id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
3670	Praczyk, Piotr Adam; Nogueras-Iso, Javier	Automatic Extraction of Figures from Scientific Publications in High-Energy Physics	2013-12-22	28	.pdf	application/pdf	11209	860	53	results of the algorithm consist of metadata, raster images of a figure, but also vector graphics, several methods of automatically extracting and processing graphics appearing in PDF documents. present a page box-cutting algorithm for the extraction of tables from PDF documents.12 single operator does not trigger rendering of elements from different detected entities (figures, Graphical areas detected by a simple clustering usually do not directly correspond to figures. number of graphical and textual operations in the content stream of a figure candidate. 2 shows a fragment of a publication page with indicated text areas and final figure candidates A fragment of the PDF page with boxes around every detected text area and each figure During the first step of the caption detection, all text clusters from the publication page are tested Matching between figure candidates and captions happens at every document page separately.	./cache/3670.pdf	./txt/3670.txt