siftCU: An Accelerated Cuda Based Implementation of SIFT

Scale Invariant Feature Transform (SIFT) is a popular image feature extraction algorithm. SIFT’s features are invariant to many image related varia-bles including scale and change in viewpoint. Despite its broad capabilities, it is computationally expensive. This characteristic makes it hard for researchers to use SIFT in their work especially in real time application. This is a common prob-lem with many image-processing related algorithm. Utilizing graphical pro-cessing unit (GPU) high computing power through parallel programming is an affordable solution for this issue. In this paper we present a GPU-based imple-mentation of SIFT using Compute Unified Device Architecture (CUDA) pro-gramming framework. We compare our CUDA-based implementation, namely siftCU, with CPU-based serial implementations of SIFT both in feature matching accuracy and time consumption.