Abstract

Abstract—SIFT is a methodology for pattern recognition invariant to translation, rotation and scale. Due to its high efficiency and accessibility, SIFT has many applications in many fields such as image retrieval, image stitching, 3D modeling, machine vision and video tracking. After SIFT was published, it hasn?t stopped improving. The most important improved methodologies are PCA-SIFT, A-SIFT, C-SIFT, SURF and W-SIFT. In this paper, we first thoroughly analyze SIFT. Then we explain the variants, mentioning the deficiencies and improvements of each one. Finally some applications in different fields

Author-supplied keywords

• c-sift
• local features
• pca
• sift
• surf
• wavelets

Find this document

There are no full text links

Authors

Cite this document

Choose a citation style from the tabs below