From imagery processing perspective, the mosaic dataset model (10.1) works eventually

larryzhang

Just for sharing:

From imagery processing perspective, some processing operations work eventually with the mosaic dataset model (10.1).

We recently use the mosaic dataset model in ESRI (10.1) to try and replace some of tradtional imagery processing tasks, including geo-correction with GCP (ground control points) and mosaic operation. On our tests, the results are pretty good and the workflow are also efficient!

In fact, it really expedites massive imagery processing and delivery in operation. For example, it effectively improves geometric accuracy by the tool ???Georeferencing??? via only UPDATE Georeferencing (no need ???Rectify??? anymore in ArcGIS or Erdas LPS). Secondly, the mosaic dataset model with ???seamlines??? function (for overview creation) can also produce acceptable results, when merging 'color-slightly unbalanced' tiles of SPOT5 imagery.

Of course, right now we still are not able to export those ???processed??? SPOT 5 images with ???overviews??? as a ???batch??? for interoperation (used offline, or for other engineering and earth-science systems). In operation, it is expected to be automatically tiled and named on one of ???fields??? from the polygon feature class.

Many thanks, raster team

Larry ZHANG at Aramco

++++++++++
Note,

1. Those SPOT 5 imagery 2012 are pre-processed by SPOT system, including 2.5-m pan-sharpened;
2. The edge detection algorithm is used when building 'seamlines' for overviews creation;
3. The BLEND mosaic operator is also combined to use, when merging overlapping imagery tiles;
4. On hands-on comparision, the algorithm ???BLEND??? still needs to be improved, against ???feathering??? function in ERDAS Imagine. In other word, the BLEND mosaic operator in mosaic dataset model looks not VERY functional, when zoom-in to the larger scales.

larryzhang

Just for sharing:
.....

1. .....
2. The BLEND mosaic operator is also combined to use, when merging overlapping imagery tiles;
3. On hands-on comparision, the algorithm �??BLEND�?? still needs to be improved, against �??feathering�?? function in ERDAS Imagine. In other word, the BLEND mosaic operator in mosaic dataset model looks not VERY functional, when zoom-in to the larger scales.

For update,

With Peter Becker (Esri) assistance this issue is solved via ONLY 'Seamlines' for mosaic methods at larger scales (full-resolution).

Regards,