Once you have collected your dataset, you can move on to the next step, image processing through photogrammetry software. Because the requirements for data differ from one software to the next, we suggest following these general data verification tips but confirming any specific data requirements directly with your photogrammetry software provider.
Photogrammetry is a crucial step in accurate aerial mapping, as the images you choose to process will ultimately determine the quality of the 2D or 3D dataset that will then be used by our autonomous drafting platform.
Data Quality Check
The first step to take before submitting data for processing is verifying the following settings and ensuring that they are within the recommended quality metrics.
- Image Overlap
- Flight Grid
- Camera Angle
- Weather & Lighting Conditions
- Sensor Megapixel
Ground Control Verification
As one of the most important steps in the pre-processing stage, verifying that your ground control is yielding accurate readings is crucial. You should check that you are getting accurate readings from your GCPs and that they are within an acceptable margin of error. If you conduct multiple readings, make sure you cross-check and compare all of the readings for consistency. Above all, to ensure the best possible results, you need to ensure the proper spread and placement of your GCPs. Refer to our section on GCP placement to see our recommendations.
The specific software provider you choose for your processing is ultimately your choice as they all have their own benefits. The following are our general recommendations for data verification before processing for AirWorks use:
- Ensure creation settings like GSD and orthomosaic are at least at our recommended minimum. The opposite is true for point cloud density, which should be at our recommended maximum.
- Avoid downsampling the dataset below our recommended parameters. The AirWorks web-app thrives on dense data and will uphold its fast processing times despite large file sizes.
- Verify that your project is geolocated before processing. Tagging your GCPs is recommended in order to get pixel accuracy during processing in the AirWorks app.
- If you do not want your project to be geolocated, you do not need to follow our GCP recommendations, as not every project requires it and AirWorks accepts non-geolocated datasets as well.
- Ensure that the data is located in a coordinate system with an EPSG code, AirWorks does not accept data located in arbitrary coordinate systems.
Once these have all been met, you are ready to process your aerial image set into your photogrammetry software of choice.
For use in AirWorks, ensure that an orthomosiac .tif file is generated, along with a .las or .laz file. Our web application accepts multiple files, as well as tiled .tif and .las files.
AirWorks yields best results with data collected with unmanned aerial vehicles (UAVs) but can process any aerial dataset that has been processed with standard photogrammetry software.
Types of Drones
These types of aircraft have multiple rotor blades that move in a circular pattern around a fixed mast, which is known as the rotor. Common rotorcraft types in aerial mapping are helicopters, tricopters, and quadcopters. Because these do not require a constant vertical thrust, they are much easier to deploy and land due to their ability to lift directly upward. For this reason, they are also easier to maneuver and handle, making flights in dense areas less complicated. One of the downsides of these types of drones is their shorter battery life, resulting in shorter flight times.
Given the benefits and disadvantages of these drone types, we recommend rotorcraft for smaller scale mapping projects.
These types of aircraft have rigid wings such as those of traditional airplanes, thus requiring forward thrust and a runway or launcher to take off for flight. Even though they have a simpler structure and can carry significantly greater payloads in terms of cameras and sensors, fixed-wing aircraft are better suited for more specific aerial mapping projects than the more versatile rotorcraft types.
We recommend using fixed-wing drones for large-scale mapping projects, corridor work, highways, oil & gas pipelines, utility projects, etc.
The sensor that you choose to fly with your drone will vary depending on the site profile, accuracy standards, and deliverable needs of your project. That being said, there are a few types of sensors that might be best suited for your aerial mapping needs.
We recommend a LiDAR sensor that has a higher density of points for collection, which can be reduced in post-processing. Sensors we usually work with range from 200k to 1.5 mil points per second and 100 to 400 scan lines per second; for high accuracy applications, we recommend the highest collection rate in both categories, and for site monitoring or low accuracy applications, we recommend staying in the middle range and balancing points per second with scan lines per second.
The field of view usually has a range of 75-330 degrees, which is okay for most applications but requires tighter flight lines to maintain the 1/2 swath width side overlap. The target echos, or a number of returns from a single pulse, allow for more detail under vegetation and a more accurate bare earth, and less noise. The typical range is 1-15 target echos and we recommend a minimum of 5 target echos for most applications.
Beam divergence is important for identifying the usable swath and flight altitude, the typical range for beam divergence is 0.35 mrad – 3.2 mrad (35-320 mm divergence at 100 m distance to the ground). We recommend the lowest beam divergence to maintain low noise and allow for a higher flight altitude and more space between side overlap but if you have a sensor at the upper range of the divergence, you can compensate by flying at a lower altitude and with greater side overlap.
