The testing of the controlled illumination method to Marmousi data shows very reasonable results. However, there are some aspects which can be improved.
One of the most important assumptions in this technique is the unitary property of the wave propagation operator. In the actual numerical implementation of the wave propagation operator, however, this assumption is not valid, especially when the velocity changes a lot. In such case, we should use inversion technique to find synthesis operator. One good candidate is least squares technique. Ji (1992) showed least square datuming can be done very fast using conjugate gradient method.
The other important factor is missing traces. Most of conventional acquisition patterns leaves many missing traces at near and far offset. For optimizing the controlled illumination process, interpolation of the near and far offsets is necessary. The interpolation scheme discussed by Ji and Claerbout (1991) in the CMP point domain would be a good candidate for this purpose. When we illuminate reflectors in a deep depth, the far offset information is very important.
In the future, I'll apply the least-squares technique to find the synthesis operator and consider many interpolation algorithms to improve the resolution of the deep reflector images.