Recently, feature-based tracking has received major attention in the research community. It requires a continuous extraction of stable features in the camera images. This paper presents NMF-SIFT, an extension to the well-known SIFT (Scale Invariant Feature Transform) algorithm based on NMF (Non-negative Matrix Factorization). A preliminary study, which compares NMF-SIFT to three standard algorithms for feature extraction (Harris, Kanade-Lucas-Tomasi, and SIFT) in bronchoscopic images, shows that the complexity of the SIFT keypoint descriptor can be reduced by NMF-SIFT while keeping a high number of keypoints for each image.     «

Recently, feature-based tracking has received major attention in the research community. It requires a continuous extraction of stable features in the camera images. This paper presents NMF-SIFT, an extension to the well-known SIFT (Scale Invariant Feature Transform) algorithm based on NMF (Non-negative Matrix Factorization). A preliminary study, which compares NMF-SIFT to three standard algorithms for feature extraction (Harris, Kanade-Lucas-Tomasi, and SIFT) in bronchoscopic images, shows that...     »