A PREDICTIVE FUNCTION OPTIMIZATION ALGORITHM FOR MULTI-SPECTRAL SKIN LESION ASSESSMENT
| Affiliation auteurs | !!!! Error affiliation !!!! |
| Titre | A PREDICTIVE FUNCTION OPTIMIZATION ALGORITHM FOR MULTI-SPECTRAL SKIN LESION ASSESSMENT |
| Type de publication | Conference Paper |
| Year of Publication | 2015 |
| Auteurs | Li C, Balla-Arabe S, Brost V, Yang F |
| Conference Name | 2015 23RD EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO) |
| Publisher | EURECOM |
| Conference Location | 345 E 47TH ST, NEW YORK, NY 10017 USA |
| ISBN Number | 978-0-9928-6263-3 |
| Mots-clés | Embedded System, FPGA, genetic algorithm, High-Level Synthesis, High-Performance Computing, Kubelka-Munk model, Light-Tissue Interaction, Multi-spectral Image Processing, POSIX Thread, SW/HW Co-design |
| Résumé | The newly introduced Kubelka-Munk Genetic Algorithm (KMGA) is a promising technique used in the assessment of skin lesions. Unfortunately, this method is computationally expensive due to its function inverting process. In the work of this paper, we design a Predictive Function Optimization Algorithm in order to improve the efficiency of KMGA by speeding up its convergence rate. Using this approach, a High-Convergence-Rate KMGA (HCR-KMGA) is implemented onto multi-core processors and FPGA devices respectively. Furthermore, the implementations are optimized using parallel computing techniques. Intensive experiments demonstrate that HCR-KMGA can effectively accelerate KMGA method, while improving its assessment accuracy as well. |