Identifying hand drawn tables

A non-transitory computer readable medium (CRM) storing computer readable program code for processing an image comprising a table embodied therein that:

• identifies a plurality of connected components in the image that form the table;
• generates, for the table, a skeleton graph comprising a plurality of edges based on the plurality of connected components;
• calculates a typesetness score that compares the table to a template table based on the plurality of edges;
• determines a rotation angle for the table and generates a custom coordinate system for the table based on the rotation angle;
• identifies a plurality of cells in the table based on the skeleton graph, the typesetness score, and the custom coordinate system;
• links each of the plurality of cells based on the plurality of edges and validates a link between the plurality of cells;
• generates a grid of cells based on the link between the plurality of cells and comprising the plurality of cells;
• exports the grid as a high-level representation of the table.

A system for processing an image comprising a table, comprising:

• a memory;
• a computer processor connected to the memory that:
• identifies a plurality of connected components in the image that form the table;
• generates, for the table, a skeleton graph comprising a plurality of edges based on the plurality of connected components;
• calculates a typesetness score that compares the table to a template table based on the plurality of edges;
• determines a rotation angle for the table and generates a custom coordinate system for the table based on the rotation angle;
• identifies a plurality of cells in the table based on the skeleton graph, the typesetness score, and the custom coordinate system;
• links each of the plurality of cells based on the plurality of edges and validates a link between the plurality of cells;
• generates a grid of cells based on the link between the plurality of cells and comprising the plurality of cells;
• exports the grid as a high-level representation of the table.