No abstract

Abstract. This paper proposes a parallel hardware architecture for the scale-space extrema detection part of the SIFT (Scale Invariant Feature Transform) method. The implementation of this architecture on a FPGA (Field Programmable Gate Array) and its reliability tests are also presented. The obtained features are very similar to Lowe's. The system is able to detect scale-space extrema on a 320 × 240 image in 3 ms, what represents a speed up of 250x compared to a software version of the method.

No abstract

No abstract

No abstract