Generating Graph from 2D Flowchart using Region-Based Segmentation
C. Leena, M. Ganesh · 2020 IEEE International Students' Conference on Electrical,Electronics and Computer Science (SCEECS) · 2020
Visual content-based information retrieval intended message retrieval from infographics and their applications are gaining attention in many scientific domains. This seeks an understanding of the structure and correct semantics of infographics. Parsing flowchart and analyzing each element of infographics in the correct sequence is challenging. Using computer vision techniques, reconstruction of the structure in digital format is possible and can help to understand the features of the structure. This paper describes technical details to extract various elements of a 2D flowchart in a meaningful sequence, relate elements as connected in a flowchart and represent in conventional data structures. Region-based segmentation with adaptive thresholding and morphological operations are applied for feature extraction. The graph is generated by maintaining the semantics of the flowchart. The experimental results are evaluated for semantic correctness of flowchart and found that 90% of generated graphs are semantically correct as a flowchart.