MANIFOLD RANKING ON MULTIPLE LOW-LEVEL FEATURE SET NORMALIZED WITH OPTIMIZED PARAMETERS IN CBIR

136 views

Abstract

In CBIR, the image is represented by multi low-level features that describe the color, texture and shape of the image. The combination of different image features in global similarity measurements such as the EMR requires normalized data sets. In this paper, a new normalization method for vector number data such as the low level features of color images is proposed. Experimentation has shown the effectiveness of the proposed algorithm for the manifold ranking EMR, and the CBIR quality is really improved.