Algorithms and VLSI Architectures for Pattern Recognition Based on the Gabor Wavelets
Résumé
In this paper, we present some pattern recognition algorithms based on the Gabor filtering. We then select an algorithm showing best performance and readiness for a hardware implementation. We focus this paper on the problem of algorithm/architecture adequation. We present a design methodology considering algorithm study and refining, hardware resources, data widths and final cost of the system implementing the considered algorithm. The whole system is simulated using a hardware/software cosimulation tool and will be implemented following a macro-block generator design strategy. This is of great importance as to the emergence of the new techniques of IP block design. The design will thus be portable, upgradable and parametrizable.