Line Removal

Can be one of the following types:

The 'Dropout Method' property determines the technique used to remove or replace regions of the image identified by the line mask.

• Choose from methods such as filling with a solid color (e.g., white or background), background interpolation, or custom fill logic.
• The selected method affects the visual result and may impact downstream OCR or data extraction.

Example

• To fill removed line regions with white, select the "White Fill" option.
• To blend with the background, select "Background Fill".
• For forms where lines overlap text, consider using interpolation to preserve underlying content.

Guidance

• Review the output and diagnostic images to ensure that the dropout method produces the desired result for your document type.
• The choice of dropout method can affect OCR accuracy and the appearance of the cleaned image.
• For most standard forms, "White Fill" is sufficient. For complex backgrounds, "Background Fill" or interpolation may yield better results.

The 'Minimum Line Length' property defines the shortest allowable length for a feature to be considered a line. This property accepts unit-aware values for both horizontal and vertical lines (e.g., 0.2in, 10px, or 5pt).

• Set higher values to ignore short marks, stubs, or noise that may be present on forms.
• Set lower values to detect shorter lines, such as underlines, checkboxes, or grid stubs.
• Configure horizontal and vertical values independently to accommodate forms with different line structures.

Example: For forms with long table lines, set to 0.2in. For forms with short checkboxes, set to 4pt.

Tuning Guidance:

• Increase to avoid detecting small marks as lines.
• Decrease to ensure short but important lines are detected.
• Review diagnostic images to verify that only intended lines are detected and removed.

The 'Maximum Line Thickness' property sets the thickest a feature can be to be considered a line. Thickness is measured perpendicular to the line's orientation (e.g., vertical thickness for horizontal lines).

• Set lower values to detect only fine or single-pixel lines.
• Set higher values to allow detection of bold, double, or shaded lines.

Example: For standard forms, use 4pt. For bold or shaded lines, increase to 6pt or higher.

Tuning Guidance:

• Increase if thick lines are not being detected.
• Decrease to avoid detecting boxes or filled shapes as lines.
• Review the "Trim Mask" diagnostic to see the effect of this setting.

The 'Maximum Line Gap' property defines the largest break (in display units, e.g., 1pt) that can occur within a line while still treating it as a single, continuous line.

• Increase to allow detection of dashed, broken, or degraded lines as single lines.
• Decrease to require lines to be more solid and uninterrupted.

Example: For forms with faint or dashed lines, set to 2pt. For clean, solid lines, use 1pt or less.

Tuning Guidance:

• Increase for noisy or degraded images where lines may be broken.
• Decrease to avoid merging separate features into a single line.
• Use diagnostic images to verify the effect on line continuity.

The 'Minimum Aspect Ratio' property sets the minimum ratio of length to thickness for a feature to be classified as a line. For horizontal lines, this is width divided by height; for vertical lines, height divided by width.

• Higher values require lines to be longer and thinner, filtering out boxes or blobs.
• Lower values allow detection of shorter or thicker lines.

Example: A value of 10 requires a line to be at least 10 times longer than it is thick.

Tuning Guidance:

• Increase to avoid detecting boxes or thick marks as lines.
• Decrease to allow shorter or thicker lines.
• Use the visual example above to understand the effect of aspect ratio.

The 'Minimum Line Fill' property defines how solid a feature must be to be detected as a line. Fill percentage is the ratio of black pixels along the line's path to the total possible pixels.

• Set to 1.0 (100%) to detect only solid, unbroken lines.
• Lower values allow detection of dotted, dashed, or faint lines.

Example: Set to 0.75 (75%) to allow for some gaps or noise along the line.

Tuning Guidance:

• Increase to require more solid lines and reduce false positives from noise.
• Decrease to detect faint, broken, or dashed lines.
• Review diagnostic images to ensure the setting matches your document's line quality.

The 'Maximum Edge Noise' property defines the highest percentage of non-line (noise) pixels permitted along the edge of a line for it to be considered valid. Edge noise typically results from specks, text overlap, or scanning artifacts adjacent to lines.

• Set lower values to require cleaner, more distinct lines.
• Set higher values to allow detection of lines with more adjacent noise or touching marks.

Example: For clean forms, use 0.65 (65%). For noisy or degraded images, increase to 0.8 (80%) or higher.

Tuning Guidance:

• Decrease to avoid detecting lines with significant edge artifacts.
• Increase to detect lines in noisy or low-quality scans.
• Review diagnostic images to assess the effect on line detection and false positives.

The 'Trim Distance' property sets the radius (in display units, e.g., 2px) around each detected line within which connected components (such as text or marks) will be trimmed or removed. This helps ensure that lines are cleanly removed from the image without leaving behind fragments or damaging intersecting features.

