Texture-based volume rendering

from Wikipedia, the free encyclopedia

Texture-Based Volume Rendering is a method for the spatial representation of cross-sectional images such as those created with CT or MRI images.

In this process, textures of the volume to be displayed are geometrically placed next to one another as parallel layers / levels (slicing). The polygons of the layers are provided with texture coordinates and define the alignment of the volume (e.g. vertical sections through an object). The corresponding data values ​​for display can be calculated using the texture coordinates ( compositing ).

2D-based texture-based volume rendering

One approach to displaying 3D graphics is to use textures that are strung together and that make spatial images appear almost continuously as 2D individual layers by means of bilinear interpolation in a three-dimensional proxy geometry.

This proxy geometry, consisting of a stack of 2D layers, gives the viewer a spatial impression of an object. The interpolation is used to determine and fill the points between the layers in order to give the impression of a coherent three-dimensional object. The image synthesis takes place along the polygons of the stack of layers to be displayed from the rearmost image points to be displayed to the foremost ( back-to-front-order ).

Rotations that affect the scene to be displayed must be added to the textures (see Texture Mapping ).

Relation to the local space (object)

On the local coordinate system of the object, due to the alignment on the object, a coordinate (for example on the z-axis) is constant along a slice in each case, provided the slices are oriented on a main axis of the local space.

Relation to global space (camera)

Viewing the texture-based volume requires shifting or changing the texture layers when navigating the scene or rotating the object at critical points: With a perspective that extends over a rotation of 45 °, the undefined spaces can theoretically be removed see between the textures.

advantages

It is to be regarded as a simple procedure due to its low complexity and possibly a powerful instrument. The interpolation can be calculated quickly on the graphics hardware and, if necessary, in parallel steps. Since the textures are 2D, OpenGL can be used to implement almost any common hardware quickly. These advantages contrast with alleged quality losses.

The visualization technique using 3D textures is an alternative and now more popular .

Problems

When looking at the polygon disks in great detail, deviations or artifacts may appear at the edges of the disks . Additional filtering (e.g. Gaussian filter) can possibly produce a softer gradient over the layers, whereby a loss of quality of the view (a so-called blurred image) must be risked. As a result of the precaution to avoid unfavorable moments of perspective, three copies or data sets of the volume data set must also be kept (in the memory). This can lead to hardware performance problems and a reduction in immersion for the recipient.

2D-based multi-texture-based volume rendering

A major difference to the 2D-based method is that the planar polygons are used to interpolate trilinearly between four voxel corner points of two layers . A new grid point with corresponding data values ​​( usually opacity and color in image synthesis ) is calculated using eight grid points and bridges the gap between two voxel cells .

literature

  • Engel, K. et al .: Real-Time Volume Graphics , p. 49 ff., AK Peters Ltd. 2006. ISBN 1-56881-266-3