Go to the USGS Home Page

USGS MIPS Sidescan Sonar Mosaicking Procedure






Method 1: MANUAL MOSAICKING

This method assumes the differences between two GEOMETRICALLY PROJECTED sonar strips are simply a line/sample linear offset. Using the method outlined here the user will compute an average line and/or sample difference for identifiable points between two adjacent strips, and using DIGMOS, move one strip (the slave) to match the other strip (the master) in the mosaic. The following leads the user through the procedure used by method 1:
  1. Display the compressed mosaics using VISTA. Select one strip in the mosaic that will not be moved, this will be the master image. Select one of its adjacent strips, this will be the slave image. Vista's POSITION dialog will be needed to get the line/sample coordinates of the 3 to 5 strip-to-strip control points (or more) needed to adjust the slave image.

  2. While the odd and even mosaics are displayed zoom in to 1x or 2x so that a common feature on both images can be identified. Once a feature can be identified on both strips, write down the line/sample coordinates from both images. Calculate and write down the differences in the line and sample directions. The user should repeat this step until an acceptable number of control points for this pair of images have been selected.

  3. Once all the differences have been calculated, compute an average line and sample difference. These difference indicate how much the slave image should be moved in order to minimize/improve the geometric mismatch.

  4. Find the uppermost line and the leftmost sample pixel of the slave image within the mosaic and write them down. The user now has all the information needed to move the slave image.

  5. Using DIGMOS create a new mosaic containing only the master image. Using MAKEIM create an image that has the same number of lines and samples as the new mosaic. Copy the label file of the new mosaic to this image. NOTE: The files created here can be "added to" when generating a mosaic containing multiple strips; MAKEIM is used only with the first pair of strips for the given mosaic.

  6. Using DIGMOS, manually insert/drop the slave image into the image created by MAKEIM. Use the default values for ISL,ISS,INL,INS. The values for OSL and OSS are calculated as follows:

    1. Add the average Line difference (step 3) to the uppermost slave line in the mosaic (step 4). This will be OSL value used in DIGMOS.

    2. Add the average Sample difference (step 3) to the leftmost slave sample in the mosaic (step 4). This will be OSS value used in DIGMOS.

  7. Display these new images to see if the results are acceptable. The mismatch should be improved between the two strips. If there are still areas that are unacceptable, they can be corrected with more points or stenciling. If the mismatch is variable throughout the strips and still unacceptable, then the strip-to-strip method below should be used instead because it can perform non-linear correction.

  8. The slave image now becomes the master for its adjacent image. Display the new mosaic (one with the slave moved) and the mosaic with its adjacent strip. Repeat steps 2-7 until all the necessary strips have been processed using this procedure.

  9. Once the mismatches have been improved and are acceptable for all the sonar strips, the final mosaic can be generated. Remember, the differences calculated will be 2x to 4x larger when generating the final mosaic due to the higher pixel resolution. The calculated OSL and OSS values will also need to be adjusted for the 2x to 4x compression.


Method 2: STRIP-TO-STRIP REGISTRATION

This method also uses strip-to-strip control points to improve the geometric mismatch, but it adapts for differences between strips using a non-linear line/sample offset at different strip locations. This method uses the thermal printer to identify the strip-to-strip control points in the DELTAV image results. Once strip-to-strip control points have been chosen, the program TUNENAV is used to manipulate the navigation file of the slave image to match the master. The following leads the user through the procedure used by method 2:

  1. As stated earlier, create the odd and even mosaics at a compressed resolution. Using blink, visually inspect them to see where mismatching occurs. Select one strip in the mosaic that will not be manipulated, this will be the master image. Select one of its adjacent strips, this will be the slave image.

  2. Print the DELTAV image results of these two sonar strips using the thermal printer. Each strip should be printed using the SAME compression factor; the compression factor should be selected such that it fits on a single page, length does not matter. NOTE: Once a strip has been printed, it does not need to be printed again (save it for later use with adjacent strips).

