Visible stripes in the orthomosaic and DTM along the flight direction (Ebee X + Aeria X)

PIX4Dmapper
1

Hi community,

I’m experiencing a recurring issue when processing my images with Pix4Dmapper. After generating the orthomosaic and the DTM, I notice visible stripes along the flight lines (in the direction of the flight). These stripes affect the visual quality of the final product and also make it harder to carry out certain analyses based on the data.

In the DTM, the problem is even more critical:
there are noticeable elevation differences between the stripes, which creates an artificial pattern on an otherwise flat agricultural field.

Project details

  • Drone: Ebee X
  • Camera: Aeria X
  • Frontal / lateral overlap: 75% / 75%
  • Terrain type: Agricultural, very flat
  • Software: Pix4Dmapper
  • Flight time: 12:00 PM (stable lighting conditions)

I’ve tried adjusting parameters during the calibration step and DSM generation, but the issue remains.

Has anyone experienced this same effect?
Do you have any recommendations on how to avoid or correct these stripes, both in the orthomosaic and especially in the DTM?

I suspect this might be related to soil reflectance affecting the camera sensor. Has anyone found a way to mitigate this?
Maybe some kind of image calibration or preprocessing?

Thanks in advance for any help!

Pix4DProblem3
Pix4DProblem31456×645 113 KB

2

Hi maicon,

Striping is often due to insufficient side or lateral overlap. However, can you send us your PDF quality report? We can look it over to see if there are any calibration issues.

3

Quality Report.pdf (3.2 MB)

4

Hello @maicon, As Mike said, usually visual stripe are the result of the low overlap or issue on the images. I would suggest the following.

  1. Image overlap of 80%.
  2. Use Ag RGB (Alternative Calibration) Template for process the flat land.
5

Thank you for your reply. However, I still have some doubts regarding the Ag RGB template:
Is the alternative calibration method actually more accurate than Precise Geolocation and Orientation?
I also noticed that the resulting point cloud has a low density and was generated at 1/4 scale, which theoretically represents lower-quality parameters than the ones I used.

Finally, I would like to point out that an 80% x 80% overlap produces a massive amount of images.