作者: Scott Cohen , Leonidas Guibas
DOI: 10.21236/ADA358270
关键词:
摘要: The Earth Mover''s Distance (EMD) between two finite distributions of weight is proportional to the minimum amount work required transform one distribution into other. Current content-based retrieval in Stanford Vision Laboratory uses EMD as a common framework for measuring image similarity with respect color, texture, and shape content. In this report, we present some fast compute lower bounds on which may allow system avoid exact, more expensive computations during query processing. effectiveness tested color-based system. addition bound work, also show how under translation. problem, points are free translate, goal find translation that minimizes other distribution.