Time-lapse image registration using the local similarity attribute

3-D field data

difc,difc2
Figure 6. Application to the Duri field data. Base image and time-lapse differences: (a) before registration, reproduced from Lumley (1995a), (b) after registration.

Finally, Figure 6 shows an application of the proposed method to time-lapse images from steam flood monitoring in the Duri field, reproduced from Lumley (1995a,b). Before registration, real differences in the monitor surveys after 2 months and 19 months are obscured by coherent artifacts, which are caused by velocity changes both in the shallow overburden and in the reservoir interval [Figure 6a]. Similarly to the results of the synthetic experiments, local-similarity registration succeeds in removing artifact differences both above and below the reservoir level [Figure 6b]. After separating the time-shift effect from amplitude changes, one can image the steam front propagation more accurately using time-lapse seismic data. We expect our method to work even better on higher-quality marine data.

Time-lapse image registration using the local similarity attribute