Earth Imaging Journal: Remote Sensing, Satellite Images, Satellite Imagery
Breaking News
BlackSky Announces Next Generation Dual Payload Satellite Architecture to Deliver High Resolution and Nighttime Imaging Capabilities
HERNDON, Va.- BlackSky, a leading provider of global monitoring...
Draganfly Selected as Sole Provider of Smart Vital Sign and Social Distancing Equipment
Los Angeles, California - Draganfly Inc. (OTCQB: DFLYF) (CSE:...
Esri and AfroChampions Launch Partnership to Promote GIS in Africa
Redlands, Calif., United States:  Esri, the global leader in location...
Verizon deploys remote network-connected drone during Big Hollow Wildfire
PORTLAND, Ore.- The Federal Aviation Administration (FAA) granted Skyward,...
United Launches Online ‘Map Search’ Feature, A First Among U.S. Airlines
CHICAGO - Let's say you live in Chicago, have $250 to spend...


Terrasolid added new functionality to TerraScan, the company’s point cloud processing software. For two years the software has had the ability to automatically vectorize 3-D building models based on classified point clouds and building roof points. Now it’s  possible to automatically create wall textures from oblique images.

Automatic Building Vectorization

The purpose of automatic building vectorization is to produce approximate 3-D vector models of buildings and building roofs rapidly. The automatic building vectorization is based on properly classified point clouds. One must first be able to recognize and classify ground points and point hits on building roofs.

TerraScan makes it easy to add texture to building models by classifying laser points (top), automatically vectorizing buildings (middle) and then adding texture (bottom).

This can be done automatically with specific algorithms in TerraScan. The point clouds first must be calibrated and tied into measured reference points. This is often done with TerraMatch, which uses trajectory data to find the best solution for the match. Once the multiple flight passes have been calibrated with each other, the overlapping point clouds should be classified to their own class, often called overlap.

The next step is to classify the points into different classes, such as ground, vegetation, building roof, etc. It’s advisable to check the classification result with aerial photographs for any anomalies. It’s possible to run the building vectorization right after the automatic classifcation, but it’s advisable to manually verify classifications, as there may be some strange points left in the classfication resulting from errors or misclassified data, such as roof windows.

Once classification is complete and the automatic building vectorization has run, each building has a unique identifier. There’s a tool to review each building and see if it meets the conditions or if it needs to be edited. In addition, there’s a set of tools to adjust the roof edges or edit details. Aerial images with accurate camera x,y,z positions and orientations support this process. Because laser data provide known equations for roof planes, one can measure edge positions from a single image. This measurement method is often called monoscopic measurement.

Laser point density has a major effect on the accuracy of the automatically generated models. The higher the point density the better and more accurate models can be produced. Low-density point clouds (< 2 points/m²) produce good models of large buildings, but there are problems with small buildings. Medium-density point clouds (2-10 points/m²) already produce good models, and with high-density point clouds (> 10 points/m²), users can produce very accurate and detailed building models.

Automatic Wall Texturing

TerraScan’s latest feature uses oblique images for wall texturing for more realistic city models. The images can be captured during LiDAR collection, ideally with five cameras—one looking down and the other four in oblique angles forward, left, right and backward to capture most of the facades. Alternatively, other available oblique images can be used, such as those from Pictometry. The only requirement for the oblique images is that the camera calibration parameters and exterior orientation are known. The software automatically textures the walls when the images are aligned and positioned. The building vector models don’t have to be produced from LiDAR data, as even manually vectorized building models can be texturized with this method.

For more information, visit Terrasolid at

Comments are closed.