Topics

1

General / Missing elevation points in DEM

« on: August 25, 2020, 01:01:42 AM »

Hello!

I generated DEM with dense cloud for multispectral images collected with the RedEdge 3 sensor mounted on a Matrice 100 UAV. The attached images show that the DEM does not have the elevation for few points. It appears as though those places are empty, but actually there are plants at those locations. When I checked the dense cloud model, it did not have the correct elevation values for those points as well.


I have processed several similar datasets, but didn't have this issue with any of them. I am not sure, what could be causing this issue. Can some one suggest how I retrieve the height points in the model?

Thank you,

Vaishali

2

General / DEM Discrepancies

« on: June 25, 2020, 01:49:48 AM »

I have two sets of DEMs from images captured on consecutive days over a cotton field. There were no changes on field, but the dense cloud/DEM have differences between the two days. As you can see from the images, the DEM in the first image does not capture the full canopy due to which the heights are lower than they are in the second DEM. I used ultra high settings to generate the dense clouds for both days. Can someone tell me what could be causing this and how to get more consistent and realistic DEMs? I am interested in determining the height from the DEMs.

3

General / Validating invalid tie points

« on: June 17, 2020, 02:58:48 AM »

Hello again!

I have shared this question as a ticket with the support team, but thought I will share it in the forum as well to get more insight into how this could be done. Here is my problem/question:

I am working with Rededge MX multispectral images that were acquired from a ground based platform. This query is kind of related to my previous post as well. My alignment processes result in broken surfaces that have different orientations. I checked why this was happening, and it seems like the tie points between the images from the two surfaces are marked invalid. Can you let me know what  needs to be done to convert the invalid tie points into valid tie points? I have tried the different accuracy settings and even tried adding manual tie points. Nothing seems to work on my data. The raw images are good quality ones, captured with low oblique angling.

EDIT: I have come across similar posts in the forum, but never really found a reply.

Thank you!

4

General / Distortion after adding GCPs

« on: June 17, 2020, 02:50:52 AM »

Hello all, I need some help to figure out a way to remove distortions in the orthomosaic caused by the GCP addition. As you can see from the attachment, apart from the GCP itself appearing distorted, the plants around the GCP are also distorted.

I tried correcting by selecting few camera that cover this region in the Photos pane and aligning just those cameras selectively. It did not seem to work either. I have positioned the markers as accurately as possible on all the images that contain the GCP. Further, I tested by removing the markers from images that don't show the GCP very clearly.

Any help or suggestion will be greatly appreciated.

5

General / Alignment issuse after adding GCPs

« on: May 29, 2020, 02:30:15 AM »

Hi, when I processed different sets of  multispectral (RedEdge 3) UAV images, I noticed the following issues.

1) On the first set of data (all images collected during the same flight mission), the alignment was initially good. But I noticed that after adding GCPs, a portion of the tie points on the left side appear disconnected from the rest of the surface (Image 1). This gives me totally off DEM values.

2) Ran a second set captured over the same area, on a different dat, and used the same set of GCPs. The tie points appear fine, expect the DEM shows abnormally low values at the bottom low corner (Image 2). Not all the markers that appear on the screenshot were used for correction. The markers with the 'point' prefix were added later to derive something else from the orthomosaic. Another issue with this DEM is the patchy surface on the right side, which has been encircled in Image 3. The actual surface doesn't look anything like this.

3) I have attached the DEM generated from Zenmuse 3 RGB camera for reference (Image 4). This is the closest reconstruction of the original surface. The same GCPs were used, and the markers with prefix 'point' were added later for a different purpose.

I appreciate any help or advice on this issue.