Samenvatting

This report describes the development of a document recognition service. Goal of the service is the recognition of “Muster 4” prescriptions, as an extension for the HMM Deutschland GmbH product DeTouro. A detailed analysis has been conducted, based on which a number of requirements have been defined. To identify a suitable framework for the document recognition, a complex research has been conducted. Next to a number of premade services, an open source alternative, based on object and text recognition services, is also evaluated. The open source alternative failed to produce meaningful results.
The Form Recognizer service, offered by Microsoft as part of the Azure platform, has been identified as most suitable. Based on that the service has been recommended for the implementation. During the design phase the process, as well as the structure, of the service have been planned. Additionally, the authentication strategy is also explained. The implementation part describes the setup of Docker, as well as the configuration of Continuous-Integration- and Continuous-Deployment Pipelines, based on GitLab. After that the implementation of the API and the conversion of the recognition results is described. Additional further explanations for a prototype, for converting photos to scan-like images, have been provided.