Intelligent nested tables detection and text extraction from images and scanned documents
Methods, systems, and apparatus are provided for detecting and extracting data from a nested table. A contour image of a first table disposed within a cell of a second table from an input image is identified. The first table on the input image is masked to generate a masked image of the second table that omits the first table. Cells of the first table and the second table are detected by identifying a plurality of regions of interest (ROIs), determining a row of a plurality of cells based on a common y-coordinate among of the plurality of ROIs, and generating cell parameter information for each of the plurality of cells that identifies the y-coordinate, a size, and a relative position of each determined cell. An output file is generated based on the cell parameter information of the plurality of cells.
1 . A method for detecting and extracting data from a nested table, comprising:
identifying a contour image of a first table disposed within a cell of a second table from an input image;
masking the first table on the input image to generate a masked image of the second table that omits the first table;
detecting cells of the first table and the second table, comprising:
identifying a plurality of regions of interest (ROIs),
determining a row of a plurality of cells based on a common y-coordinate among of the plurality of ROIs, and
generating cell parameter information for each of the plurality of cells that identifies the y-coordinate, a size, and a relative position of each determined cell; and
generating an output file based on the cell parameter information of the plurality of cells.
2 . The method of claim 1 , further comprising:
preprocessing the input image to generate a stencil image consisting of horizontal and vertical lines that represent a layout of the first table and the second table.
3 . The method of claim 2 , wherein
the preprocessing further includes applying adaptive thresholding to generate the stencil image with only two contrasting colors, and
the method further includes:
identifying a plurality of contours of the stencil image.
4 . The method of claim 1 , further comprising:
calculating an average height of the plurality of ROIs;
determining a height of an ROI of the plurality of ROIs; and
comparing the average height with the determined height of the ROI.
5 . The method of claim 4 , further comprising:
identifying a cell having a merged row in response to the average height being less than the determined height of the ROI.
6 . The method of claim 5 , wherein
the cell parameter information for each determined cell comprises:
an end row value, and
a cell height value; and the method further comprises:
determining the end row value of the cell having the merged row by identifying another determined cell having an identical sum of the y-coordinate and the cell height value.
7 . The method of claim 6 , wherein
the end row value of the cell having the merged row is the same as the end row value of the another determined cell having the identical sum of the y-coordinate and cell height value.
8 . The method of claim 4 further comprising:
identifying a cell not having a merged row in response to the average height not being less than the determined height of the ROI; wherein
the cell parameter information for each determined cell comprises:
a start row value,
an end row value, and
the method further comprises:
determining the end row value of the cell not having the merged row to be the same as the start row value.
9 . The method of claim 1 , further comprising:
determining a binary target variable indicating whether the input image is suitable based on:
evaluating a plurality of image features,
optically recognizing text within the input image, and
performing a spell check of the optically recognized text.
10 . The method of claim 1 , wherein the output file comprises:
a first sheet generated based on the cell parameter information associated with the first table; and
a second sheet generated based on the cell parameter information associated with the second table.
11 . A system comprising:
a processor;
a memory in communication with the processor storing instructions executable by the processor to configure the system to:
identify a contour image of a first table disposed within a cell of a second table from an input image;
mask the first table on the input image to generate a masked image of the second table that omits the first table;
detect cells of the first table and the second table based on:
an identification of a plurality of regions of interest (ROIs),
a determination of a row of a plurality of cells based on a common y-coordinate among of the plurality of ROIs, and
cell parameter information for each of the plurality of cells that identifies the y-coordinate, a size, and a relative position of each determined cell; and
generate an output file based on the cell parameter information of the plurality of cells.
12 . The system of claim 11 , wherein the memory further comprises instructions executable by the processor to configure the system to:
preprocess the input image to generate a stencil image consisting of horizontal and vertical lines that represent a layout of the first table and the second table.
13 . The system of claim 12 , wherein
the preprocess further includes an application of adaptive thresholding to generate the stencil image with only two contrasting colors, and
the memory further comprises instructions executable by the processor to configure the system to:
identify a plurality of contours of the stencil image.
14 . The system of claim 11 , wherein the memory further comprises instructions executable by the processor to configure the system to:
calculate an average height of the plurality of ROIs;
determine a height of an ROI of the plurality of ROIs; and
compare the average height with the determined height of the ROI.
15 . The system of claim 14 , wherein the memory further comprises instructions executable by the processor to configure the system to:
identify a cell having a merged row in response to the average height being less than the determined height of the ROI.
16 . The system of claim 14 , wherein the memory further comprises instructions executable by the processor to configure the system to:
identify a cell not having a merged row in response to the average height not being less than the determined height of the ROI; wherein
the cell parameter information for each determined cell comprises:
a start row value,
an end row value, and
the memory further comprises instructions executable by the processor to configure the system to:
determine the end row value of the cell not having the merged row to be the same as the start row value.
17 . The system of claim 15 , wherein
the cell parameter information for each determined cell comprises:
an end row value, and
a cell height value; and
the memory further comprises instructions executable by the processor to configure the system to:
determine the end row value of the cell having the merged row by identifying another determined cell having an identical sum of the y-coordinate and the cell height value.
18 . The system of claim 17 , wherein
the end row value of the cell having the merged row is the same as the end row value of the another determined cell having the identical sum of the y-coordinate and cell height value.
19 . The system of claim 11 , wherein the memory further comprises instructions executable by the processor to configure the system to:
determine a binary target variable indicating whether the input image is suitable based on:
an evaluation of a plurality of image features,
optically recognized text within the input image, and
a spell check of the optically recognized text.
20 . The system of claim 11 , wherein the output file comprises:
a first sheet generated based on the cell parameter information associated with the first table; and
a second sheet generated based on the cell parameter information associated with the second table.