Support Website Contact Support Blog

Reoptimize issue with merged Micasense Rededge projects (only one band processed)

Hi: 

I wonder any one see the problem below?

I merged three Micasense Rededge projects from three different, adjacent flights. All appear good after merging. However, After marking 6 ground check points (each project includes 3 out of the 6 GCPs), I reoptimize the merged project. However, it looks only one band is normal in results. All other bands become fragments and there are a large amount of uncalibrated images.

This is hard to understand since if the procedure works for one band, it is supposed to work for other bands.

The problem does not happen with RGB and thermal projects I did.

I recognize that the 5 bands in Rededge are actually treated as rig cameras. Can this issue be related to this? Do I need to change any camera rig setting?

Thank you very much.

Stephen

Sorry, some update. The images below are what before reoptimization.

Hi Chengyuan Xu, 

This might be related to how you marked the tie points. Please make sure that you mark the tie points in images from all bands. You can check this with the name of the images that you mark. 

Also, if you merge projects, I would recommend to add tie points in the subprojects before merging. This enables you to reoptimize the subprojects separately and then merge them with the tie points already marked, i.e. no need to reoptimize the merged project separately. 

Let us know how this went :slight_smile:

Hi Pierangelo:

I checked the rig camera setting and found a problem that may be the key reason to cause this issue.

When merging three Rededge projects, Pix4D treat the band from each project separately and set total 15 bands (5 from each project) marked from 1-15. 

The 15 bands are bulked into three rig cameras, one for each project.

Normally for Rededge, Green band is set as the reference to provide the camera position information to overlap the images of all 5 bands.

In this case, each rig camera should use its own green band. It is correct for the first rig camera shown in the above two images. However, Pix4D appears to confuse the green band from different projects and used one (marked as 2) for all three rig cameras.


In this case, green band 2 is used as the reference for all three projects as reference while green bands 7 and 12 were set as new cameras and were not used as the reference of their own projects.

In this case, because most images in green band 2 is different from those in band 7 and 12, using that as a reference would screw up. I suspect the only processed band is green band but not 100% sure.

I tried to reset rig camera but cannot find out how to choose green band 7 and 12 to substitute green band 2 in the two projects.

Could you please advise whether edit rig camera bands is possible? If Pix4D automatically take green band of the first project as the reference camera in rig setting and does not allow to change, that is clearly a bug to be fixed.

Thank you very much.

Stephen

Hi Stephen,

Indeed this mix of bands is a bit confusing and it should not happen. After merging what you should see is 3 camera rigs with 5 bands per rig.

From your description, it seems that this was after you add the GCP’s therefor, I would suggest you as Pierangelo said to add and mark your GCP’s individually in each project before merging. When you mark you need to make sure that you mark it in all corresponding bands. 

What I believe that it could have happened here is that you might have marked the GCP,s only in green bands of each project which could lead to this problem of identifying one green band as the master of all three rigs. 

That being said can you give it a try and add the GCP’s/MTP’s individually in each subproject before merging, then merge all of them and see if it solved the problem? 

To double check, the merging workflow take a lookhere.

Additionally, could you check if you Duplicate camera parameters option is activated?

Let me know how it worked?

Cheers,

Ina

 

Hi Ina and Pierangelo:

I have done what suggested by Pierangelo. I marked GCPs in all sub-projects and then merged them, without reoptimization. ‘Duplicate camera parameters’ is the default option and I did not change it. When marking GCPs, I marked all bands. Actually, after marking one bank and click automatic marking, the system automatically mark other band.

The  mix of bands happens from the beginning of the merge of projects, before marking any tie points. The screenshot below is taken when the project was just merged and the initial process was generating.

It seems the software does not properly recognize the bands or set the reference band of the first project in default, although it can be fixed if I can edit rig camera settings.

Cheers.

Stephen

 

Hi,

Is this happening also when you try to merge first 2 projects and then the resulting one with the third?

Cheers,

Ina

Hi Ina:

Yes. I tried to merge two project first and the bands are still screwed up. See Image Properties Editor below. Thus I think the software may have some issue to merge Micasense Rededge projects. Any other suggestions?

Thank you very much.

Stephen

Chengyuan Xu, 

 

Could you please send me two samples of these two projects that you have tried to merge so I could test it from my side? The samples should preferably be from near the common area so I can merge them as well.

 

Thanks,

Ina 

Hi Chengyuan,

I have checked again and if you uncheck duplicate camera parameters the bands do not get mixed. Could you try it from your side?

For now, we will do further tests to see why the green band from one rig is considered as global master with duplicate camera parameters enabled. Soon I will have updates I will let you know.

 

Cheers,

Ina

Hi Ina:

I did another try. If uncheck duplicate camera parameters, it seems the program bulk images of all sub-projects together, respectively for each band. It can be processed. 

However, the sub-projects usually have different images of reflectance standard panel for radiometric processing and calibration, which are taken at different time. When merged with this approach, it only allows to use one standard panel image for processing and calibration (see image below), which is not ideal. Therefore, if the green band rig issue can be worked out, it would be very helpful.

Thank you very much.

Stephen

 

Chengyuan Xu,

With merged multispectral images we support only one calibration target: eg assuming that you have a large dataset from one flight you could split it and process individually for processing time reasons.

Our developers are currently looking into the issue with the green band being the global master band but even though in that case you can put individually radiometric target the reflectance maps will not be merged; you will get 10 maps. Hence this procedure would not be of your interest because you will get no extra benefit compared with processing projects individually. 

 

Cheers,

Ina

 

 

Hi Ina:

I further checked the calibration target. Actually when projects are merged with duplicate camera parameters box ticked, it generates  and seems to support multiple calibration targets, one for each project (see screenshot below). That looks all ok. 

Thus, it seems the issue is just that the green band is set as the global master when projects are merged. I agree that technically it is similar to separately processing three project. However, it is the approach that keeps the most complete information for all projects when multiple calibration targets are merged. In my case, the problem is that the customer wants one image for the whole farm with high resolution, so I have to merge data from three flights, which appear too large when processed together. Anyway, I greatly appreciate the support and hopefully to see green band global mastering issue will be solved.

By the way, does the calibration target information within each projects get lost and needs to be re-entered after merging projects?

Thank you very much.

Stephen