Application of SIFT Feature Combined with Global Characteristics in Image Retrieval
Hong Liang · Journal of Engineering of Heilongjiang University · 2010
A feature extraction algorithm of SIFT feature combined with global characteristics is proposed in this paper which can solve the problem of inefficacy feature description in objective feature.Firstly,the steps of constructing scale space,detecting extrema and confirming key-points are employed to create SIFT feature vector.Secondly,the global color characteristics is combined to SIFT feature to construct the union feature which makes the image content information more comprehensive.Thirdly,PCA dimensionality reduction algorithm is applied to reduce the dimension of union feature.Finally,the retrieval is implemented through distance match.The experiments indicate that a good application result of certain object description could be achieved in retrieval with the proposed algorithm.