• Increase to remove more of the area around lines, which can help with thick or fuzzy lines.
• Decrease to minimize the impact on nearby features, preserving intersecting text or marks.

Example: For standard forms, use 2px. For bold or wide lines, increase to 4px.

Tuning Guidance:

• Review the "Trim Mask" diagnostic image to visualize the effect of this setting.
• Adjust to balance thorough line removal with preservation of important intersecting features.

The 'Minimum Run Length' property defines the shortest continuous sequence of black pixels (in display units, e.g., 0.2in) that will be considered as part of a line during detection. This helps filter out small marks, specks, or noise that do not form meaningful lines.

• Increase to require longer uninterrupted runs, reducing false positives from short marks.
• Decrease to detect shorter lines or underlines.

Example: For forms with long table lines, use 0.2in. For forms with short underlines, reduce to 0.1in.

Tuning Guidance:

• Adjust based on the typical line length in your documents.
• Use diagnostic images to verify that only intended lines are detected.

Can be one of the following values:

The HoughLevel enumeration controls the use and precision of advanced line detection based on the Hough transform. When enabled, the Hough transform is used to detect faint, skewed, or non-axis-aligned lines that may not be found by standard morphological methods. Higher accuracy levels increase the sensitivity and precision of line detection, but may also increase processing time.

Use advanced line detection for documents with faint, broken, or slightly skewed lines, or when standard detection misses important features. Choose the lowest level that provides reliable results for your document set to balance accuracy and performance.

• 'Off': Disables advanced line detection; only standard methods are used.
• 'Low', 'Medium', 'High': Enable the Hough transform with increasing accuracy and computational cost.

Review diagnostic images to assess the effect of each level and tune for your specific needs.

The 'Angle Tolerance' property sets the allowable angular deviation from 0° (horizontal) or 90° (vertical) for detected lines. This is useful for handling forms that are slightly skewed or scanned at an angle.

• Lower values require lines to be nearly perfectly horizontal or vertical.
• Higher values allow detection of lines with more skew.

Example: For well-aligned forms, use 5.0 degrees. For skewed or hand-filled forms, increase as needed (up to 45°).

Tuning Guidance:

• Increase if lines are missed due to minor skew or rotation.
• Decrease to avoid detecting diagonal or unintended lines.
• Review diagnostic images to ensure the setting matches your document alignment.

Can be one of the following types:

The Dash Detector enables Grooper to recognize dashed lines—such as those used for signature lines, table borders, or form separators—by analyzing the spatial arrangement and characteristics of dash-like marks in the image.

This class is typically used as part of an image preprocessing pipeline, where it supplements standard line detection by identifying lines that are not solid but are instead composed of repeated dash segments. The detection process is configurable, allowing you to control the minimum number of dashes required to form a line and the maximum allowable gap between dashes.

How Dash Detection Works

The Dash Detector scans the image for clusters of dash-shaped blobs that are sufficiently aligned and spaced to be considered a line. It evaluates both horizontal and vertical orientations, filtering out noise and short or irregular sequences. When a valid dash sequence is found, it is treated as a line for the purposes of downstream processing, such as table extraction or form field recognition.

Configuration Guidance

• Use the 'Minimum Segment Length' property to specify how many dashes must be present in a sequence for it to be considered a line. Increasing this value reduces false positives from short or accidental dash groupings.
• Use the 'Maximum Dash Spacing' property to control how much space is allowed between dashes. This helps distinguish between intentional dashed lines and scattered marks.

The Dash Detector is most effective on documents where dashed lines are used for structure or guidance, and where consistent dash size and spacing can be expected. For best results, ensure that the image is properly binarized and free of excessive noise before applying dash detection.

Can be one of the following values:

The 'Comb Removal' property controls whether and how combs are detected and removed from the image. A comb is a short line segment (such as a checkbox, tick mark, or grid stub) that is connected to a perpendicular line at one or both ends. Removing combs is important for cleaning up forms with checkboxes, comb fields, or dense table grids, where these features can interfere with data extraction or be mistaken for field boundaries.

The 'Minimum Length' property sets the shortest length (in display units, e.g., "4pt") for a line segment to be classified as a comb. This value must be less than the 'Minimum Line Length' property, as combs are defined as short lines that do not meet the main line length threshold but are connected to a perpendicular line.

• Set this value to match the size of checkboxes, tick marks, or comb field stubs on your forms.
• If set too low, noise or specks may be misclassified as combs; if set too high, valid combs may be missed.

Example: If 'Minimum Line Length' is "0.2in", set 'Minimum Length' to "4pt" to target small checkboxes or grid stubs.

