, distributed ray tracing techniques is combined with this eye model to produce a variety of visual results using the programmable shader in NVIDIA OptiX environment
We present a multilayered appearance model for snake skin, inspired on its anatomy: The top layer is a thin film layer producing a specular iridescent reflection, while the bottom layer is a diffuse highly-absorbing layer, that results into a dark diffuse appearance that maximizes the iridescent color of the skin.
In the context of Industrie 4.0, it is necessary for several applications, to encode characteristics of a Boundary Representation of a manifold M in an economical manner. Two related characterizations of closed B-Reps (and the solid they represent) are (1) medial axis and (2) Reeb Graph. The medial axis of a solid region is a non-manifold mixture of 1-simplices and 2-simplices and it is expensive to extract. Because of this reason, this manuscript concentrates in the work-flow necessary to extract the Reeb Graph of the B-Rep. The extraction relies on (a) tests of geometric similarities among slices of M and (b) characterization of the topological transitions in the slice sequence of M . The process roughly includes: (1) tilt of the B-Rep to obtain an unambiguous representation of the level sets of M ,(2) identification and classification of the topological transitions that arise between consecutive level sets, (3) sample of Reeb graph vertices inside the material regions defined by the level sets, (4) creation of Reeb graph edges based on the type of topological transition and the 2D similarity among material regions of consecutive levels. Although the Reeb Graph is a topological construct, geometrical processing is central in its synthesis and compliance with the Nyquist-Shannon sampling interval is crucial for its construction. Future work is needed on the extension of our methodology to account for manifolds with internal voids or nested solids
The generation of realistic natural scenarios is a longstanding and ongoing challenge in Computer Graphics. LiDAR (Laser Imaging Detection and Ranging) point clouds have been gaining interest for the representation and analysis of real-world scenarios. However, the output of these sensors is conditioned by several parameters, including, but not limited to, distance to scanning target, aperture angle, number of laser beams, as well as systematic and random errors for the acquisition process. Hence, LiDAR point clouds may present inaccuracies and low density, thus hardening their visualization. In this work, we propose reconstructing the surveyed environments to enhance the point cloud density and provide a 3D representation of the scenario. To this end, ground and vegetation layers are detected and parameterized to allow their reconstruction. As a result, point clouds of any required density can be modeled, as well as 3D realistic natural scenarios that may lead to procedural generation through their parameterization.
With the increase in popularity of procedural urban modeling for film, TV, and interactive entertainment, an urgent need for editing tools to support procedural content creation has become apparent. In this paper, we present an end-to-end system for creating a library of reusable procedural parts in a rule-based setting to address this need. No trivial extension exists to perform this action in a way such that the resulting ruleset is ready for production. For procedural reusable parts, we need to handle the rulesets extracted from the source graphs, and later on, merge them with a target graph to obtain a final consistent ruleset. As one of the main contributions of our system, we introduce a library of reusable parts that could be seamlessly glued to other graphs and obtain consistent new procedural buildings. Hence, we focus on intuitive and minimal user interaction, and our editing operations perform interactively to provide immediate feedback.
Owing to the lack of accessibility of postpartum rooms nowadays, which are needed to teach the correct assessment of newborn health status to medical students, virtual reality and simulation are increasingly used for teaching and assessing visual perception tests that evaluate the condition of the newborn. This paper aims to explain the operation of the Apgar test evaluation simulator in a virtual reality environment. This virtual environment can be manipulated externally from a web browser to visualize and control the course of the simulation in real-time. In addition, an offline version would allow initialization and visualization of the Apgar test parameters without the need for synchronization with the virtual environment.
In this paper we present preliminary results of a new approach for interactive visualization of historical fabrics at yarn level based on models using Signed Distance Functions. They are compared with the results of Virtual Loom, an interactive application for rendering historical fabrics from the 15th to the 19th century, resulting from the European SilkNow project [PSA ∗ 18]. Yarns used on these fabrics has an special complexity due to a metallic cover that appears in some elements of the fabric. This new approach is based on rendering using Sphere Tracing. We will propose a signed distance function that allows both the visualization of a single yarn and a complete fabric in real time.
Random bin picking is still one of the main tasks for robotics in the current days. When the environment is very cluttered, the calculation of grasping positions can be highly demanding in terms of time and computing power. To ease the computation load, some parts arranging operations can be performed before the segmentation stage. For instance, for small and light parts, a feeder-vibrating table system can be used to separate the components, allowing them to be easily grasped, and increasing the overall performance of the solution. However, as the geometry and requirements for piece type are different, one or more feasibility tests need to be done for each case. These analyses are usually very time and cost intensive and require the use of expensive hardware such as robots, grippers, and prototype cells. The use of virtual reproductions of the environment like digital twins or physical-based simulations could help reduce the time and effort spent on designing the settings, nevertheless, their correct configuration is not trivial. This paper presents a simulation based analysis method for picking small-sized parts. It aims to supply the tools and define a streamlined procedure for efficient feasibility testing. Those concepts are applied in a specific bin picking scenario of multiple small electronic components. For each part type, a set of case-specific initial and boundary conditions are taken into account, then a series of performance metrics for both bin and vibrating table part picking are computed. The obtained information is decisive to make strategic decisions regarding the hardware requirements, the profitability, and the success probability of the project.
