Digitisation / Cataloguing of non-textual objects: A standardised and optimised process for data acquisition from digital images of herbarium specimens

点击翻译按钮获取中文摘要

The project will develop and document a software-driven standard process for extracting metadata from images of herbarium specimens (i.e. dried pressed plants or plant parts mounted on cardboard and stored in natural history collections). We will address a large proportion of science collections: approximately 22 million herbarium specimens exist as botanical reference objects in Germany, about 500 million worldwide. Metadata like plant name, collection site and date, collector, accession numbers, etc. are also glued flat on the sheet and thus visible on the specimen image. Up to now most of the data capture is manually fed into collection databases, but increasingly, imaging techniques are employed (also to ensure that the on-line metadata can be verified). The standard process shall replace or add to the manual data input as much as possible. Image processing software detects objects on the digitized record and classifies them. Text objects will be transformed into structured information using text mining algorithms. For handwriting, author identification is attempted. The project will evaluate and enhance existing software to conform to standard interfaces and integrate it into an overall open software architecture on the basis of established IT standards. Finally, the requirements for the process will be formulated as a standard and the actual application will be documented.