China
Africa
Explore chapters and articles related to this topic
Image in Multimedia
Published in Sreeparna Banerjee, Elements of Multimedia, 2019
An image is a spatial representation of an object or a two-dimensional (2D) or three-dimensional (3D) scene. A picture, which is a representation of a momentary event from a 3D spatial world, is an example of a 2D image, whereas a computer game can be a sequence of 3D images [1–7]. An image is modeled as a continuous function defining a rectangular region in a plane. Images formed from actual convergence of light rays are real images, whereas images formed from the extrapolation of light rays that are non-existent are imaginary. The image intensity distribution is proportional to the radiant energy received in the electromagnetic band to which the sensor/detector is sensitive (sometimes referred to as the intensity image) [1,2,6]. A digital image is represented by a matrix of numeric values, each representing a quantized intensity value I (r, c), where r is the row and c is the column. An image can be obtained in digital form by sampling the image intensity function at discrete intervals. The sample points representing the smallest discrete intensity value is called a pixel. A pixel [1–6] is essentially a picture element (pel) in two dimensions. A voxel is a 3D picture element of a very thin slice, or a volume element. This process of selecting the discrete points is referred to as sampling. The value of the intensity at each pixel is obtained by approximating the continuous value to a bin value. This approximation is referred to as quantization, and the approximation corresponds to the number of bits per pixel required to obtain this intensity value.
Software Phantoms for X-ray Radiography and Tomography
Published in Paolo Russo, Handbook of X-ray Imaging, 2017
Voxel phantoms are based on discretization of an object into tiny cubes, called voxels, with uniform characteristics. One approach to obtain voxel models is by digitizing an object into voxels. Such an example is shown in Figure 57.6a, where a water CSG slab phantom is transformed to a voxel water phantom. Another source for voxel phantoms are the 3D scanning technologies: CT (computed tomography), MRI (magnetic resonance imaging), Tomosynthesis, PET (positron emission tomography), as well as cross-sectional photographs of a cadaver through which detailed patient-specific anatomy may be obtained. Images obtained with these modalities represent a matrix of pixels. A consecutive set of such images can be considered as a 3D matrix made of voxels, where each voxel belongs to a specific organ or tissue. Another example is shown in Figure 57.6b, where a model of a patient head is designed initially from CT data and then converted to solid based geometry. This phantom is defined as an assembly of 3D cubes. In general, voxel elements may not be cubes, rather they can be parallelepipeds. The properties of each voxel are stored within each voxel. These are indexes that correspond to given material.
Segmentation, Registration, and Fusion of Medical Images
Published in Alexander D. Poularikas, Stergios Stergiopoulos, Advanced Signal Processing, 2017
The result of segmentation is a classification that labels every voxel to be part of a certain region. This is referred to as binary segmentation since a voxel either shares a property with its neighbors or not. Because medical imaging techniques like computed tomography (CT) or magnetic resonance imaging (MRI) produce discrete volume grids, certain voxels may represent two different materials, e.g., on object boundaries. The so called partial volume effect leads to an uncertainty whether the voxel has to be assigned to the one object or the other. In contrast, fuzzy segmentation only computes a probability that a voxel belongs to a certain region. In the remainder of this chapter we will, however, focus on binary segmentation since most medical imaging classification techniques target at a clear distinction of the detected structures. For an overview of fuzzy segmentation techniques see Yoo [1].
Fabrication of parts with heterogeneous structure using material extrusion additive manufacturing
Published in Virtual and Physical Prototyping, 2021
Aggelos Vassilakos, John Giannatsis, Vassilis Dedoussis
A unified methodology for the representation and fabrication of highly heterogeneous parts via MEX is investigated in this paper. The proposed methodology employs a voxel modelling approach for the definition of material deposition density at specified part locations. The voxel model of a specific part can be created either directly, employing a voxel editing software, or indirectly, i.e. by ‘voxelizing’ (decomposing in voxel form) a previously designed CAD model. This offers some sort of flexibility to the designer who can choose whether to construct the structure by editing voxels or by using standard CAD modelling approaches. Voxels are subsequently decomposed into layer elements (laminae) which are employed for representing heterogeneity in the layer level. Based on the locally defined deposition density, laminae deposition paths are generated and joined to form a single infill path for each one of the fabrication layers.
Opportunities and challenges of quality engineering for additive manufacturing
Published in Journal of Quality Technology, 2018
Bianca M. Colosimo, Qiang Huang, Tirthankar Dasgupta, Fugee Tsung
One key feature of x-ray CT is that it produces voxel-based reconstructions. A voxel is a three-dimensional extension of a pixel. As pixel-based images can be modeled as a grayscale on a bidimensional grid, voxel-based data can be represented as a grayscale in a three-dimensional or volume grid (Figure 7). Surface reconstruction and geometry modeling starting from voxel-based data is an interesting area where more research is needed. Existing literature on data modeling and monitoring in biomedical applications of x-ray CT will possibly represent a starting point (Kalender 2006). Furthermore, multisensory data fusion in dimensional and geometrical metrology will possibly be further exploited to combine data provided by different metrological systems to enhance surface reconstruction for AM applications (Weckenmann et al. 2009; Colosimo et al. 2015; Wang et al. 2015; Xia et al. 2011).
A voxel-based method for designing a numerical biomechanical model patient-specific with an anatomical functional approach adapted to additive manufacturing
Published in Computer Methods in Biomechanics and Biomedical Engineering, 2019
Augustin Lerebours, Frederic Marin, Salima Bouvier, Christophe Egles, Alain-Charles Masquelet, Alain Rassineux
This work presents an original approach combining B-rep and a voxel-based representation. This modelling system has a number of features that make it a valuable alternative, namely an ability to control accuracy at a desired level, Boolean and algebraic operations, connection of slightly discontinuous surfaces, and undemanding modelling. Voxelization involves converting geometric objects into a set of voxels (configurable numerical cuboids) that best approximates the original object (Jense 1989). Voxel-based model are insensitive to object complexity and make for ease of display and model manipulation (Boolean operations) (Jense 1989). Little work has been published on the use of voxel-based models for geometric reasoning in engineering design (including for AM).