Indoor image matching algorithm based on improved SIFT algorithm

Abstract

Aiming at the indoor positioning by using SIFT algorithm, because the structure information is more or similar, and there is no obvious identification, which will lead to low accuracy and poor applicability, this paper proposes an improved SIFT algorithm for image matching. Choosing Junction feature which is more suitable for indoor structure, and combining with the improved CPW energy function method to eliminate mismatching points. By testing the actual image to compare and analyze get the feasibility and validity of the new algorithm.