Error increasing after inputting marker accuracy

[MeganG]

Hello!

For context, I'm working with drone data collected along shorelines, and we measured ground control points at our sites. I've been having an issue with some of my sites in which when I add in the Marker accuracy (pix) and Marker accuracy (m) under the reference settings, the Error (m) of my GCPs increases substantially. The Marker accuracy (m) is based on the error of the GCPs and the base station collected in the field. This substantial increase in error is only happening at some of my sites even though the marker accuracy is very similar across sites (~0.025m). Before entering this information, my Error (m) is usually around 0.03, but afterwards, it's about 0.25.

Does anyone have any insights as to why this might be happening? Happy to provide more information if needed!

Thank you,
Megan

[jrp]

By the sounds of this, you are more experianced than I am with GCPs anyway, but incase it's useful:

I've seen loosly similar results when I've had a point with a substantial error in it, have you checked all the input data for anomalies? Best place to start looking would be to sort the GCPs by error, and turn them off one or two at a time starting with the worst ones then refresh, and see if everything pops into place.

I've seen it where the turned off point is then reported to have an error of nearly exactly a meter, and it turns out to be a simple typo. (university, student projects, we usually end up with the points written down with a pencil in a notebook as it's less of a learning curve)

[MeganG]

Unfortunately, that doesn't seem to be the problem. None of the GCPs have high error from the measurements collected in the field, and when I uncheck some in agisoft metashape, I still have very high error. After inputting the marker accuracy in the software, almost all of my GCP errors increase. I tried unchecking about half of them, and I still have an error of 10cm, which for my sites is very high and unusable.

[jrp]

Are you able to post some screenshots of the reference panel in diferent states? (ideally with a good and bad project)

[MeganG]

Absolutely! These are from the same site but different years. I'm happy with the error for 2022, but not for 2023. In 2023 we used a lot more GCPs than 2022, and we used some elevated GCPs. I thought this might be causing an issue with the error, but I've tried processing 2023 without the elevated GCPs, and I still come back with a high error. Also, I've processed some sites where we used elevated GCPs, and the error was acceptable. Both of these screenshots were taken after adding in the marker accuracy in the reference settings.

[Paulo]

Hi MeganG,

i think the problem is that your camera positions are in WGS64 geographic and your GCPs are in UTM but the program thinks the camera positions are in projected UTM (thus immense error in cameras). So you should  load photos as WGS84 then convert to your projection UTM and then import the GCPs reference in same UTM... You should get better results...

[MeganG]

Hi Paulo,

I did change the coordinate system in the reference settings (screenshot attached). Do you know of another way to do this? or another place I should be changing the coordinate system?

Thanks,
Megan

[tutoss]

Hi MeganG,
With the marked button on the attached image, you can choose your desired coordinate system to convert.

Best regards,

tutos

[macsurveyr]

Hi Megan

It appears to me that your error estimates for marker pixel error is probably too loose in 2022 and way too tight in 2023. 2022 marker pixel errors is over 1 pixel which isn’t good. 2023 shows marker pixel errors far far below 0.1 pixel which is way too tight. Is that perhaps a typo that you meant say 0.15 but put in 0.015? Such an error estimate will force any mis marking error into the control values.

Tom

[Paulo]

Hi Paulo,

I did change the coordinate system in the reference settings (screenshot attached). Do you know of another way to do this? or another place I should be changing the coordinate system?

Thanks,
Megan

Hello,

the camera reference should be set to WGS84 as these coordinates are lat, lon, alt with lat lon in degrees.

Also it seems your marker image measurement accuracy is way too tight (0.054 pix) should be around 0.5 pix

[MeganG]

Hi MeganG,
With the marked button on the attached image, you can choose your desired coordinate system to convert.

Best regards,

tutos

Using the "convert" button didn't change anything. I believe it does the same thing as changing the coordinate system in the reference settings. Thanks for the tip though! I didn't know the button existed.

[MeganG]

Hi Megan

It appears to me that your error estimates for marker pixel error is probably too loose in 2022 and way too tight in 2023. 2022 marker pixel errors is over 1 pixel which isn’t good. 2023 shows marker pixel errors far far below 0.1 pixel which is way too tight. Is that perhaps a typo that you meant say 0.15 but put in 0.015? Such an error estimate will force any mis marking error into the control values.

Tom

Thanks for the tip! I'll try resetting GCP locations to get a new pixel error and report back on the results.

[MeganG]

Hi Paulo,

I did change the coordinate system in the reference settings (screenshot attached). Do you know of another way to do this? or another place I should be changing the coordinate system?

Thanks,
Megan

Hello,

the camera reference should be set to WGS84 as these coordinates are lat, lon, alt with lat lon in degrees.

Also it seems your marker image measurement accuracy is way too tight (0.054 pix) should be around 0.5 pix

Changing the reference setting didn't do anything for the error. I believe because my cameras are not checked off, this is not what's affecting my error. But thank you for the tip about the pixel error! I'll try reprocessing and report back.

Thanks,
Megan

[MeganG]

Hi Paulo,

I did change the coordinate system in the reference settings (screenshot attached). Do you know of another way to do this? or another place I should be changing the coordinate system?

Thanks,
Megan

Hello,

the camera reference should be set to WGS84 as these coordinates are lat, lon, alt with lat lon in degrees.

Also it seems your marker image measurement accuracy is way too tight (0.054 pix) should be around 0.5 pix

Hi Megan

It appears to me that your error estimates for marker pixel error is probably too loose in 2022 and way too tight in 2023. 2022 marker pixel errors is over 1 pixel which isn’t good. 2023 shows marker pixel errors far far below 0.1 pixel which is way too tight. Is that perhaps a typo that you meant say 0.15 but put in 0.015? Such an error estimate will force any mis marking error into the control values.

Tom

Hi all,

I reprocessed 2023 and ended up with a pixel error of 0.559. My error is still very high (see screenshots attached). I also tried changing the coordinate system, and this isn't changing my error either.

Thanks,
Megan