Question

Identifying Image Mistatches

  • 23 April 2018
  • 9 replies
  • 10 views
J

I have several county-wide 4 band GeoTiff imagery datasets that I need to review. Sometimes I identify the exact same image in two different tiles. I would like to automate a method for identifying this error. Does anyone have a workflow for identifying this or have any ideas on moving forward?

9 replies

danilo_fme
Rank
Userlevel 4
Badge +30

Hi @joeyb2018

There are so many transformers for Raster files.

The transformer RasterPropertyExtractor can to help you extracting the properties of a raster file RasterPropertyExtractor

 

After you can use the transformer Tester to validate the informations between Source datas.

Thanks,

Danilo

 

J

Hi @joeyb2018

There are so many transformers for Raster files.

The transformer RasterPropertyExtractor can to help you extracting the properties of a raster file RasterPropertyExtractor

 

After you can use the transformer Tester to validate the informations between Source datas.

Thanks,

Danilo

 

Thanks @danilo_fme,

 

I think I should clarify the question a bit. So in this case, the imagery is being applied to the wrong image tile. This often times creates duplicate images within the entire imagery set for the county. As such, I am not sure which raster properties would be useful in making a comparison. Is there a method that could compare histograms for the entire imagery dataset, assuming that each unique image will have a unique histogram?

 

Thanks,

 

Joseph

 

J

Hi @joeyb2018

There are so many transformers for Raster files.

The transformer RasterPropertyExtractor can to help you extracting the properties of a raster file RasterPropertyExtractor

 

After you can use the transformer Tester to validate the informations between Source datas.

Thanks,

Danilo

 

@danilo_fme ,

 

 

I forgot to mention that when the imagery is also taking on the same name as the tile. So same image, two tiles, two names.
danilo_fme
Rank
Userlevel 4
Badge +30
@danilo_fme ,

 

 

I forgot to mention that when the imagery is also taking on the same name as the tile. So same image, two tiles, two names.
Hi @joeyb2018

 

 

 

Please, see the custom transformer RasterStatisticCalculator.

 

 

jdh
Rank
Badge +22

You basically need to compare each image to all the others (irrespective of name) to make sure that aren't duplicates. I wonder if doing checksum comparisons might work here. See the CRCCalculator.

Alternatively, send them to the Matcher, though that might be considerably slower.

EDIT: If the tiles have embedded georeferencing, they won't be detected as matches, even if they are pixel-wise identical, you'll want to strip that before either transformer. (RasterGeoreferencer Upper Left set to 0, spacing set to 1)

danilo_fme
Rank
Userlevel 4
Badge +30

You basically need to compare each image to all the others (irrespective of name) to make sure that aren't duplicates. I wonder if doing checksum comparisons might work here. See the CRCCalculator.

Alternatively, send them to the Matcher, though that might be considerably slower.

EDIT: If the tiles have embedded georeferencing, they won't be detected as matches, even if they are pixel-wise identical, you'll want to strip that before either transformer. (RasterGeoreferencer Upper Left set to 0, spacing set to 1)

Do you want to compare name of files rasters?

 

 

jdh
Rank
Badge +22
Do you want to compare name of files rasters?

 

 

I suspect the original poster is finding that file R1C1 and file R2C1 are the same image, and is looking for an automated way of detecting that.

 

J
I suspect the original poster is finding that file R1C1 and file R2C1 are the same image, and is looking for an automated way of detecting that.

 

That is correct. We sometimes find the same raster image in two or more locations in an image dataset of roughly 1000 images. This can be difficult to find manually and embarrassing to miss..:). Thank you both for your help, I am testing a couple of your ideas now. If you think of anything else, please let me know. I will make sure to let you know if I happen to solve this one in the meantime.

 

 

jdh
Rank
Badge +22
That is correct. We sometimes find the same raster image in two or more locations in an image dataset of roughly 1000 images. This can be difficult to find manually and embarrassing to miss..:). Thank you both for your help, I am testing a couple of your ideas now. If you think of anything else, please let me know. I will make sure to let you know if I happen to solve this one in the meantime.

 

 

@danilo_fme idea of the RasterPropertyExtractor could be a good way of filtering out obvious mismatches. If two tiles have a different number of bands, rows and columns, then they aren't identical, and don't need further testing.

 

 

This doesn't help of your tiles are in some sort of slippy map and have all the same properties.

 

Reply


Community Stats

28,670
Posts
109,365
Replies
36,531
Members
Terms & ConditionsCookie settings
A product of
Safe Software respectfully acknowledges that we live, learn and work on the traditional and unceded territories of the Kwantlen, Katzie, and Semiahmoo First Nations.