Consider the computational costs associated with the geometry/weight matrix in CIT (e.g. a 100x100 pixel grid in latitude and latitude will have 10000 unknown pixel values to solve, but if it is extended to 100 pixels/voxels deep there will now be 1000000 values to solve in each iteration) and the fact that this matrix is VERY sparse ( ∼99% of cells are zero in a 100x100 grid; the greatest number of pixels a ray can pass through on the reconstruction grid is 141 pixels, at a 45 degree angle of elevation above the horizon) and will be even more sparse for 3D/4D reconstruction. What is the benefit of performing higher dimension reconstructions rather than using 100x100 grids for each slice of longitude and stacking these plots to for a quasi-3D/4D product?
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix