The Project

Printers’ ornaments are the decorative features of books printed in the hand press period. Fleuron includes images of all kinds of printers’ ornaments, from those hand-cut in blocks of wood or metal, to cast ornaments, and the decorative pieces of type known as fleurons (our namesake). The database also includes engraved ornaments and illustrations. Full-page illustrations are excluded, but these can be found on Eighteenth-Century Collections Online . You can find out more about the different types of printers’ ornaments and their history in the Glossary. Fleuron was created from 32 million page images provided by Gale-Cengage Learning. The page images are originally from Eighteenth-Century Collections Online .

We developed a program to recognise printers’ ornaments and extract them from the full page images into a new database. The program was developed by Machine Doing Ltd. with the support of Research Software Engineering at the University of Cambridge.

The Code for Ornament Extraction

There are different ways to extract printers ornaments each with their own trade-offs. The approach adopted here is a morphological one. A series of morphological operations (e.g. filtering, dilution, erosion, etc) is applied on each image followed by a series of heuristics to filter out those connected components that are deemed to be printers ornaments. This is possible because ornaments do not occur randomly on the page but have clear structure. For example they are typically horizontally centered with the text and surrounded by whitespace or serve as capitals with clear bounds on aspect ratio. This helps distinguish them from illustrations and blobs of text that end up glued together.

That being said, the approach is by no means fool proof and will make mistakes, particularly where the source image is degraded or the morphology of the ornament closely resembles the morphology of the text. It should, however, serve as a useful baseline to help bootstrap more advanced machine learning based methods with less magic numbers.

Using The Database

You can browse or search the ornaments in the database by a number of criteria, including date, author, printer/publisher, place, and subject, as well as searching for ornaments according to their dimensions. If you want to navigate straight to a specific book, you can look it up using its unique English Short Title Catalogue (ESTC) number. If the book you’re searching for doesn't appear, it may be that it does not contain any ornaments — not all books do. Each ornament has a unique and searchable identification number, which you can make a note of if you want to return to an ornament later.


Hazel Wilkinson - Principle Investigator

I am a Junior Research Fellow in English at Fitzwilliam College, University of Cambridge. My research concerns the history of the printing trade in the eighteenth century. I spent the three years of my Ph.D attempting to identify the printers of eighteenth-century editions of Spenser by tracing their ornaments. This labour intensive process led me to begin work on Fleuron in 2013/14, in an attempt to make the process of finding and comparing printers’ ornaments more efficient.

Dirk Gorissen - Software Engineer and Computer Vision Expert

Dirk Gorissen has a background in Computer Science, Artificial Intelligence and Computational Engineering. He has worked in academic and commercial research labs in the US, Canada, Europe & Africa and his interests span machine learning, robotics, and computational engineering. He has also been a regular consultant for the World Bank and closely involved with a number of Drone related startups. He currently is a senior engineer in autonomous vehicle startup Oxbotica and on the side is an active STEM Ambassador, organiser of the London Big-O Algorithms & Machine Learning meetups, and is active in the Tech4Good / ICT4D space. Homepage.

James Briggs - Research Software Engineer and Web Developer

James is a Research Software Engineer with the Research Computing Service at the University of Cambridge. He has a MPhys in Computational Physics from the University of Edinburgh and has a research background in Astrophysics and High Performance Computing. His interests are in Big Data engineering, machine learning, and web development. He previously worked at Stephen Hawking Centre for Theoretical Cosmology as a software developer and High Performance Computing specialist.

Filippo Spiga - Research Software Engineer

Filippo leads a team of Research Software Engineers at the University of Cambridge working on several software projects across various Schools and Departments, from Bioinformatics to Engineering to Social Sciences, pursuing the mission of "Better Software for Better Research". He has a Computer Science degree from the University of Milano-Bicocca, obtained completing completing a thesis on hybrid MPI+OpenMP parallelism during a visiting period at the Edinburgh Parallel Computing Centre (EPCC).

Starting from August 2013 he has also been appointed as one of the Directors of the Quantum ESPRESSO Foundation.

The Future

Removing False Positives

Currently, the database contains some images of chunks of text, ink blots, manuscript notes etc. These have been mistakenly identified as ornaments by our ornament detection software. We are in the process of refining the software and cleaning these up, and they will gradually disappear!

Image Matching / Duplicate Detection

We will soon be adding an “image match” function, which will allow you to find similar or identical ornaments within the database at the click of a button.


The creation of Fleuron was generously sponsored by The Bibliographical Society’s Katherine F. Pantzer Jr Research Fellowship. Additional funding was received from the School of Humanities and Social Sciences at the University of Cambridge.

The data for the project was provided by Gale Cengage Learning. Technical support and equipment was provided by University Information Services, Cambridge.