Camera Offset Survey

[Jim Campi]

I now have the Camera Offset Survey Option enabled and have a few questions after reviewing the materials:

There are apparently 5 types of points:

Target – this is the unknown point of interest.

Vantage – this is the location of the camera (LS) when collecting an image. The angle between various Vantage and Target points should be 90 degrees (when possible). Vantage points should be designed in an arc at varying heights. The minimum number of vantage points for an optimal solution is 5.

Tie – this is an unknown point, common to multiple images utilized by the software to link images containing the target point. An optimal solution contains 4-5 tie points common to all images. At least 2 images should have a 90 degree angle between the Vantage and Tie points per Target point.

Control – A known point that should be visible on at least one image. Accuracy might increase if visible on more than one image.

Check – A known point used for accuracy assessment. Check points are not used in adjustment calculations. For an optimum result use 1 -2 check points. The angle between the Vantage and Check points should be 90 degrees.

A few questions:

Can a Target be a Tie point?

Obviously the images must be collected in the field. What portion of this process can be completed in the office? What points must be identified on each image, in the field at the time of survey?

Any insight/thoughts would be helpful.

 

[Shawn Billings]

Everything can be done in the office outside of collecting the images. There is no need to identify any points in the field. Just make sure that you can see points of interest in the photos clearly (known points and tie points).

I'm not sure where you got the different point types. I know of these:
Reference Points: points with known coordinates which can be used for control or as a check
Tie Points: points with unknown coordinates that are visible in at least three photos (when using control points or 5 photos without control points)

Congratulations on getting the feature. For the right project it is incredibly useful.

 

[Jim Campi]

Thanks Shawn for the information.

This is a new process for me. The point types I identified above were from the 23 page "Camera Measurements" doc that Javad provides. The point types are not defined as I have them above however; I have attempted to differentiate them as used in the document to better understand the process and develop an algorithm that is repeatable. For example, it appears that a Target (point) really should be at a 90 degree angle with Vantage points (when possible). I also gather that it's not as critical to have the tie points at a 90 degree angle with Vantage points (as long as there are 2 per Target point).

You mention Tie and Reference points. I presume that Tie points can and in many instances are the Target (unknown point of interest)?

 

[Shawn Billings]

Yes a tie point is basically anything that isn't known.

 

[Jim Campi]

Have you tried using RAMS when working on the LS for a significant amount of time manipulating points/images?

 

[Jim Campi]

I attempted to use RAMS. The system lags a bit and would not make processing images more efficient...perhaps I need to adjust the configuration to increase responsiveness.

 

[Matt Johnson]

Have you tried using RAMS when working on the LS for a significant amount of time manipulating points/images?

Are you just wanting to be able to click on the photos with a mouse and use a keyboard? In the next version of the OS that will be released USB keyboards, both wired and wireless, can be used with the LS. A USB mouse can also be used with the OS but we do not have an option to show the mouse cursor in J-Field yet.

 

[Jim Campi]

Yes, it would be much more convenient using a mouse and keyboard for intensive office work on the LS.

 

[Shawn Billings]

I hope that we can add some sort of desktop processing option someday. I like the flexibility of being able to work in the field, but it would be easier to work at a computer monitor and keyboard.

 

[Mikhail Drakin]

In fact, if one considers sensible taking 10-20 or more images with small displacement between them, they may be fed in Photoscan, and most probably it will be able to process them - fully automatically and with 3D model generation. The key things in this case are: 1) large enough number of images with relatively small vantage point change between them (intersection angle between outermost shots still should be about 90 degrees); 2) feed separate CSV with coordinates into Photoscan, as EXIF geolocation tags in LS images are only single epoch data. As Photoscan is not too good at estimating accuracy, I'd recommend using check points in this workflow as well.

 

[Jim Campi]

Autodesk Recap 360 accepts photo projects. Any idea if it's compatible with JAVAD's format?

 

[Shawn Billings]

I don't know anything about Recap. It would be interesting to find out, Jim.

 

[Jim Campi]

Apparently Recap 360 is free. It's included in the Infrastructure Design Suite I have on subscription however; you don't need to buy a product to obtain this version. They also offer Recap 360 Pro with more features that is a paid version.

I also believe both versions support UAVs. Triumph F1?

 

[Scott Harmon]

I really want to use this feature. We do ALTA Surveys and building heights are a requirement, it would be much easier to collect a point and take a picture of the building to get this data.

How can I tell if my unit has the "Camera Offset Survey Option enabled"

 

[Shawn Billings]

Hi Scott,
Go to Home, Support, Options