Earth Observation with UAV's  

Image-based modelling (IBM) refers to the techniques of acquiring 3D object information from two or more images. This includes three traditional photogrammetric algorithms (feature extraction and matching, Bundle Block Adjustment and orthophoto generation) and new techniques from the Computer Vision community (such as structure from Motion, Visual Odometry and Semi-Global Matching) to derive 3D point information from an image sequence. These techniques can be used to process both terrestrial and airborne images. Among the innovative platforms for data capture, Unmanned Aerial Vehicles (UAV, better known as drones) are becoming a valid alternative to traditional Geomatics acquisition systems, as they close the gap between higher resolution terrestrial images and the lower resolution airborne and satellite data. UAV can be remotely controlled helicopters, fixed wind airplanes or kites. Different sensors can be installed on-board to acquire data. Many applications ranging from 3D building modelling to crop and forest monitoring can profit from these data acquisition platforms. In this course the advanced IBM techniques and, in general, the 3D geo-information processing will be explained, with focus on the use of data acquired by UAVs. The course is composed of two main parts. In the first part, the four main steps of the modern IBM process (image orientation, point cloud generation, orthophoto generation and quality assessment) to retrieve 3D information from images will be defined. The peculiarities of IBM process using UAV images will be discussed in detail, showing the differences with the traditional acquisition of airborne images. During the second part the participants will gain hands-on experience on the use of UAVs. In this period, the students will learn how to process images acquired with different sensors and for different applications. Specifically, participants will learn the principle of IBM methods and they will design three simple solutions (feature extraction, feature matching and relative orientation) by adopting these methods in simple Matlab codes. Lectures will be always coupled with demonstrations and practical sessions on the theory delivered. The second part of the course will allow the participants to experience the UAV data acquisition and processing workflow. They will understand how a UAV related project is planned and executed with their involvement to a real UAV acquisition project. Then, they will apply the learned IBM techniques using a commercial software (Pix4D – www.pix4d.com) to process the acquired data and extract 3D information. They will finally analyse and compare the data using the available ground truth and dedicated tools and software (Matlab scripts and CloudCompare) to evaluate their results. Multi-spectral and thermal image acquisitions from UAVs will be also part of the course topics. Participants will learn how to process these images and how to better use them for different applications. Additional presentations will be finally provided to describe the use of UAVs in six different domains covering different perspectives of ITC Departments: land administration, disaster mapping and management, natural resources and crop monitoring, water management and flood monitoring, maintenance of UAVs and real-time processing. Participants do not need prior knowledge on the topics of the course.
English
2022-11-10T23:00:00Z
The course will be composed of lectures (with the use of flipped classrooms when necessary), practicals, supervised and unsupervised assignments and fieldwork for UAV image acquisitions. The student will learn how to correctly process the acquired images receiving both the theoretical and practical knowledge and gaining in self-confidence and independence during the course.
English
Participants do not need prior knowledge on the topics of the course.,All M-GEO and M-SE students are accepted. Note that we offer two UAV courses in Q5. GFM students should choose "Scene understanding with UAVs", while all the other M-GEO and M-SE students should join this course. In general, all students should have basic knowledge of remote sensing.
English
201900053
Earth Observation with Unmanned Aerial Vehicles
English

UNIVERSITY OF TWENTE

Faculty of Geo-Information Science and Earth Observation