Advanced Topics in Computer Vision by Giovanni Maria Farinella Sebastiano Battiato & Roberto Cipolla

Advanced Topics in Computer Vision by Giovanni Maria Farinella Sebastiano Battiato & Roberto Cipolla

Author:Giovanni Maria Farinella, Sebastiano Battiato & Roberto Cipolla
Language: eng
Format: epub
Publisher: Springer London, London


7.4 Image-Based Localization

A large number of interesting applications for micro aerial vehicles such as surveillance, agriculture, inspection, or simply 3D reconstruction for visualization share one common feature: The need for highly accurate localization of the vehicle with respect to the scene. According to the work of Konolige and Bowman [33] on lifelong maps, long-term visual localization requires an accurate map with good features, but should also allow incremental mapping and recovery from localization failures. An additional requirement for MAVs is the possibility to switch between GPS and visual localization, as visual maps are unnecessary when flying high above buildings or in wide open spaces.

In the previous sections, we have shown how to create metric, geo-referenced 3D models by taking pictures with a consumer camera from eye-level above ground or by using micro aerial vehicles as acquisition tools. Now we demonstrate how the concept of virtual views helps to incorporate prior knowledge and fulfills the requirements listed above. We present a method for monocular localization with an accuracy comparable to differential GPS in a global coordinate system (Fig. 7.13). Further, we introduce an incremental feature update step which allows to fuse in-flight information back into the original scene model, and we exploit additional sensor information to ensure localization robustness.

Fig. 7.13Localization in a geo-referenced, metric visual landmark allows to convert the resulting local ECEF coordinates back to a GPS position. It is therefore possible to seamlessly switch between GPS and visual navigation, as well as to exploit onboard GPS and IMU data in the localization process. ©2012 IEEE. Reprinted, with permission, from [69]



Download



Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.