10 Times Speed-up: Producing High-Quality 3D Maps from Lidar

DIPPER, a spin-off company from the University of Twente (Enschede, the Netherlands), provides a breakthrough solution on processing massive lidar data in an accurate and efficient way. It offers comprehensive services on lidar data processing and 3D scene modelling. The self-developed software is highly automatic, enabling DIPPER to create high quality 3D maps for 10,000 buildings within only one week, by one operator with one laptop. It means that at least 10 times speed-up has been achieved. The high efficiency and accuracy boost large scale applications, such as asset management, smart city, securities development, and urban planning. Up to now, DIPPER has successfully created 3D models for 5 international cities.

Figure 1. A 3D city scene of Den Haag, the Netherland. The 3D scene has buildings, ground, river, and roads, and are reconstructed by automatic-processing software which are developed by DIPPER. The 3D map of cities could be used in solar energy calculation, flood control, noise simulation, and sensor layout design.

Lidar, or Light Detection and Ranging, is a remote sensing technology that measures distance by illuminating a target with a laser and analyzing the reflected light. It is able to create intricate three-dimensional maps in places where bad weather or thick vegetation hamper traditional aerial mapping. In addition, airborne lidar system provides 3D data with 5cm accuracy on vertical direction, which is much better than 50cm accuracy achieved by dense matching from stereo images. However, current lidar data processing needs heavy labor works, making it very expensive and time-consuming. In the end, most companies on the market are either offering accurate 3D maps with high price, or producing affordable 3D maps with compromised quality. Therefore, it becomes extremely difficult to achieve advanced lidar applications.

DIPPER, a spin-off from University of Twente, was founded in December 2014 in Enschede, the Netherlands. Supported by the top research group on lidar mapping, DIPPER has successfully converted the 20-years pioneering scientific achievements to commercial products. The company has developed novel algorithms which enable data processing in an ultrafast and highly-automatic way. Meanwhile, it also provides reliable 3D ICT services including solar energy calculation, flood control, noise simulation, and sensor layout design. With the unique technology, DIPPER is confident and flexible to provide customized products and services according to the situation, application and requirement.

DIPPER’s technology can significantly improve the efficiency by creating high-quality 3D building models in a remarkable automatic way. The quality levels of 3D maps can be determined by how many details they provide. Level of Detail 1 (LoD1) is a level of detail at which buildings are reconstructed in just one height, which leads to the only flat roofs for all the building models. In comparison, at Level of Detail 2 (LoD2), the 3D models are very detailed; the roof constructions including dormers can be clearly modeled. Great efforts are needed to improve the level of detail from LoD1 to LoD2.

Figure 2. A 3D scene of very high voltage transmission line, China. The 3D scene has pylon, wire, building, and trees, and are reconstructed by automatic-processing software which are developed by DIPPER. The 3D map of transmission corridor could be used in vegetation clearance management, risk management, and new plan design.

The DIPPER software can construct 3D building models at LoD2, using lidar data collected from a helicopter or an airplane. During the data processing, by the software, the lidar points on buildings are firstly recognized and segmented into individual roof faces. The roof structures are then inferred and primitive sub-buildings are detected and modeled. A building with arbitrary complex structures is hereby modeled by composing a set of primitive sub-buildings. More impressively, if an error is discovered in the roof construction, the software is able to automatically recognize and memorize the error patterns, and then correct repeated errors in other buildings models. Therefore, this algorithm ensures the high quality of the 3D model.

The market of 3D map application is at the booming time right now. The international market reached USD 1.90 billion in 2015, and will go up to to USD 16.99 Billion by 2020, at an estimated CAGR of 55.0%, reported by M&M, a premium market research company. Regionally, the Dutch government is developing the national service for large-scale 3D topography. Cities like Amsterdam, Rotterdam, and Den Haag etc., have launched projects to invest the application of 3D maps for city visualization, management, and communication. Moreover, companies focusing on solar energies, insurance, security and city planning are tending to use 3D information for their business development and information support. Accordingly, the demand of cheap and accurate 3D lidar processing and modeling is growing fast, and DIPPER is ahead of this trend.

In 2015, DIPPER has developed advanced toolkits for lidar data processing. One important function of the toolkits is to model the urban scene, such as buildings, terrain, power corridor and trees, with high efficiency and accuracy. It means that DIPPER is able to combine 3D building models with environmental parameters and provide detailed overview and smart suggestions on many advanced applications, including solar energy analysis, noise management, flood control and urban planning. Meanwhile, DIPPER has established close connections with many government organizations in the Netherlands and China, such as Kadaster, municipalities, and utility bureau. By far DIPPER has made high-quality 3D maps for five Dutch cities and Chinese power bureau. With these achievements, DIPPER is also exploring business opportunities within non-governmental organizations such as solar panel companies, insurance companies and LoRa Alliance for “internet of things”.

“Our aim is to provide easy-to-use applications. We are transforming massive lidar data into semantic and concise 3D vector data.” Xiong, the CEO said. After accomplishing his PhD degree in University of Twente, Xiong launched this start-up with four passionate colleagues: “We have successfully processed lidar data for city, forest, railway, power corridor, and industry scenes. The data processing varies from multi-scan registration, point cloud classification, and 3D scene modeling. Since we work closely in a flexible while effective environment, we are able to response quickly to the challenging problems raised by customers every day.

As DIPPER’s slogan, Showing, Solving and Leading can perfectly explain what we are trying to achieve, that is, to show a new view of the world via 3D lidar maps, to solve complex problems and help with smart decision making by developing toolkits, and to lead lidar processing technology by keeping innovation alive. Moreover, as an innovative company aiming to build up a bright future, DIPPER is also willing to contribute to the new concept development, such as self-driving car and smart city development.”

In 2016, DIPPER is going to make a detailed 3D map for the whole Netherlands from airborne lidar data. It will be the first countryside 3D map globally at LoD2, which has extreme distinct roof structures. In the next five years, DIPPER is planning to work with other pioneering countries who are eager to have their accurate 3D maps. As an innovative company, it will never forget to invest on research and development to improve the software and workflow. Applying cutting-edge technologies including deep learning, big data mining, cloud processing for