Inertial Measurement Units (IMUs) put together information from various sensors to provide accurate measurements for orientation, velocity, and pressure altimeter of the drone. Some common IMUs are accelerometers, magnetometers, and gyroscopes. It is possible that these are already built into your aerial mapping UAV. Our application supports data collected with either of these types of sensors, as such, we do not have an official recommendation as sensor choice will depend on your specific project.
Even though you are probably using GCPs to ensure the accuracy of your data, enabling GPS will further validate the accuracy of the aerial dataset. Depending on the accuracy standards your project requires you to meet, you can choose to use either RTK or PPK.
The number of GCPs you choose will also depend on the GPS system you select. Below are our recommendations for the number of GCPs per site depending on whether you select RTK or PPK.
- Minimum of 10 ground control points if using RTK
- Minimum of 10 ground control point if using PPK
- 3 usable ground control points are the bare minimum needed. You should always collect more in the event your RTK/PPK fails.
- Minimum of 10 ground control points per 100 acres, regardless if using RTK or PPK
As mentioned before, the angle that you select will depend on the profile of your site. Visit our recommendations for camera angles in our site profiles section.
Typically, your camera’s auto settings will work well enough to adequately collect data. We recommend you follow our recommended flight parameters table to ensure best results depending on the specifics of your site and your project. One important thing to note is that you should ensure GPS location is included in the metadata for every image you collect during your flight, which is typically done automatically for most cameras.
This setting control how much light comes into the lens of your camera, and usually refers to the length of the lens, F, over the diameter of the aperture, f/X. Having a larger aperture setting will yield brighter images, while a smaller aperture setting will result in darker images. We recommend setting your camera’s F-Stop at F/4 or F/5 to get light-filled images in which site details are more easily recognizable.
Your ISO setting determines how sensitive your camera’s sensor is to light. A high ISO increases the sensitivity of your sensor towards light, which results in a brighter image that also better displays visual distortions. A low ISO is the opposite and will result in darker images with obscured visual distortions. We recommend setting the ISO between 100 and 250 to reduce the number of visual distortions in the imagery.
The speed of your shutter controls the sensor’s exposure time, which also determines the amount of light that enters the lens as you are capturing an image. High shutter speed settings shorten the sensor’s exposure time, resulting in sharper but darker images. Low shutter speeds, on the other hand, lengthen the sensor’s exposure time, yielding lighter but blurrier images. We recommend a shutter speed of 1/250 or above to get images that are crisper and show more details while still remaining light-filled.
Global shutter vs. Rolling shutter
Depending on the type of camera your drone is equipped with, you may need to double-check whether it is automatically using global or rolling shutter settings. Rolling shutter cameras will typically yield a vertical displacement/distortion that results from the shutter capturing the image in a frame-by-frame manner, causing the “rolling shutter effect”. Global shutters, on the other hand, capture the entire image frame at the exact same time, which eliminates distortion and creates more accurate images for the purposes of aerial mapping and CAD drafting. For best results, we recommend using global shutter cameras whenever possible to avoid image distortions.
At AirWorks, we try to follow the recommended American Society for Photogrammetry and Remote Sensing (ASPRS) practices for Ground Control Point (GCP) usage and placement. GCPs are used in aerial mapping as a way to ensure the accuracy of the map that is then created from the collected dataset. The type and style of GCP targets you choose to use are not as important as ensuring you place them both strategically and correctly in order to ensure accuracy.
The best suggestion we can give in this case is to verify that GCPs are spread evenly throughout the entire acreage of the site, with each acre having a uniform amount of GCPs within it.
Here is our checklist for best GCP placement for accurate map rendering:
- Spread GCPs evenly throughout the entire acreage of the site.
- For regularly-shaped polygons, place a GCP at each corner of the site. This applies to both regular and irregular polygonal-shaped sites.
- Position GCPs at both high and low points of the terrain throughout the site.
- Set the accuracy of your GCP target location to 0.05 ft to meet ASPRS standards.
- Measure the location of your GCPs both before and after flight. If using a non-permanent target, also measure both before and after flight.
- You should measure your points with different constellations of satellites to get an independent reading. Usually an 8-hour buffer.
- Implement the utilization of checkpoints for quality reports. These points should follow a similar layout to your GCPs and can be redundant in placement.
- Redundant placement of GCPs throughout the site is not an issue, as more GCP placements will give you more options for geolocating the project during the photogrammetry and image processing phase.
- AirWorks recommends using a 50/50 split of GCPs and checkpoints in your site.
- Checkpoints should not be used to locate the project, just to check for output location.
We recommend using a ground control target that looks like this:
Number of Images
The biggest factor in deciding the number of images you should collect during your flight will depend on your choice of photogrammetry software, as each one will have different requirements, as well as limits. A general rule of thumb to follow is to not try to process projects with more than 5,000 images per flight, but collecting more images will always render better results than attempting to process small datasets with limited images.
