ARAM
2.2
|
ARAM is a C++ library for Augmented Reality on Medical Applications based on the extraction and identification of fiducial markers in real time. Based on Opencv, several technologies and approaches are combined to resolve the various problems related to the accuracy, robustness and occlusions. This library is provided under BSD license.
The library deals with the identification of markers in the presence of occlusions without sacrificing the accuracy of the pose camera estimation. These markers are generated using Hamming Code with error correction. In order to assess the accuracy of pose estimation, specifically when high levels of accuracy are required on medical field, it is mandatory to carry out a thorough analysis of the errors occurring in the whole process. Consequently, this library deals with the performance analysis of the 3D pose estimation. The suggested analysis focuses on the error estimation affecting the edge detection process.
The image processing software performs real-time 3D camera pose estimation. Firstly, the camera is carefully calibrated off-line. The extraction of edge information is the first fundamental processing step of the whole pose estimation procedure. The edge or contour point detection relies on well-known gradient-based methods. Following contour point extraction, chains of contour points are determined to form possibly closed contours. This latter are then reduced to a minimal number of vertices according to a specified tolerance. Finally, only the polygons containing the “good” marker are selected to compute the camera pose.
The processing steps of AR marker detection procedure can be summarized as follows:
The library main classes are divided as following:
Tested with Microsoft Visual C++ (Express Edition 2010) and gcc 4.6.3
A video, with correct calibration camera, to test samples if you don't have a camera
FirstApp circle corners of "good" tag.
MultiTracking use ARAM abilities to track a chessboard of tag.
This work is contributed by Alexandre KORNMANN in the framework of his final project assignment in Altran Research Medic@ under the responsibility of Dr. Abdelkrim BELHAOUA. Altran Research Medic@ is research program aimed to integrate innovative augmented reality technology in order to improve the medical operating services.
Please report any bugs, comment about error and/or suggest enhancements to suppo rt-m edica @alt ran.c om