Click on image to view larger version.
FIGURE 14. Gradient algorithm. This plot displays difference between estimated and measured projections (vertical axis) as function of values in 2 pixels of an image. Black lines are contour lines. Goal of algorithm is to find lowest point. From initial estimate for image (point A), step along steepest descent (dashed arrow) to reach point B. Then, at B, step along steepest descent (solid arrow) to reach minimum (point C). Values for 2 pixels at location C give solution. Note that, depending on location for starting point A, minimum can be different.
