Comparative Analysis of Existing Similarity Measures used for Content-based Image Retrieval
Miroslav Marinov, Irena Valova, Yordan Kalmukov · 2019
Searching through millions or billions of images is a complex, computationally intense and time-consuming task, but it is the most important and desired feature of image databases. Search results should be sorted by similarity in descending order. The relevance is usually calculated by an appropriate similarity measure that fits the chosen type of metadata. This paper presents a comparative analysis of four universal similarity/distance measures and determines if they could be used for comparing and ranking images as well.