RPCs GENERATION

RPC Generation & Bias-Correction from Sensor Orientation Data

There are basically two possible approaches to the generation and bias-correction of RPCs when ALOS metadata is available, as indicated in the following flowchart:

RPC-only option Pushbroom model option

i) Import of metadata

Metadata, including orbit and attitude observations and camera mounting data, is imported when the images are imported. This is achieved by:

File - Import Image(s) - Formatted Image File(s)

In the example illustrated below, four image chips of an ALOS nadir scene are chosen, these being listed in the Import Image(s) dialogue box. The Import Pushbroom Metadata option must then be ticked to ‘on’, after which the Import button is selected to import the images, their thumbnails appearing in the thumbnail view, along with the metadata nodes.

ii) RPC-only option

· RPC generation: To generate RPCs, right-click on the pushbroom icon appearing in the Barista thumbnail view and choose Generate RPCs. A dialog box will appear asking for the elevation range of the region of interest. It is quite important that all object point heights for the scene are within this range. Once the operation is performed, the RPC icon will appear in the Barista thumbnail view. Note, however, that if there is already an existing RPC node for the image, the current RPCs in the project will be replaced with the newly calculated RPCs.

· RPC Bias Correction: Once RPCs are available for the image or images, bias-corrected RPCs can be produced via the standard adjustment procedures detailed in the Barista User Manual.

iii) Pushbroom model option

· Sensor Orientation Adjustment: Sensor orientation adjustment visa the pushbroom sensor model is performed via Adjustment__ - Start Adjustment. Further details on the process can be found in the Barista Users Manual. In the context of generating RPCs, the pushbroom scanner model offers a rigorous means of correcting the orbit and attitude data for biases prior to generation of the bias-corrected RPCs. For ALOS, this is the preferred approach, though it does require more ground control than the RPC generation followed by bias correction to the RPCs. This option must be followed in cases where image strips (say >2 images) are processed.

· RPC generation: To generate RPCs for the adjusted image orientation, which will now be bias-free, follow the procedure outlined above in (ii).

Experimental testing with ALOS PRISM and AVNIR-2 imagery has shown that the generation of bias corrected RPCs by the two methods produces agreement in the back-projection of object points into the image of around 1 pixel, which is adequate for practical applications. __

RPC generation for an ALOS PRISM TRIPLET

The generation of RPCs for an ALOS PRISM image triplet follows the pushbroom__ model option described above, in which the sensor orientation data is first refined via a bundle adjustment using the pushbroom sensor model. RPCs, free of bias, are then generated directly from the adjusted orientation parameters for the Forward, Nadir and Backward images forming the triplet.