Spatial Frequency Domain Entropy Dissimilarity Measure

An entropy-based spatial frequency domain 2D image dissimilarity measure is proposed and demonstrated. The metric is an extension to pixel-pixel based entropy dissimilarity measures (EDM) proposed by Tsai and Wu. The new spatial frequency domain EDM maintained all desired features of EDM while overcoming limitations such as requiring the two digital images to be compared to have the same size in pixels. The research in inspired by the need of analyzing the Advanced LIGO time series data sets in developing a real-time and blind gravitational wave event detection algorithm. Compared to other dissimilarity measures, frequency domain entropy dissimilarity measure (FD-EDM) is spatial information sensitive and agnostic to image sizes and resolutions. It also further improves the computational efficiency of normal pixel-pixel EDM.