New Sequential and Parallel Derivative-Free Algorithms for Unconstrained Minimization
SIAM Journal on Optimization
Automatic morphological detection of otolith nucleus
Pattern Recognition Letters
Hi-index | 0.00 |
We present a semi--automatic system that detects and counts the yearly growths rings of cod otoliths. The system is based on morphing an angular section of the otolith to a rectangular region where the vertical directions follow B--splines that we place with an optimization algorithm such that they cross the rings in a close to perpendicular manner. The rectangular area is treated with standard Fourier techniques and classical filters. We obtain a large number of intensity profiles which we further analyze to count the annual rings. The preliminary results achieved on a small subset of our large database are encouraging. The manual steps of the system are easy to be performed automatically as well, however, we postponed their implementation concentrating at the beginning on the global system design.