Image Retrieval using Spatial Chromatic Histograms
Patrick Lambert, Nathan Hervey, H. Grecu · Conference on Colour in Graphics Imaging and Vision · 2004
In this work, a new method for color image indexation is proposed. First the color space and the spatial domain are both projected on a 1D space and partitioned in a fuzzy way. Then, the 2D histogram of the membership degrees to these two partitions is computed. This histogram can be used to index each image of a large database. Therefore, retrieval is simply performed by computing the Euclidean difference between the 2D histogram of the query image and the 2D histogram of a database image. This method is applied on a real database.