The 'Minimum Fill' property defines how solid a rectangle must be to be classified as a comb. Fill percentage is the ratio of black pixels to the total area of the rectangle. A perfect comb (such as a solid vertical bar) has a fill of 100%.

• For single-connected combs, the fill must meet or exceed this value.
• For double-connected combs, the fill must be at least half this value.

Use this property to distinguish combs from other shapes, such as parentheses or the letter "I", which may fit the size requirements but have a lower fill percentage.

Example: Set to 0.8 (80%) to require combs to be mostly solid, reducing false positives from hollow or irregular shapes.

The 'Quiet Zone Size' property defines a region on each side of a single-connected comb that must be relatively free of feature pixels. This helps prevent skinny vertical characters (such as a lower-case "j" with an underline) from being mistaken for combs.

The quiet zone is visualized in the following example:

◄──│──►
───┴───

Tuning Guidance:

• Increase the quiet zone size to require more whitespace around combs, reducing false positives.
• Decrease if valid combs are being missed due to nearby text or lines.
• Review diagnostic images to see the effect of this setting.

Example: Use the provided SVG and text examples to visualize the quiet zone in your forms.

The 'Minimum Weight' property works with the quiet zone to ensure that the area around a comb is sufficiently clear of other features. In the region that includes the comb and its quiet zone, the percentage of feature pixels belonging to the comb must meet or exceed this value. A value of 100% requires a completely clear quiet zone.

• Increase this value to require stricter separation between combs and surrounding features.
• Decrease if valid combs are being missed due to nearby marks or noise.

Example: Set to 0.9 (90%) to require the quiet zone to be almost entirely clear, minimizing false positives from adjacent text.

The 'Maximum Speck Size' property sets the largest size (in display units, e.g., "1pt") for artifacts (specks) that touch the edge of a line and should be removed during comb removal. Specks are small, isolated marks or noise that can interfere with line detection and form analysis.

Specks must have an aspect ratio less than or equal to 3. In the figure below, the solid black pixels outside the line (outlined in red) are artifacts that are touching the line.

Best Practices:

• Set this value to match the size of noise or marks you want to remove.
• Increase to remove larger artifacts; decrease to preserve small marks that may be important.
• Use the SVG diagnostic to visualize the effect of this setting on your forms.

Binarization is the process of converting an image to pure black and white, which is a critical first step for accurate line detection. The 'Binarization Settings' property allows you to select the binarization method and configure its parameters to best suit your document type and image quality.

• Auto: Automatically determines the optimal threshold using Otsu's method. Recommended for most standard scanned documents.
• Adaptive: Calculates a local threshold for each pixel, making it robust to uneven lighting or background shading. Use for documents with variable backgrounds.
• Simple: Applies a fixed threshold. Use when you know the ideal threshold value for your images.

Tuning binarization is essential for reliable line detection. If lines are faint or broken, try Adaptive or adjust the threshold. If too much noise is present, increase the threshold or use Auto.

Best Practices:

• Review the 'Binarized' diagnostic image to ensure lines are clearly defined and separated from the background.
• Adjust binarization parameters if lines are missing or if noise is interfering with detection.
• For forms with shaded backgrounds or inconsistent lighting, Adaptive is often the best choice.

Font dropout removes text from the image before line detection, helping to prevent text from being mistaken for lines or interfering with detection. The 'Dropout Font Size' property sets the maximum font size (in display units, e.g., "14pt") to be dropped out.

• Set to a value matching the largest font size you want to remove. For example, "14pt" will remove all text up to 14 points in size.
• Leave empty or set to zero to disable font dropout.

Tuning Guidance:

• Use font dropout when text overlaps or touches lines, or when text is dense and interferes with detection.
• Review the "Preprocessed" diagnostic image to ensure that text is removed without damaging lines or boxes.
• If lines are being removed or damaged, reduce the dropout size or disable font dropout.
• For forms with large text or headings, set a lower value to avoid removing important structure.

Example: Set to "14pt" to remove standard form text, or "0" to disable and preserve all text.

The 'Command Type Name' property shows the user-friendly name of the IP Command as it appears in the Grooper UI. This helps users identify which command is being configured or executed, especially when working with IP Profiles containing multiple commands.

This property is read-only and is automatically set based on the command's type.

The 'Supported Pixel Formats' property lists the image pixel formats that the IP Command can process. If an image does not match one of these formats, the command will be skipped for that image.

• By default, all pixel formats are supported unless a derived command restricts this list.
• Example formats include Pixel8bppGrayscale, Pixel24bppBgr, and Pixel1bppIndexed.

Use this property to verify compatibility when building IP Profiles or troubleshooting why a command did not execute.