Word Spotting

locatedThis project started from the simple idea of trying my algorithms that I developed in the basic research of Computer Vision and Image Processing and which I have used for Image stitching, and see if they could be used for Word Spotting. These algorithms allows for a more relaxed matching, which is necessary to cope with the variations found in handwriting.

Semi-Automatic Transcription

a2This project is a continuation of the word spotting project and aims at making the transcription using word spotting. The idea is that the user should mark up the word that needs to be transcribed and then the word spotter finds all occurrences of that word. Hence, the word needs to be transcribed only once, and the process can be performed in arbitrary word order by multiple transcribers.

Making Collections Searchable

Berlin1This project too use the word spotting framework to make collections of documents searchable that has not yet been transcribed. Especially, if no annotated data exists, machine learning methods cannot be used, while the learning free word spotter can be. This still makes it possible to query documents and can be a useful tool for researchers in the field of humanities.

New Eyes on Sweden’s Medieval Scribes

Scribal Attribution using Digital Palaeography in the Medieval Gothic Script

DHIn the first paper about letter spotting we searched through the SDHK dataset consisting of almost 11 thousand medieval charters, trying to find which scribes used specific variants of the letters ‘f’,’g’ and ‘h’. The idea is to find certain documents that might have been written by the same scribe that wrote the example letters. The image shows the probability in a 3D space, where the magenta coloured spheres have the highest probability.

The second paper deals with certain script features such as core height compared to ascender and descender length. This work is still in progress.

islandThis paper present a visualisation framework for visualising handwritten script signs. Signs or characters with similar features forms distinct clusters that can be examined further. The visualisation have the look of an atlas over small islands that form each cluster.

 

 

From quill 2 bytes

q2b-whiteThis cross disciplinary initiative takes its point of departure in the analysis of handwritten text manuscripts using computational methods from image analysis and linguistics. It sets out to develop a manuscript analysis technology providing automatic tools for large-scale transcription, linguistic analysis, digital paleography and generic data mining of historical manuscripts. Our mission is to develop technology that will push the digital horizon back in time, by enabling digital analysis of handwritten historical materials for both researchers and the public.

  • Since January 2017 I take part in the following project within q2b: 2016 – 2021: Riksbankens Jubileumsfond, Jubileumsutlysningen Nya utsikter för humaniora och samhällsvetenskap, 13.4 MSEK. New Eyes on Sweden s Medieval Scribes. Scribal Attribution using Digital Palaeography in the Medieval Gothic Script. (Dnr NHS14-2068:1, PI Lasse Mårtensson).

GeoMemories

bannerI became part of the Geomemories project during my ERCIM Alain Bensoussan fellow period at IIT, CNR, Pisa in Italy. The problem of aerial photography matching and stitching became one of my main research interest for several years, resulting in many publications in the areas of key point detection, feature matching and outlier removal, etc.