[HOT] Bing imagery offset

Severin Menard severin.menard at gmail.com
Tue Nov 18 21:59:35 UTC 2014


Hi Althio,

Thanks for pointing this out. Actually I had made the changes but seems my
not really reliable internet did not load my last edits for the
instructions.
Actually I had already moved the data when I noticed the Bing imagery
change, and it is not an easy operation, at all. I did this to maintain the
whole quite consolidated, otherwise it would be changed here and there
through the TM tasks with likely streets becoming snaky. Supposedly what
remains to be moved is only a few meters and can be done through the tasks
by regular mappers. But a validation process on the road network once the
job will be done will not hurt, I guess.

Sincerely,

Severin

On Tue, Nov 18, 2014 at 9:02 PM, althio forum <althio.forum at gmail.com>
wrote:

> Hi,
>
> Offset is always a pain with different or updated imageries... Or so it
> seems.
>
> The new workflow is certainly better and more streamlined in the current
> conditions.
>
> A suggestion though regarding imagery/offset instructions:
> " Align the WordView-2 imagery on the vector data (and not the contrary).
> Using the offset_db plug-in will help you to get offsets points that will
> likely almost aligned completely the imagery (having to solve a few meters
> offset is easy, where it is several tens, it takes more time to figure out
> where this building is supposed to be). "
> Since a "rectification process was performed to reduce offset" on
> WordView-2 and since any offset from database is outdated in this area:
> this imagery/offset instructions could be simplified and reduced to the
> first sentence only " Align the WordView-2 imagery on the vector data (and
> not the contrary). " (And no offset_db plug-in will help you there...
> Unless all old offsets are completely deprecated and new correct ones are
> created.)
>
> Also a question regarding process: how difficult or effective would it be
> to offset globally all data and align on new Bing imagery instead of
> relying on each mapper on his individual tile? Alternatively the other face
> of the question is: how easy is it for a regular mapper editing with iD to
> do a re-alignment for a lot of existing objects?
>
> althio
> On Nov 18, 2014 7:53 PM, "Severin Menard" <severin.menard at gmail.com>
> wrote:
>
>> Hi,
>>
>> I just updated the job <http://tasks.hotosm.org/project/591>
>> instructions as proposed below. SUmmary: first Bing 2013, then WV2 to
>> complete.
>>
>> Sincerely,
>>
>> Severin
>>
>> On Tue, Nov 11, 2014 at 3:48 PM, Severin Menard <severin.menard at gmail.com
>> > wrote:
>>
>>> Hi,
>>>
>>> Yes the adjustments for the latest WV2 imagery, put in the offset db,
>>> were based on the Bing imagery, supposedly (almost no GPS traces over Juba,
>>> unfortunately). BUT Bing updated its imagery over Juba; it is now dated
>>> June 23, 2013 and the georeferencing is slightly different, making the
>>> previous offset_db points obsolete. Tried to deprecate or delete them, did
>>> not get it, if someone knows the tip, I would be interested.
>>>
>>> Considering the two imagery are now quite close, I would suggest to:
>>> - draw first on Bing and consider its georeferecing is good
>>> - then display the WV2 imagery, correcting the offset manually (was more
>>> complicated with the previous Bing imagery that was quite dark and quite
>>> old, what made the offset correction not easy considering the city is
>>> growing/evolving fast), and add the missing parts (from the test I made,
>>> new buildings here and there and more new buildings in the city limits, as
>>> expected).
>>>
>>> It this makes sense for everyone, I edit the job's instructions
>>> accordingly.
>>>
>>> Sincerely,
>>>
>>> Severin
>>>
>>> On Thu, Nov 6, 2014 at 9:51 AM, Sander Deryckere <sanderd17 at gmail.com>
>>> wrote:
>>>
>>>> The reason for this is because relative offsets are much harder to deal
>>>> with than absolute offsets. When the relative positions are correct, you
>>>> can still calculate how far things are from each other, how big they are,
>>>> ... You don't need the absolute position for it. Even on the field, when
>>>> using a GPS, the quality of the GPS position is often so low that you won't
>>>> notice an absolute offset.
>>>>
>>>> To minimise the number of relative differences, it's advised to base
>>>> your data on a source that's more or less good in quality, and has a big
>>>> coverage. As such, the Bing imagery was chosen as a base (without an offset
>>>> applied, as any new mapper would get it in his freshly opened editor).
>>>>
>>>> When there are better sources available in the future (f.e.
>>>> interpolated, high-precision positions), then everything in that region can
>>>> be shifted by the then-known Bing offset.
>>>>
>>>> Regards,
>>>> Sander
>>>>
>>>> 2014-11-06 11:55 GMT+01:00 althio forum <althio.forum at gmail.com>:
>>>>
>>>>> Hi HOT
>>>>>
>>>>> I am on task #591 - South Sudan Crisis, Cholera outbreak in Juba,
>>>>> mapping with WorldView-2 imagery.
>>>>>
>>>>> Instructions includes:
>>>>> Check in the vector data is correctly aligned on Bing imagery. Bing
>>>>> imagery is the reference for the georeferencing.
>>>>>
>>>>> My question is about the first step i.e. What is the recommended
>>>>> offset for Bing imagery:
>>>>>  (a) get from database
>>>>>  (b) set to offset: 0.00; 0.00
>>>>>  (c) other
>>>>>
>>>>> Cheers,
>>>>>
>>>>> althio
>>>>>
>>>>> _______________________________________________
>>>>> HOT mailing list
>>>>> HOT at openstreetmap.org
>>>>> https://lists.openstreetmap.org/listinfo/hot
>>>>>
>>>>>
>>>>
>>>> _______________________________________________
>>>> HOT mailing list
>>>> HOT at openstreetmap.org
>>>> https://lists.openstreetmap.org/listinfo/hot
>>>>
>>>>
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20141118/20f4618c/attachment.html>


More information about the HOT mailing list