CNCSIS grant CODE 16, type TD, Title: Robust techniques in image registration
Based on this method, the project proposes the use of image registration techniques to solve the node localization problem in a wireless sensor network based on video sensors. Moreover, the proposed solution adds the video-field overlap estimation for classical spatial localization. Several types of registration algorithms are analyzed and tested for performance evaluation.
This project comes under the Agreement n° 764672, concluded between the Coordinator and the DG General Justice and Consumers.
In this project, a new feature method based on robust 2D registration images is proposed. Image distortion is modeled as a four-parameter-like transformation, sequentially estimated by 1D transformations, resulting in increased density compared to 4D spatial processing. By adopting a mean shift estimator, the advantages of RANSAC and M-estimators can be combined in a unique and sound theoretical framework.
ACTIVITIES AND RESULTS:
- Choosing a mobile device to implement and test the concept.
- Choosing one of the existing courses.
- Development of an eLearning and mLearning platform
- Testing the platform on a group of students.
- Conclusions and other directions.
PROJECT DATA
Director
PhD student Daniela Fuiorea Bulucea
Value
5,210 RON