  3. Find an area where the strips can be laid out. Overlap the slave strip on top of the master. The white edge of the thermal print should be cut off to help visualize the overlap better. Try to identify common features on each strip. Once a feature has been found on both prints, mark the feature on the thermal prints; repeat this step identifying 3 to 5 points. It is best to spread the points out as uniformly as possible and over the majority of the strip length, if possible.

  4. Once an acceptable number of features have been chosen on the thermal prints use VISTA to display the DELTAV image results; the master image on the left half and slave image on the right half of the screen. Steps 5-7 are used to identify line/sample coordinates that will be used to modify the slave navigation file to match the master's navigation file. Once both images have been displayed, select Zoom from the Options pull down menu in VISTA. Also select Position from the Show pull down menu. Move these dialog boxes from obstructing the view of the sonar images.

  5. From the Zoom Options box, turn off the Right zoom -- click the button next to Right/Bottom zoom. VISTA will now only zoom in/out on the LEFT half of the screen. Make sure the Left/Top zoom button is pushed in. On the screen find one of the features identified on the thermal print of the master.

  6. Once the feature from the master is displayed, turn off the Left zoom -- click the button next to Left/Top zoom. Turn on the Right zoom. VISTA will now only zoom in/out on the RIGHT half of the screen. Find the same feature on the slave image.

  7. The same feature is now displayed from both images. Using the mouse and Position box, get the line/sample coordinates where the feature appears to be the same. Write down the line/sample for both images. The user should repeat steps 5-7 until line/sample coordinates for all features have been obtained.

  8. After selecting the strip-to-strip control points, create a control point file, with a text editor, that will be used as input to the program TUNENAV. The format for the control point file is as follows:



    point slave slave master master
    number, line, sample, line, sample


    For example, the users control point file may look like this:



    1,101,235,201,1521
    2,3036,175,3351,1457
    3,5782,117,6421,1652


    In this example point 2 indicates that slave line 3036 sample 175 will be mapped to master line 3351 sample 1457.

  9. Run the program TUNENAV. This program allows the navigation information of the slave image to be manipulated based on the master image, using the strip-to-strip control points selected in the steps above. TUNENAV uses the latitude/longitude values of the master image at each control point to calculate new lat/longs for the slave image. It interpolates between control points to generate new latitude/longitude values for each line. It generates a new image file equivalent to the DELTAV image, but with a modified navigation file.

    The navigation information between any two strip-to-strip control points is calculated using a linear interpolation between points. The navigation information above the first control point and below the last control point, can either be recalculated based on the closest point (default) or extrapolated based on the two nearest control points. Most often the default works best; however, any combination of interpolation/extrapolation for the top and bottom can be used, and depending on the given data one procedure may be better then the other at different times.

  10. Use the DELTAV file TUNENAV generated as input to SNRGEOM. Compress the pixel resolution for this file the same as before.

  11. Using DIGMOS, create a new mosaic containing only the master image. Create another mosaic containing only the corrected slave image. NOTE: As in method 1, the files created here can be "added to" when generating a mosaic containing multiple sonar strips.

  12. Display the new images to see if the results are acceptable. The mismatches should now be improved between the two strips. If there are still unacceptable areas, select more control points and/or do interactive stenciling to minimize the remaining problems.

  13. The slave image now becomes the master for its adjacent image. Display the new mosaic (one with the slave "corrected") and the mosaic with its adjacent strip. Repeat steps 2-12 until all strips have been corrected.

  14. Once mismatching has been minimized to an acceptable level for all sonar strips, the final mosaic can be generated at the desired pixel resolution.


Back to USGS MIPS Procedures

Authors: Miguel Velasco, Jo-Ann Bowell, Stuart Sides and Pat S. Chavez Jr.

U.S. Department of the Interior | U.S. Geological Survey | Jobs
Home | Search | Disclaimers & Privacy Web Rings
Send comments to webman@TerraWeb.wr.usgs.gov
URL:
Last Modified: Thu Oct 31 15:08:22 MST 2002
FirstGov: Your First Click to the U.S. Gov ernment