SBIR-STTR Award

TMSSequoia proposes to develop a software system to allow conversion of paper-based legacy engineering drawings, technical manuals and reports, to a digital format. This process will include automatic extraction of data from the scanned images that is input into a Configuration Management Information System (CMIS) for storage, query and retrieval. TMSSequoia will then identify current software tools and/or applications that support this effort. TMSSequoia will then take the identified software and define how the available tools can be used to minimize any actual software development. Additional software development will be identified that brings current technologies together into a complete system for scanning of documents, identification of document type, extraction of index data from image , storage into CMIS, and ultimate retrieval and viewing. Much of the research in this project will be in the identification and evaluation of possible input data to ensure that proper extraction and indexing takes place.

Keywords:
DIGITAL AUTOMATION FORMS PROCESSING DOCUMENT RECOGNITION

Conversion of legacy drawings to usable product models is an important requirement for all the services. This is especially true of USAF which is trying to keep operational several time-tested airplanes. The DoD Automated Data Conversion System (ADCS) project addresses this issue. A key objective of this project has been the conversion of vectorized CAD drawings to feature based CAD models so that they can then be used to automatically generate NC machine code - in other words, help with automated manufacturing. Our proposal directly addresses this conversion. We have a unique approach in that we perform feature extraction in the 2D domain. Feature extraction is a crucial aspect of the conversion process and there has been little success within this is attempted directly in the 3D domain. Having performed many experiments with our approach we are confident in proposing our architecture called FlexiCAD to solve the conversion problem. Our Phase I proposal will complete an architectual validation study that will prove that when implemented it will a) semiautomatically correct errors in the vectorized data and b) automatically convert vectorized drawings to feature based output in a format that the Air Logistics Centers can use for automated production of spares.