This paper brings an innovative approach to generate 3D scene views from real world data using a matrix range sensor and a 2D video sensor. Besides the enhanced visualization, combining 2D and 3D data provides also valuable information for object detection and recognition as well as for other image processing methods. The scope of this work includes a discussion on extracting good features from low resolution images, necessary for the system calibration. The registration process is based on a geometrical model which is derived from the calibration parameters of the cameras. Finally, a FPGA based solution is provided, which generates the 3D scene on-the-fly. The proposed architecture achieves its maximum performance with a fully pipelined implementation from the raw data to the generated fusion coordinates, while limited hardware resources are consumed. In contrast to standard processor approaches, a significant performance boost is achieved with the FPGA architecture.
Fast fusion of range and video sensor data
Schnelle Datenzusammenführung aus Entfernungs- und Video-Sensordaten
2007
16 Seiten, 13 Bilder, 2 Tabellen, 7 Quellen
Conference paper
English
Fast Fusion of Range and Video Sensor Data
British Library Conference Proceedings | 2007
|Fusion of range and intensity data in a scanning sensor
SPIE | 1992
|Variational approach to sensor fusion using registered range and intensity data [1956-21]
British Library Conference Proceedings | 1993
|High-Altitude Precision Landing by Smartphone Video Guidance Sensor and Sensor Fusion
DOAJ | 2024
|