Ed/ Autumn/ Dave,

I am trying to get my head around required settings for using the Inspire for recording geospatial data. With this regard could you be so kind as to obtain the following information please as I am trying to calculate a Ground Sampling Distance and require the following information

1.    What is the Camera sensor width in mm =

2.    What is the Camera Focal Length in mm (Real focallength) =

3.    How do I calculate the Image width in pixels for a givenaltitude e.g. if flying at 30m altitude what is the image width at ground level in pixels

The above information will enable me to calculate the Ground Sampling Distance (GSD).  This is the distance between 2 consecutive pixel centers measured on the ground. The bigger the value of the image GSD, the lower the spatial resolution of the image and the less visible details. The GSD is related to the flight height: the higher the altitude of the flight, the bigger the GSD value.

For example a Ground Sampling Distance (GSD)  of 10 m means that one pixel in the image represents 10 m on the ground.

Would appreciate your help in providing the above camera specifications.


Thanks in advance..


GB

Just had two chats on the DJI Support On-Line Chat.  Result, both Chats were ended by DJI Xin without answering my questions.    Then DJI ask me to submit how well they performed !  Simply unbelievable.

Come on DJI, a little help here please.  

DJI has the product information, specifications, technology, engineers and the ability to answer these questions.  Please help.

I doubt you'll get much help out of them mate.
Might be speedier for you to find a sports ground or other flat space and head out with a long tape and some cones.
Lay out a grid, make it as accurate as you can and vary altitude with camera pointed straight down until you hit the extremities of the FOV.  
You should then be able to back-calculate the data. Maybe not the most accurate approach - but might be close enough?

Farnk666 Posted at 2015-8-15 21:05
I doubt you'll get much help out of them mate.
Might be speedier for you to find a sports ground or  ...

Farn,

Thanks for the input.  

I will probably be out tomorrow flying and may take a certified 30m tape out with me and do just that.  Take some stills at specific altitudes and see actual ground distance recorded.

Unfortunately I still need the exact camera specs to calculate the altitude I should be flying to capture the quality of data required.

I need to calculate the real focal length calculated by :

FR [mm] = Camera Focal Length (assumed is 20mm from DJI Manual)

therefore, at the minute I have a calculation :

Real focal length = FR [mm] = (20mm * SW) / 34.6FR

What is the SW ?

SW = real sensor width in mm


So would be helpful if DJI can confirm what the Real sensor width is please

Ok guys here's an update with some Inspire Camera specifications obtained via a third party ( Would be good if Ed or Autumn could confirm these specs) :

Sw = the sensor width of the camera (mm) = 6.17
FR = the focal length of the camera (mm) = 3.55137
imW = the image width (pixels) = 4000

With these settings, we should be capable of determining the Ground Sampling Distance (GSD) when collating geospatial data collection and need to know what the ground spread/distance of one image would be.

For example if flying at an altitude of 48m this will suggest a GSD of  2.0848m which is basically the footprint width covered on the ground by one image in width.  Imagine a Pyramid, with the aircraft at the peak, the footprint covered on the ground e.g the base of the Pyramid would be 2.0848m in width.

I also done some specific ground measurements at various altitudes recently to determine actual spread recorded by the Inspire across known accuratley measured Ground Points.  Can provide this data if anyone is interested.


Hope this helps ...

GB

I really like the spreadsheet that Pix4D folks created for doing all the heavy math. https://support.pix4d.com/hc/en- ... OOLS-GSD-Calculator
It also includes a nice visual of the various variables...
When I plug in the values you have GB44, the GSD is 2.08483 Centimenters, NOT meters.  The GSD is  the Pixel size, not the image foot print.  If there are 4000 pixels along the width of the photo, then the image captures 4000 pixels that are 2.085 cm each for an overall width of ~273 feet image width
I had earlier found some slightly different values for camera spec, but won't muddy the water by sharing them here.

I just found this thread. I will see if I can get one of our engineers to answer your question on camera specs. I checked the data I have and it was not there.

Dave

pinpointmapping Posted at 2015-8-19 02:05
I really like the spreadsheet that Pix4D folks created for doing all the heavy math. https://support ...

Further to your comment.

