Seamline problems with mosiac dataset

2417
1
04-12-2012 10:44 AM
grahamlangley-edwards
New Contributor II
Hi, I have a couple of Landsat 5 tiles from different dates that I am trying to mosaic together with the Create Mosaic Dataset tool.  The problem is that the tile that is displayed on top, where the two tiles overlap, changes for some reason.  It seems that the two tiles are not "sewn" together at a point where they overlap as I thought they would be, but are just displayed one on top of the other.  The one that is displayed on top changes, apparently depending on which tile is taking up the most space in arcmap at the moment.  This creates a problem because there are clouds in one of the tiles and none in the other, so the images seems to shift.  I have attached two screen shost showing the intersection of the two tiles.  The first one, North_Tile, shows the image when the cloudy northern tile is displayed on top.  The second, South_Tile, shows the cloud free southern tile displayed on top.  In the left corner where they intersect you can see a cloud in the first tile and no cloud in the second tile.  

Is there a way to get the tiles to fuse together into one image with the north tile on top?

Also, I would like to create a more seamless image.  The seam is rather obvious and contains black pixels, shown in the Seamline screen shot attached.  I have already color corrected, the Default Mosaic Method is Seamlines, the Default Mosaic Operator is Max.  

So, is there a way to create a more seamless image?

Some info on the images: 

Pixel type: unsigned integer
Pixel Depth: 8 bit
Format: AMD
Number of Bands: 7

Any help is greatly appreciated...

Regards,

Graham
0 Kudos
1 Reply
grahamlangley-edwards
New Contributor II
OK, so I just figured out that you can select lock raster in the Image properties and the images will be locked in place.  That solves question 1, but  I am still at a loss as to how to get rid of the black line where the two images meet up.  Any help on that is still needed and greatly appreciated.

Thanks,

Graham
0 Kudos