A study of the impact of X-ray views and their number on the accuracy and robustness of 3D/2D registration, which enables the registration of 3D models and 2D images, was published in a serial SCI publication Springer - Lecture Notes in Computer Science.
COBISS.SI-ID: 6185044
A new robust registration method for geometrical registration of 2D and 3D images was developed. In the comparison with the existing methods, the proposed method was proven more robust and fast and, thereby, feasible for the registration of 3D models to 2D images of the inspected objects. The paper was published in a SCI journal with a high impact factor of 3.275, which ranks in the 4th place out of 227 journals in the field of electrical engineering.
COBISS.SI-ID: 6611540
A new method for the segmentation of tablet images was developed. The method is based on iterative border tracking that follows large intensity gradients and on robust mechanism for border contour verification. The results indicated that accurate and robust segmentation can be performed in less than 1.5 ms. The proposed method is under review in an SCI journal and it was installed in the industrial machine for tablet quality inspection.
In the year 2008, the post-doc project leader had successfully defended his second PhD thesis at the Utrecht University, The Netherlands.
COBISS.SI-ID: 238884608
A new robust registration method for geometrical registration of 2D and 3D images based on gradient reconstruction was developed. In the comparison with the existing methods, the proposed method was proven more robust and fast and, thereby, feasible for the registration of 3D models to 2D images of the inspected objects.
COBISS.SI-ID: 661077