Too many Components in Aerial image Alignment

I took images using drone from a height of 8 meter and the both side overlap was 85%. But after the alignment process I found too many components and it divided into several parts. But I want to make a single model from all the images together. Also the model quality is not up to the mark. (see attached screenshot)

So is there any way to get a better solution. I tried all the ways described in this link. But result didn’t improved. https://support.capturingreality.com/hc/en-us/articles/115001485012-How-to-put-together-more-components-

Can anyone have a solution or suggestions.

Thanks in advance.

 

From what I can see on the screenshot the flight pattern is not very regular. My guess is that the camera angles are too big for RC to find matches. There are many threads with similar problems - did you look into those as well?

I used Sony alpha 5100, 24.3mp camera with 16mm lenses with an image overlap of 85%. The camera was in Nadir direction, just under the UAV and flight height was 8m from ground. What do you mean by camera angle? Can you please explain a little more. I also checked some threads but still have same problem. 

Ah, ok. If it was vertically down then never mind. In your screenshot it seemed like they were all pointing in different directions.

Is it possible that the images are not sharp?

Can you post your alignment settings?

Here is my alignment setting. Letter I tried to change the value of ‘max. feature per image’ and ‘preselector features’ option. But result didn’t changes much. I used Manual focus to take images from 8 meter height. Please check the attached screenshot and for images, use this link: https://1drv.ms/f/s!AkiMezhLnO0ghBk6yLxYHaX_nK13 

Dear Nasim,

Please consider few recommendations:

  • Make sure all your imagery is sharp all over the image, not just in the center. 

  • Make sure you fly in calm wind conditions. If you experience any wind, the vegetation such as rice field might get blurry in the imagery and RC will have problems in alignment, reconstruction and texturing too. 

  • For vegetation mapping with very high resolution GSD, make 2 flights - one low altitude with 80%-80% overlap as you did (did you ?), another at 50m altitude in order to better match lower altitude dataset. Vegetation is generally one of the problematic patterns for photogrammetry so ensure you have also higher altitude overlap available. 

  • Geotag imagery in EXIF part of your JPG files. This information will provide RC with initial position in space and it can be later used also for rough model measurements and volume/area calculations. Depending on your output requirements, you may export to other SW such as GIS/CAD for further data analysis.

  • In can you require volume calculations or you just want to achieve a very good detail in 3D model from all angles, make sure you capture also oblique imagery with sufficient overlap. Oblique imagery means camera angle is 80-45 degrees towards the ground depending on your preferences and the terrain features.

  • It’s also helpful to capture connecting images between different datasets (flight missions) with sufficient overlap. In this case you are increasing your chance to create one big component from all datasets at the end. 

  • For UAV missions use flight planner apps that will ensure accurate image overlap. Do not use manual mission data capture as in most cases you will not achieve such overlap accuracy. 

Hope it helps.

Best Regards,

Michal

 

Hello,

I just want to add that I can see from your second screenshot that you did not raise the Preselctor features value. It is the number of features that will be used in alignment from the detected ones. The default value is 10000, like in your screenshot. 

Thank you very much Michal Rus for your important suggestions.

Yes, the vegetation would have been my next target - quite difficult.

The image I looked at seemed sharp enough though.