Image Acquisition

How do I capture clear images for mapping and survey?

It is important to have a clear image acquisition plan before flying your drone for mapping and survey. In order to capture clear images, we recommend to follow these 3 steps:

Step 1: Plan for image acquisition

Existing maps and satellite images are a good place to start evaluating the target site. We would also recommend to carry out a site recce whenever possible, to determine the whether the target area is safe for drone flying.

Changes in altitude during a drone flight will result in variations in the Ground Sampling Distance (GSD) of your input images. When generating 3D maps and models, it is recommended that the maximum GSD of input images is no more than 2 times of the minimum GSD.

For example, if the minimum altitude of your drone images is 150m, the maximum altitude of images in the same project must not exceed 300m in order for processing to be successful.

Step 2: Follow general image acquisition guidelines

There are 2 important aspects to take note of:

1. Image overlap Adjacent images must have sufficient overlap in order to be properly stitched together into 3D maps and models. An easy way to ensure enough overlap is to capture images with a wide-angle or fisheye lens.

For example, 80% overlap” indicates that any two adjacent images should capture the same area in approximately 80% of their frames.

2. Camera Angle Depending on whether you are creating a 2D/3D map or 3D model, you may also need to adjust the angle at which your camera is pointing. This ensures that all desired surfaces are visible in some of the images captured.

In most cases, this means flying several passes over the same area with the camera pointing at a different angle. We would like to recommend you to follow the settings for 2D/3D maps and 3D models.

Step 3: Geotag images in preparation for processing

Images must be geotagged before you upload them to Garuda Plex. If you used a camera with an automatic image geotagging feature, this data would already have been included for you.
If your camera doesn’t have the function, you have to manually geolocate your images.Please kindly refer to ‘Geotagging’ section.

Recommended Settings for 2D/3D Maps and 3D Models

Recommended settings for 2D/3D maps

  • Frontal Overlap: At least 85%
  • Side Overlap: At least 70%
  • Camera Angle: 0°(vertical/nadir)
  • Flight Altitude: Typically 120-300 m for 16-24 MP cameras.

Note: Do note that capturing images at a greater altitude ensures more overlap and allows for more reliable map generation but reduces the resolution (GSD) of the final map.

Recommended settings for 3D models

  • Image overlap (with GCP): 80%
  • Image overlap (without GCP): 90%
  • To ensure that all surfaces of the target area/structure are captured, we recommend a minimum of three flights, each with the camera pointing at a different angle:

Flight 1: Camera pointing at 45° from vertical
Flight 2: Camera pointing at 30° from vertical, altitude greater than Flight 1 [See note 1]
Flight 3: Camera vertical (nadir, vertically downwards)

Note: You may add flights between Flight 2 and Flight 3 if you wish to improve the final 3D model. Each subsequent flight should take place at a greater altitude and a smaller camera angle from the vertical (increasing pointing downwards).

You may also use handheld cameras to capture close-up images of the target structure from the ground. Maintaining a high level of image overlap in these ground-level images is critical if you want the final 3D model to have detailed facades.


What is geotagging?

Geotagging (aka geocode, georeference) is a process of including geographical information (e.g. latitude, longitude, altitude…) to various media, primarily images captured by camera, in the form of metadata.

Why is geotagging of images important?

Geotagged images allow a faster and high quality image processing for photogrammetry, 3D modeling, 2D/3D map generation.

How do I manually geotag the images captured by camera with no built-in GPS?

For video tutorial, please click this link: ‘Mission Planner: Geotag Images

Things to look out for include:

  • Individual images should be oriented correctly to the base map.
  • Important like the end of takeoff (WP1) or when the UAV turns to determine whether the images are “late” or “early”. Cross-reference actual images in the source folder.
  • If you decide that the time offset needs to be corrected (images have been tagged too early or too late), use the following formulas: Total offset correction = Number of images to shift x Seconds between images
  • If the images are “late”, the offset value needs to be less positive/more negative. If images are “early”, the offset value needs to be more positive or less negative.

Map Quality

What is map stitching?

In short, every photo taken by drone contains specific features such as buildings, roads, trees or other heterogeneous objects, will be matched by a complex mathematical process and aligned on top of each other. The end result will be a map consists of all the photos stitched together.

How to improve map quality?
  • Fly higher – To cover more features in each photo and to achieve higher overlap which helps in homogeneous imagery like plantation.
  • Fly on an overcast day – Cloud is a good light diffuser which provides soft & even lighting on your target/area of interest and reduces shadow and glare from reflective surfaces. It is an effective technique for 3D models reconstruction.
  • Modify flight path – To achieve shortest flight time and to fly cross-wind.
  • Increase overlap – To increase the number of photos captured simply by shooting faster in a shorter interval.
  • Increase sidelap – To get more matched features in imagery at the expense of longer flight time / reducing total coverage of a drone can cover in one flight.
  • Only nadir images – Images contain horizon which are normally captured during banking turn should be removed before processing.
What are the common issues in images captured by drone?
  • Motion blur – It is due to high flying speed of drone or low camera shutter speed. It can be avoided by reducing the flying speed of drone or increasing camera shutter speed.
  • Unfocused camera – Turn on autofocus mode.
  • Vignetting images – It is due to lack of light. It can be solved by flying your drone under sufficient lighting condition, especially during day time.
  • Insufficient overlap – Increase overlap in order to have more map detail over a smaller area.
  • Non-nadir photos – Horizon (aka infinity) leads to distorted internal distance. Thus, only include images containing the area of interest right below and remove non-nadir photos (captured during banking turns) before processing.
  • Photo captured at low altitude – It causes problem to stitch properly due to insufficient coverage of ground in one flight compare to a higher altitude flight.
  • Homogeneous imagery – It is difficult to stitch them due to few distinctly recognizable features (e.g. sea or lake).

Drone & Camera

What drone can I use to capture data for processing?

You can use any UAV setup to capture images. All images must be in JPG/JPEG, geotagged, facing the area of interest, have enough overlap (at least 70%, with sidelap 60%) and at least have 20 images.

Do I need LiDAR drones to create elevation map?

Nope. In fact, elevation maps can be created by using standard geo-referenced images captured by drones.

What type of SD card is recommended for mapping large areas (e.g. 500 ha)?

High speed SD card with at least 16 GB will be ideal.