Drone Mapping

Mapping creates unique types of deliverables that provide perspectives and value beyond individual aerial images: the orthomosaic, point cloud, mesh and digital elevation model. Below is a look at these products. Can drone mapping produce a survey? Read on.
Also, here is a look at how drone mapping works.


The value of an orthomosaic is to provide a view from above over a large area.

An orthomosaic (also sometimes called orthomap) is a photographic map. It “blends” together the perspectives captured in all the drone images (dozens or hundreds) produced in the mapping flight. (The actual processing involves more than just blending images.) The orthomosaic is flat, without perspective distortion. Compared to an individual photo of the same area, the orthomosaic avoids occlusion by tall structures: nothing is hidden behind anything else.

The amount of detail carried in an orthomosaic depends on the resolution of the camera and on the altitude from which the images were captured. See a comparison of photos from different altitudes.

Orthomosaic of about 3 acres of land. Actual resolution of the original is about 1.7cm per pixel in this case. Resolution depends on flight altitude and camera sensor.

Point Cloud

A point cloud is the right work product for a project where the value lies in analyzing the composition, presence or the shape of objects in a scene.

The point cloud is just that: points floating in space. The points represent features found by the algorithm: edges and details on surfaces of objects. A point cloud looks like a 3D model, but unlike a “regular” 3D model, it contains no surfaces – just clouds of points, clustered together in interesting ways.

It turns out that when we look at a point cloud, we recognize the objects it represents just fine. Point clouds often have odd holes that we don’t expect, but they do the job of showing a scene quite well.

Dense point cloud, about 12 million points. In the left image color represents an auto-classification of points. In the right image, color is derived from the drone photos. The orange markers in the front are for a distance measurement. The green markers in the back are Ground Control Points, which tie the geometric model into a real-world location.

There is a trick to showing point clouds: the point size. When each point is shown as a tiny dot (a single pixel or close), the point clouds looks semitransparent and wispy. When points are shown bigger, covering more screen real estate, then many neighboring points overlap, creating the illusion of a surface on a solid object.

Comparing a dense point cloud shown as small points (left) to one shown as larger points (right.) With the larger points, neighboring points overlap, providing a more u0022solidu0022 look of the scene, but debris also becomes more visible. Some tools also automatically vary point sizes to improve visual appearance.

Point clouds come in two flavors: sparse and dense. The sparse cloud is often referred to as tie points, and the dense point cloud often is simply called the point cloud. The sparse cloud is an interim product of the computation, which is needed as an input to compute the dense cloud. The sparse cloud consists of an initial set of points representing features that appear in multiple photos. The dense cloud also contains such feature points, but a lot more of them. After making the sparse cloud, the algorithm “has a general idea” about where things are, so it can “take another look” and find a lot more details, to make the dense cloud.

Tie points (left), dense point cloud (right)

Usually the dense point cloud is the useful product, since it shows a meaningful level of detail from close distances. The sparse cloud, on the other hand, can be useful if the project is extremely large, and details are not important. In that case, the costly creation of the dense cloud can be avoided.


A mesh is right for projects where surface details and textures are important in the context of a scene.

A mesh is a 3D model: textured surfaces, delineated by lines (“wireframe”) connecting dots in space. The geometry for the mesh is a simplified, smoother version of the point cloud. Textures, based on the drone photos, can make the mesh look very realistic. But edges that are straight or smooth in the real world sometimes look wrinkled in the mesh. To achieve an accurate mesh for an important object, such as a building, it helps to capture extra imagery of the object when flying the drone mission. For example, orbits around the object, with oblique camera angles, can be added to a normal grid-based mission flightpath.

Comparing the same dense point cloud to a 3D model (mesh) with textured surfaces. The mesh handles buildings well, but struggles with vegetation and debris. The quality of the input data (GPS accuracy, camera lens sharpness) can make a huge difference in the quality of the end result.

Digital Elevation Model (DEM)

Best for showing terrain surface.

A digital elevation model represents the altitude of the ground at every point. This is a professional format used with geographical information systems (GIS), to perform analytics of terrain, such as determining drainage (important for construction site prep) or building coverage. In general, “DEM” is a term that encompasses both digital terrain models (DTM) that model the bare ground surface and DSM (digital surface models) that include buildings and other objects.

With terrain models, the exclusion of features other than the bare ground is a challenge, with or without drone images. Addressing this challenge, in the point cloud examples above, we can see that a point cloud classification algorithm does a decent job in differentiating the ground, vegetation, buildings and man-made features, as long as the ground is not completely covered by vegetation.

“How About a Survey?”

A survey is produced by a licensed surveyor, delivering accurate details such as property lines and distances. Surveyors use high-precision equipment and techniques to produce these accurate results.

Graf Systems does not claim to produce surveys. The primary value of drone mapping is in providing a rich view of a site, allowing the viewer to make decisions about the site in context of what is currently there.

However, Graf Systems can collaborate with a licensed surveyor when capturing drone mapping data. The surveyor can use the drone mapping data to produce a survey.

Graf Systems