Support Website Contact Support Blog

Using street view images to create a point cloud

Hi all. I am relatively new to pix4D. I have access to a 360-degree camera set up with 6 individual cameras to create a street view 360 type image. I was wondering if anyone has experienced or tried to limit the distortion on the individual photographs enough to create an accurate point cloud based off of just the street view cameras images. If anyone has any thoughts or suggestions on what I could try please let me know! thank you.

 

Hi Harriet,

I have to admit that from our end, we have never tried to test a similar setup. However, it was a good idea of yours to post here on the community. Maybe, other users will share some good insights from their experience. 

Nevertheless, the tip I can share with you is to learn how to remove the sky from the images. I think you find our article How to remove the Sky from the Point Cloud using the Annotation Tool very useful while processing your dataset. 

Could you tell us also more about your camera/s? 

Best

I am using Applied Streetview original 360 camera (http://www.applied-streetview.com/original-360-camera/)

It is a series of 6 fisheye lenses. As far as sky goes that will be a challenge as well as geolocating the exact location of the car and camera. Will update how it goes, I should begin processing within the next few days!

Hi Harriet,

Thanks for the details.

In the case of a camera with many lenses, you can initiate your project in Pix4Dmapper in different two ways.

The first would be by defining a rig - indicating to the software that there are six lenses with defined parameters. However, to do it, you would need to have a rig add-on for Pix4Dmapper. Please contact our Sales Department if interested. In the rig, there is an establishment between the reference camera and other cameras (distances and angles from the reference camera to other cameras which the camera manufacturer should know.)

The second possibility is to process all the images together without defining the cameras separately. In this example, however, you have to check if the camera parameters that Pix4Dmapper reads from EXIF are correct. If not, you have to redefine them. Note that the internal parameters should be accurate in order to start the processing without errors since the beginning. 

It is hard to say with which method will work better with that camera. That’s something to be tested. 

Please keep me updated about the results you get. If you will have some problems with that project. I would highly advise you to submit a ticket to our Pix4D Support Group. It would improve our communication and data transfer. 

Best,