If you do have a flight with more than 5000 images, we recommend breaking up the project into multiple flights and processing them as separate datasets during the photogrammetry stage, then combining them into one single dataset in post-processing.
Point Cloud Density
If you have a point cloud that is over 3 GB in size, we recommend breaking up the project into multiple las files. If you have a flight with a point cloud density greater than 5000 points per square meter (ppsm), we recommend thinning the point cloud to reduce the density of points.
The altitude of your flight will depend on the type of aircraft you select, and the sensor quality it possesses. In general, for aerial imagery, we recommend a flight altitude of 150 ft along with a 20-megapixel camera resolution and 0.5-inch ground sampling distance (GSD). For aerial LiDAR, it is dependent on the number of returns and beam divergence. If your sensor has multiple returns and low beam divergence, you can fly around 250-300′ and collect a point cloud with low noise/distortion and a high level of detail; as you have fewer returns or more beam divergence over a shorter distance, you need to reduce flight altitude to be closer to 150′.
That being said, your flight altitude can be higher or lower depending on the accuracy requirements of your specific project. Accuracy is also dependent on the number and distribution of ground control points, the control collection methods, and processing techniques.
The flight grid that you select for your data collection stage will vary greatly depending on the profile of the site you are flying, and things like vegetation density and structures or buildings will affect the camera settings required for accurate data collection.
All projects should be collected at 90 degrees/nadir for the main grid. You can add extra passes with 45-60 degree obliques to make the buildings & vertical structures look better in a 3D model but it doesn’t apply to the work AirWorks is doing nor will it provide the data quality we need to provide an accurate CAD deliverable. A double grid with perpendicular flight lines is recommended for all LiDAR collections.
|Site Profile||Recommended Camera Angle||Recommended Flight Grid|
|Dense, urban setting with many vertical structures||60 degrees, oblique||Double Grid|
|Dense, urban setting without vertical structures||90 degrees||Single or Double Grid|
|Suburban setting with vegetation||90 degrees, oblique and nadir||Single or Double Grid|
|Rural, open setting without vegetation||90 degrees, nadir||Single Grid|
|Homogenous areas (ex: snowcapped fields, woods or forests, large bodies of water, etc.)||In this case, it is best to increase flight altitude in order to add more differentiated details in the images you collect, which will also make data processing and photogrammetry|
Weather conditions can have a great effect on the quality of the images that your UAV or aircraft captures. While it may sound counterintuitive, optimal data capture happens when weather conditions are cloudy. These conditions allow for uniform lighting within an entire set of images and also reduce shadowing of objects throughout the site. If weather conditions are not cloudy, we recommend flying when the sun is at its highest point, around noon (12 pm) to reduce distortions due to shadows in the images.
In order to build an accurate 3D model that yields high-quality drafting results, later on, you must also prepare to conduct a flight which encourages sufficient image overlap, both front-overlap, and side-overlap. At AirWorks, we recommend setting your flight to 80% sidelap and 85% frontlap to get the best results from photogrammetry and subsequently your drafting results.
Start by setting the boundary of your project. This is dependent on the terrain and size of the mapping project. What is most important before flying to collect data is remembering that your flight boundary must be set beyond that of your project.
We recommend you set the flight boundary at least 100 feet beyond your selected processing boundary to ensure you have an adequate artifact distortion buffer. This will also ensure the geospatial integrity of the model.
If flight includes the use of LiDAR sensors, flight boundary should include 1/2 swath distance.
|Aircraft Type||Rotorcraft or Fixed-Wing UAV|
|Image Settings||Angle: Based on Site Profile |
GPS Metadata: Enabled
Aperture: F/4 to F/5
ISO: Low preferred, between 100 and 250
Shutter Speed: 1/250 or above
|Flight Boundary||100 ft distortion buffer from processing boundary.|
With LiDAR: Include 1/2 of swath distance.
|Weather Conditions||Fly on cloudy days for best results.|
|Flight Line Overlap||Setting at 80% side overlap and 85% front overlap.|
With LiDAR: 20% side overlap.
|Flight Grid||Depends on the site profile.|
With LiDAR: 2nd pass perpendicular to 1st pass, turns made out of AOI.
|Flight Altitude||150ft |
With LiDAR: Depends on the number of returns.
|Number of Images||Depends on your choice of photogrammetry software, usually no more than 5,000 images.|
|Ground Control||Based on ASPRS accuracy recommendations.|
The AirWorks application is also suitable for datasets collected with manned aircraft, with some limitations depending on the GSD and altitude of the data collected. To learn more about processing manned aircraft data in the AirWorks app, click here.
In order to maintain the highest standards of accuracy in our deliverables, we request that if you have data that has been collected outside of our recommended parameters, you either collect the data again or speak to someone from our support team before submitting your data to discuss accuracy expectations and what we are able to deliver.
Please contact your account manager for details on data quality and accuracy.