Petascope ingestion of non-geo-referenced images

[Alex Dumitru]

When importing this GML (​http://pastebin.com/RfhmM4T1) I get a coverage that starts at (1,1), although the origin is set at (0.5, 0.5). Is this correct behavior? I am assuming some rounding is performed somewhere for Index2D CRSes.

[Vlad Merticariu]

Can you attach the coverage description after ingestion?

[Bang Pham Huu]

I think the reason is based on this function and also want to know why need to do this? Please answer if you have time.

  def get_origin(self):
	        """
	        Returns the origin of the dataset. This is calculated using the origin from the dataset + 0.5 of an offset
	                vector, as petascope requires it so.
	        :rtype: list[str]
	        """
	        geo = self.gdal_dataset.GetGeoTransform()
	        # Add 0.5 * size of pixel to make sure there's no error in petascope
	        # Should be removed once petascope handles numeric computations correctly
	        return str(geo[0] + 0.5 * self.gdal_dataset.GetGeoTransform()[1]), \
	               str(geo[3] + 0.5 * self.gdal_dataset.GetGeoTransform()[5])

[Alex Dumitru]

@Bang: The origin in petascope is considered to be in the middle of the geopixel, that is top-left corner + 0.5 offset vector. The whole geo-model is described here ​http://www.rasdaman.org/wiki/PetascopeUserGuide
The comment delimited by # is not correct, only the one between """ """ is.

This is a bug in petascope.wcst, not in WCSTImport, the created coverage is correct.

[Bang Pham Huu]

Thanks Alex for your time I will read the article, I also would like to add the file as Vlad requested.

[Alex Dumitru]

Please leave it assigned to Vlad. This is more of a standards decision than actual implementation.
@Vlad Should the origin be in the middle of the pixel for non-georef grids as well?

[Bang Pham Huu]

​http://rasdaman.org/wiki/PetascopeSubsets I update this ticket with the correct document why it is "of half-pixel on all sides" (or + 0.5).

By the way, in here, you can see that the origin is chosen on top-left

 As often done in GIS applications, the origin of an image is set to be its upper-left corner: 
this finally means that the origin of our rectified and referenceable grid coverages shall be there too in order to provide a coherent GML/GMLCOV coverage. 
Note that placing the origin in the upper-left corner of an image means that the offset vector along the northing axis will point South, hence will have negative norm (in case the direction of the CRS axis points North!).

[Bang Pham Huu]

Replying to mdumitru:

Please leave it assigned to Vlad. This is more of a standards decision than actual implementation.
@Vlad Should the origin be in the middle of the pixel for non-georef grids as well?

Acording to ​http://rasdaman.org/wiki/PetascopeSubsets

regular axis: when a grid axis has equal spacing between each of its points, 
then it is assumed that the sample space of the points is equal to this spacing
(resolution) and that the grid points are in the middle of this interval; 

So the answer for Alex is yes as non-georeference grid is a 100% regular axis.

[Bang Pham Huu]

with the patch for using wcst_import to import all 3 types of coverage in system test, the origin of these RectifiedCoverage (mr, rgb) is set correct when GetCoverage ​http://pastebin.com/wMUWj8pW, so I think we can close this ticket here.