Modelling of atmospheric effects and, in particular, of clouds is a recurrent topic in computer graphics. Achieving realistic results poses a great challenge, and manual modelling of such complex structures is unfeasible in production. The use of procedural techniques, either through simulation or heuristic methods, is, thus, frequent. In this work we propose a model that allows atmospheric cloud modelling with small computational effort and a good degree of control by the modeller, speeding up the artistic process. The model is designed hierarchically, can be extended easily due to its modular structure and it allows an intuitive control through a graphical user interface. The tests show a variety of effects that can be generated, and the influence of the different parameters on the render time of the final scene.
Materials systems are an important element within the development of a renderer for an application such as a video game. Nowadays, the method to build a graphic style for a product involving a real-time engine implies a rendering system supporting a solid and concise materials system, as those well-established in real-time engines such as Unreal or Unity. This study presents an open-source application to serve as an editor of materials consisting of a modern real-time renderer. The application consists of a basic OpenGL real-time rendering engine to visualise 3D geometry and its appearance through the support of a node-based material editor to assemble materials in an intuitive and simple manner, without the use of programming and little technical knowledge. The culmination of the project and the achievement of its objectives was satisfactory. We concluded that this work can be used as a reference to understand real-time material systems and renderers and its state of the art in the video games industry.
Terrestrial Laser Scanners, also known as LiDAR, are often equipped with color cameras so that both infrared and RGB values are measured for each point sample. High-end scanners also provide panoramic High Dynamic Range (HDR) images. Rendering such HDR colors on conventional displays requires a tone-mapping operator, and getting a suitable exposure everywhere on the image can be challenging for 360° indoor scenes with a variety of rooms and illumination sources. In this paper we present a simple-to-implement tone mapping algorithm for HDR panoramas captured by LiDAR equipment. The key idea is to choose, on a per-pixel basis, an exposure correction factor based on the local intensity (infrared reflectivity). Since LiDAR intensity values for indoor scenes are nearly independent from the external illumination, we show that intensity-guided exposure correction often outperforms state-of-the-art tone-mapping operators on this kind of scenes.
Laser scanners enable the digitization of 3D surfaces by generating a point cloud where each point sample includes an intensity (infrared reflectivity) value. Some LiDAR scanners also incorporate cameras to capture the color of the surfaces visible from the scanner location. Getting usable colors everywhere across 360 scans is a challenging task, especially for indoor scenes. LiDAR scanners lack flashes, and placing proper light sources for a 360 indoor scene is either unfeasible or undesirable. As a result, color data from LiDAR scans often do not have an adequate quality, either because of poor exposition (too bright or too dark areas) or because of severe illumination changes between scans (e.g. direct Sunlight vs cloudy lighting). In this paper, we present a new method to recover plausible color data from the infrared data available in LiDAR scans. The main idea is to train an adapted image-to-image translation network using color and intensity values on well-exposed areas of scans. At inference time, the network is able to recover plausible color using exclusively the intensity values. The immediate application of our approach is the selective colorization of LiDAR data in those scans or regions with missing or poor color data.
Automated visual inspection is an ongoing machine vision challenge for industry. Faced with increasingly demanding quality standards it is reasonable to address the transition from a manual inspection system to an automatic one using some advanced machine learning approaches such as deep learning models. However, the introduction of neural models in environments such as the manufacturing industry find certain impairments or limitations. Indeed, due to the harsh conditions of manufacturing environments, there is usually the limitation of collecting a high quality database for training neural models. Also, the imbalance between non-defective and defective samples is very common issue in this type of scenarios. To alleviate these problems, this work proposes a pipeline to generate rendered images from CAD models of industrial components, to subsequently feed an anomaly detection model based on Deep Learning. Our approach can simulate the potential geometric and photometric transformations in which the parts could be presented to a real camera to faithfully reproduce the image acquisition behavior of an automatic inspection system. We evaluated the accuracy of several neural models trained with different synthetically generated data set simulating different transformations such as part temperature or part position and orientation with respect to a given camera. The results shows the feasibility of the proposed approach during the design and evaluation process of the image acquisition setup and to guarantee the success of the real future application. CCS Concepts • Computing methodologies → Quality Inspection; Industrial Manufacturing; Photo-realistic Rendering; CAD Models; Anomaly Detection; Deep Learning; Generative Adversarial Networks;
Three-dimensional point clouds have conventionally been used along with several sources of information. This fusion can be performed by projecting the point cloud into the image plane and retrieving additional data for each point. Nevertheless, the raw projection omits the occlusion caused by foreground surfaces, thus assigning wrong information to 3D points. For large point clouds, testing the occlusion of each point from every viewpoint is a time-consuming task. Hence, we propose several algorithms implemented in GPU and based on the use of z-buffers. Given the size of nowadays point clouds, we also adapt our methodologies to commodity hardware by splitting the point cloud into several chunks. Finally, we compare their performance through the response time. CCS Concepts • Computing methodologies → Massively parallel algorithms; Visibility; Point-based models;