I looked at this again and noted that the Pix4d calculations you mention states that although the calculation is in cm/ pixels the result is in metres (m).  See link :

https://support.pix4d.com/hc/en-us/articles/202557469

So in my example this would be the distance covered on the ground by one image at the specified altitude, but in metres.  If it were in cm, the amount of images you would need would be thousands to cover any significant area and at an approximate 5MB per image with a 12MP camera you would soon run out of memory.  If traveling along a fixed line 30m flight in cm it would need to record 15000 images based on a 2cm spread, so if this was to be multiplied over a large autoflight grid pattern the memory card required would be massive.  

This would seem to meet similar results by others using a 10MP camera at 100m altitude, which provided a ground photo width of 125m, which is not too far away from my own site measurements.. See this detailed report at :
Bosak secrets of UAV photomapping.pdf - ResearchGate



removed table as was confusing

GB44 Posted at 2015-8-19 02:47
Further to your comment.

I looked at this again and noted that the Pix4d calculations you mention ...

Aloha GB44,
Math has never been my favorite subject.  Are you saying that if/when you fly @ ~48 meters above the ground that the camera is only able to see an area 2.0848 meters wide?   The Pix4D tool does say that GSD is  cm/pixel.  We know the image is 3000x4000 pixels (thus 12mp), so a 4000 pixel wide image with 2.0848 cm pixels would be 4000x2.0848=8339.2cm or 83.392 Meters wide (Footprint Width)

pinpointmapping Posted at 2015-8-19 03:18
Aloha GB44,
Math has never been my favorite subject.  Are you saying that if/when you fly @ ~48 me ...

Yes I agree, Maths is not my strongest attribute, but have a look at my post above.  Not sure if I was in middle of editing when you posted your reply.

Would be good to get a consolidated agreement though for my benefit also, just so I havent got this horribly wrong.
No certainly not saying camera can only see 2m wide at 48m altitude - seel my site findings and measurements above.  

I think the Pix4d calculation is refereing to an approxiamte 2m ground spread to give a good desired overlap.    ......... I think !!! So using the Pix4d calculation this would give a spread of 4.34m wide for 100m altitude

pinpointmapping Posted at 2015-8-19 03:18
Aloha GB44,
Math has never been my favorite subject.  Are you saying that if/when you fly @ ~48 me ...

Pinpoint,

Just been looking at the Pix4d calculations and they are using (mm) and (m) in their calculations by multiplying the height by 100.   The formula used gives a final answer in metres.

GB44 Posted at 2015-8-19 05:59
Pinpoint,

Just been looking at the Pix4d calculations and they are using (mm) and (m) in their ca ...

The handy Pix4D form/spreadsheet doesn't include any over (side/end) lap, and only provides a GSD in centimeters.  You use the GSD to determine the area covered by each image.  The link you provided shows a formula Dw=(imW * GSD)/100  which may be causing confusion.  The 100 it is divided by is to account for conversion of cm & Meters.  The length or width an image covers is simply pixel count * pixel size.
Then to determine proper picture capture to achieve the desired overlap (I like 75%-80%), use standard R*T=D. If an image is 4000 pixels wide & 3000 pixels tall (and each pixel is one inch (2.54 cm)) then the image footprint is 3000 inches x 4000 inches.  To achieve end lap of 75% you are are only capturing 25% new data per image or 3000*.25=750 pixels ( 750 pixels that are each 1 inch in this example) If you have the inspire set to capture an image every 5 seconds, then you have 5 seconds to travel 750 pixels forward and side lap would be 4000*.25=1000 pixels or in our simple example 1000 inches  of new ground.

I do mapping as well, but not on my Inspire. I gave up and use an X8+ with Ardupilot and Pix4D.I do almost all mapping at 50m AGL and get a resolution of <1.5cm/pixel. which is the "target parameter in Pix4D. The results are very good and I have pleased clients. A simple starting point if trying to plan your flight is to establish 50m AGL as your starting parameter. Then use a spread between your runs of about 3m (try to vsualize a sheet of plywood long between your legs).  I shoot at 2 sec intervals. This is typical result of automated setup in the Mission Planner software, but I'm trying to give you a manual flight approach. Err on the "more overlap" on your runs to make sure you get enough. The digital records are cheap and you won't have to go back an re-fly.
J.

PinPpint/ Sky....

Thanks for your comments.  I guess it just takes time to get to grips with these flight plan parameters and end results.

So it would seem that we need DJI to provide some update in the App to allow for 2s intervals for recording automated still images