Column Labels and Formats - Designer - Foundation 23.1 - Foundation 23.1 - Brainware - external - Brainware/Designer/Foundation-23.1/Brainware-Intelligent-Capture-Designer/Data-Extraction/Table-Analysis-Engine/Column-Labels-and-Formats - 2024-01-22 - For each column, you must to specify a label that the engine uses to identify the column.

Brainware Intelligent Capture Designer

Platform
Brainware
Product
Designer
Release
Foundation 23.1
License

For each column, you must to specify a label that the engine uses to identify the column.

Optionally you can specify format strings to determine whether the column cells have a valid content.

In most cases, all column labels are positioned at about the same horizontal position, that means they belong to one text line. This line is called label line. In some cases, the labels extend over more than one text line. The engine can process single-line as well as multi-line labels.

The label line plays a central role for the table analysis, to detect the following information.

  • The start position of the table
  • The preliminary table layout and
  • The preliminary mapping of the table columns.

To properly assign a cell to a column, some degree of overlap with the label must exist. If a label overlaps with cells of different columns, the column is created using the cells with the most overlap.

Example

In the following table, the engine would not recognize the third column.

Note: Since the label detection is vital for table recognition, the OCR quality is very important to obtain good results. Normally, light labels on a dark background are more difficult to read than vice versa.