Abstract: “In this paper, the approach of transformation an image of UML activity diagram to its XMI data format is presented. The images of UML activity diagrams contain three major components, that is, Diagram Layout, Diagram Flow and Data Text. In order to recognize these three components, the modules of OCR, Object detection, and Line detection will be executed. Any text in UML activity diagram images are captured by OCR. On the other hand, the activity diagram elements and arrows are analyzed by the Object and Line detection. Once all the objects and their relations have been detected, an XMI file shall be created. The prototype software called IMG2XMI has been implemented in this research work. It uses an open source library for helping the part of image processing. The proposed approach is also useful for the transformation of other UML diagram images to the XMI format in order to enhance the working environment independence.”

Keywords: image processing, UML activity diagram, XMI, software process improvemen