[Talk-GB] OS 7th series, accuracy update, request for calibration help
Thanks for the offers of help. The task is a very "on again, off again" affair so expect quiet times and busy times! (Depending on availability of new scans.) The steps are as follows: 1) Load map into gimp or similar. The images are 17000 by 21000 pixels which can be a strain on computer memory. 2) Resize to 50% of orginal. This seems sufficient and makes later stages faster to process. 3) Save as jpg 4) Note 9 calibration points that are appoximately evenly placed in a 3x3 grid. The map corners are used as calibration points. Each point has a map reference and pixel position. 5) Save that information in a text file as specified here: http://wiki.openstreetmap.org/wiki/Warp-gbos#Calibration_of_Sheets 6) To check the calibration, the calibration file can be used with the warp-gbos tool to create a KML file. The warp-gbos tool needs to be compiled from source. Details at the above link. 7) This can be compared to google-earth as a sanity check. 8) Send me the calibration file :) I will send people who have expressed an interest a link to download a map sheet to play with. Regards, Tim ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] OS 7th series, accuracy update, request for calibration help
Jerry Clough - OSM wrote: > I'm right there may also be errors on Arran, which also only had 2 10km > squares published. On the other hand, Arran may have been much better > surveyed in the 19th century because of its geological interest. I have a "Crown Copyright 1972" 1:5 geological map of Arran that claims to be: "Geologically surveyed by W. Gunn and A Geikie, B. N. Peach, District Geologist. Published in One Inch Sheets 21(1901) & 13(1902). Colour-printed Edition, 1910." (followed by republish and reprint info). ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] OS 7th series, accuracy update, request for calibration help
I wouldn't be surprised if Kintyre other than NR71 & NR72 (Campbeltown) is derived from much earlier mapping. Those two octads are the only ones published for Kintyre in the 25k provisional/first series. Certainly for other parts of highland Scotland serious errors existed until the 50k 2nd series (one reason why this appeared for the Highlands first). If I'm right there may also be errors on Arran, which also only had 2 10km squares published. On the other hand, Arran may have been much better surveyed in the 19th century because of its geological interest. Jerry Clough SK53 From: TimSC To: talk-gb@openstreetmap.org Sent: Sunday, 18 October, 2009 16:11:08 Subject: [Talk-GB] OS 7th series, accuracy update, request for calibration help Hi all, I have updated the 7th series layer with improved accuracy. It seems as if the Helmert approximation of the OS grid reference to WGS84 was not accurate enough; I was getting transform errors of approximately 20m. Also, to my surprise the next biggest error seems to be surveying errors. For example, see around the Mull of Kintyre where the survey errors are around 40m (near NR730100). http://www.openstreetmap.org/?lat=55.3339&lon=-5.5737&zoom=14&layers=B000FTF ..snip Tim ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb Send instant messages to your online friends http://uk.messenger.yahoo.com ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
Re: [Talk-GB] OS 7th series, accuracy update, request for calibration help
Tim, What's the scan file size? Will happily help look at calibration as I'm doing it all the time for the 1:25k so set up to do it easily. If I know the file size I can try a similar size on my box here. How many calibration points are you needing per sheet to do a reasonable job? Cheers Andy >-Original Message- >From: talk-gb-boun...@openstreetmap.org [mailto:talk-gb- >boun...@openstreetmap.org] On Behalf Of TimSC >Sent: 18 October 2009 4:11 PM >To: talk-gb@openstreetmap.org >Subject: [Talk-GB] OS 7th series, accuracy update,request for calibration >help > >Hi all, > >I have updated the 7th series layer with improved accuracy. It seems as >if the Helmert approximation of the OS grid reference to WGS84 was not >accurate enough; I was getting transform errors of approximately 20m. >Also, to my surprise the next biggest error seems to be surveying >errors. For example, see around the Mull of Kintyre where the survey >errors are around 40m (near NR730100). > >http://www.openstreetmap.org/?lat=55.3339&lon=- >5.5737&zoom=14&layers=B000FTF > >The best way of determining the error in rectification is to get the lat >lon of a grid square corner. stick that into the OSTN02 transformation >(http://gps.ordnancesurvey.co.uk/convert.asp). The true corner position >can be found by rounding the resultant grid coordinates to the nearest >thousand. (Assuming the error is less than 500m!) Rectification errors >seem to be in the order of 0 to 20m depending on the area. If any areas >have errors that exceed this, let me know and I will investigate it. > >The next steps are to rectify the other sheets we have scanned. >http://wiki.openstreetmap.org/wiki/7th_Series > >Is there anyone who can help us with sheet calibration? Having a linux >box with a decent amount of RAM is a requirement. If yes, get in touch >with me and Steve Chilton. > >Tim > > >___ >Talk-GB mailing list >Talk-GB@openstreetmap.org >http://lists.openstreetmap.org/listinfo/talk-gb ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
[Talk-GB] OS 7th series, accuracy update, request for calibration help
Hi all, I have updated the 7th series layer with improved accuracy. It seems as if the Helmert approximation of the OS grid reference to WGS84 was not accurate enough; I was getting transform errors of approximately 20m. Also, to my surprise the next biggest error seems to be surveying errors. For example, see around the Mull of Kintyre where the survey errors are around 40m (near NR730100). http://www.openstreetmap.org/?lat=55.3339&lon=-5.5737&zoom=14&layers=B000FTF The best way of determining the error in rectification is to get the lat lon of a grid square corner. stick that into the OSTN02 transformation (http://gps.ordnancesurvey.co.uk/convert.asp). The true corner position can be found by rounding the resultant grid coordinates to the nearest thousand. (Assuming the error is less than 500m!) Rectification errors seem to be in the order of 0 to 20m depending on the area. If any areas have errors that exceed this, let me know and I will investigate it. The next steps are to rectify the other sheets we have scanned. http://wiki.openstreetmap.org/wiki/7th_Series Is there anyone who can help us with sheet calibration? Having a linux box with a decent amount of RAM is a requirement. If yes, get in touch with me and Steve Chilton. Tim ___ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
