Floating form processing based on topological structures of documents

A computer program product comprising a non-transitory computer readable storage medium having a computer readable program code embedded therein for controlling a data processing apparatus, the computer readable program code configured to cause the data processing apparatus to execute a process for processing a document image, which is an image of a filled form that corresponds to an empty form, the empty form having a plurality of form fields in a floating structure, the process comprising:

• obtaining a topological structure of the empty form, which includes, for each given text phrase among at least a subset of text phrases of the empty form, a vertical sequence containing text phrases in the empty form that have a vertically aligned relationship with the given text phrase, and a horizontal sequence containing text phrases in the empty form that have a horizontally aligned relationship with the given text phrase;
• obtaining a settings file for the empty form, which specifies text phrases of the empty form that are field names of form fields of the empty form, and for each form field, a positional relationship between the field name and a corresponding user input area;
• extracting and segmenting text phrases from the image of the filled form;
• based on the extracted and segmented text phrases, generating a topological structure of the filled form, which includes, for each given text phrase among at least a subset of text phrases in the filled form, a vertical sequence containing text phrases in the filled form that have a vertically aligned relationship with the given text phrase, and a horizontal sequence containing text phrases in the filled form that have a horizontally aligned relationship with the given text phrase;
• matching the text phrases in the empty form with segmented text phrases in the filled form by comparing the topological structure of the empty form with the topological structure of the filled form;
• based on the matching of the text phrases, extracting text phrases in the filled form that do not match any text phrases in the empty form as user-filled data;
• and
• associating each text phrases of the user-filled data of the filled form with a form field based on the settings file and positional relationship of the text phrase of user-filled data with text phrases in the filled form that match field names in the empty form.