Enhancing Scanned Images

Brief description: This section demonstrates ways of enhancing scanned monochrome images - deskewing, filtering, and eliminating distortions using calibration and four-point correction. Also, we give you some recommendations about restoring poor quality images.

Deskewing

When paper is fed through a scanner the direction of the paper often deviates from the orthogonal, therefore the resulting scanned image is skewed, sometimes rather considerably.
The special command - Deskew is used to deskew an image.
The command works in two modes:

  • Auto Deskewing - program automatically estimates the skew value, and then performs deskewing.
    Choose Deskew from the Image menu, and then -Auto.
  • Manual Deskewing - you can use this mode, if the result of Auto Deskewing is not satisfactory. In this mode you can specify a skew angle by specifying a line, which must be horizontal or vertical.
    Choose Deskew from the Image menu, and then -Manual.

Press the Measure Angle button in the displayed Deskew dialog, and then specify the deskew line with two points. You can use snap tools during this operation.
Deskew line (manual deskewing)
After deskewing

Filtering Monochrome Images

There are a number of different filters available within WiseImage, which allow you to considerably enhance the scanned monochrome raster image. You can apply filters to the whole image or to a fragment. You can also apply filters to several raster images simultaneously.

Using the filters, you can:

  • Remove raster speckles
  • Remove holes (white points in black raster lines or filling)
  • Invert images
  • Smooth raster objects
  • Thin and thicken raster objects
  • Transform filled raster area to one-pixel contour       

All commands for filtering monochrome images are located in the Filters menu.

Speckle Remover

Choose Speckle Remover from the Filters menu.

You can specify the size of raster speckle on the screen using the measure buttons of the Speckle Remover dialog.
If the Auto Estimating checkbox is on, the program automatically estimates the speckle size. You can apply the filter several times.

Before applying Speckle Remover

The result of applying Speckle Remover

When applying the filter to the image of very poor quality some data may be lost. To avoid it, you can use the operation of Separation by size.
See section Separating Monochrome Images by Size for detailed information.

Hole Remover

Choose Hole Remover from the Filters menu.

You can specify the size of raster speckle to remove on the screen using the measure buttons of the Hole Remover dialog box. If the Auto Estimating checkbox is on, the program automatically estimates the speckle size.

Before applying Hole Remover

The result of applying Hole Remover

Inversion

Choose Inversion from the Filters menu.

This filter reverses the color of the monochrome image: After applying this filter the background dots become the image dots and vice versa.

Before applying Hole Remover

The result of applying Hole Remover

Smoothing

Choose Smoothing from the Filters menu.

This filter smoothes raster object outlines, fills edge and inner background droplets, and partially removes raster speckles.

Before applying Smoothing

The result of applying Smoothing

The Smoothing filter is tuned by setting optimal values of Medianning and Threshold. You can estimate the result before applying the filter in the preview window.
Tuning Smoothing filter

Thinning and Thickening filters

Choose Thinning (Thickening) from the Filters menu.

Thinning filter makes raster objects thinner in the specified directions; by one pixel for one pass. You can specify a number of passes.

Before applying Thinning

The result of applying Thinning

If needed, you can apply the mode of thinning raster objects up to skeleton - i.e. only pixels in the middle are left.

Thickening filter makes raster objects thicker in the specified directions. You can specify a number of passes and choose any combination of thickening directions.

Contour (4-coupling and 8- coupling)

Choose Contour from the Filters menu.

Contour filters convert filled raster areas to one-pixel contours.

Before applying Contour

The result of applying Contour

Separating Monochrome Images by Size

This operation is effective in the following cases:

  • If you want to prevent important small objects from being removed while applying the Speckle Remover filter (e.g., dots in dashed-dotted lines; lines which consists of multiple small fragments, delimiters, etc).
  • If you work with an image of very poor quality.       

The principle of the separation by size operation is similar to speckle removing, except speckles are not removed but transferred to a new raster layer; and then the user can select fragments incorrectly transferred to this “speckle” layer and bring them back to the drawing.

Original image

The result of separating to basic and “speckle” layers
The significant elements have been brought back from the “speckle” layer to the basic layer; the “speckle” layer is removed

Calibration for eliminating linear and non-linear distortions using

Calibration - is applied to eliminate arbitrary (both linear and non-linear) distortions of monochrome, grey-scale and color raster images.

IMPORTANT: You should calibrate the raster image before vectorization (tracing), because if you vectorize a distorted image you will obtain distorted vector data that is impossible to correct later.

Calibration terms
The original image must contain points with known coordinates. These points are called real points. You can use nodes of a rectangular grid as real points (if you work with mapping materials) and/or other points, which coordinates are known or can be calculated.
The points, which correspond to the real ones on the distorted image, are often located with some displacement (i.e. they have different coordinates). These points are called measured point, as their coordinates value can be measured on the raster image.
By applying calibration, the raster image is transformed so that the measured points coincide with, or are extremely close to the real points.

TIPS: before applying calibration you should set the User Coordinate System by specifying its origin and scale.
See section Document Coordinat System for detailed information.

How to calibrate an image:

Choose Calibration from the Image menu.

Step 1. Defining real points
We can define rectangular grid nodes and points with known (estimated) coordinates.

Defining grid:
Grid is defined by the three parameters: start point (coordinates of lower-left node), cell size (by X and Y axis), and grid size (by X and Y axis).

  • Press Define Grid   on the Calibration toolbar. In the Calibration Grid dialog.
  • Specify the coordinates of the lower-left grid node in Origin.
    You can enter the coordinate values of the lower-left grid corner in the corresponding field or specify it with the cursor on screen.
  • Enter the number of grid cells along X and Y axis in Size field.
  • Enter the cell size along X and Y axis in Cell field.
  • If the grid is rotated about the coordinate axis, enter the skew angle relative to X-axis in Angle field.       

TIPS: Use the Preview button to make sure that the grid is properly specified.

Specifying points with known coordinates:

  • Press Add point   button from the Calibration dialog. Enter the point's known coordinates in Real field (use separating character between X and Y values).
    You can enter a name for the point in Label field; alternatively the program numbers the points consecutively.
  • Do the same for all points.      

Step 2. Specifying measured points

To begin step 2 you can press Next Point . WiseImage places the first point from the list in the screen center and highlights it with grips.

You see the real (correct) coordinates of the first point with grips.

Drag grips to the grid node on the scanned image - so you specify the “measured” point.

To shift to the next (previous) point, use Next (Previous) point.
You can also use the acceleration keys: N (Next) or Tab - to shift to the next point and P (Previous) - to the previous.

Step 3. Choosing calibration method

Choose the optimal method from the pull-down list of Calibration Method field.

What method to choose?

  • If you are sure of the proper method, choose it from the list. Estimate the accuracy of this method by pressing Estimate  button.
  • If you are not sure of the proper method, select Choose automatically from the combo-box of the calibration method

Step 4. Running Calibration

  • IMPORTANT: If you have several raster images in your document, select one(s) that need calibrating using the specified set of calibration pairs.
    See section Raster Image Selection for detailed information.
  • Run calibration by pressing the Apply button.
           

Four-Point Correction

You can use this operation for quick correction of scanned technical drawings, which contain a frame. This operation assumes that both frame and contents of the image have the same distortion.

How to correct an image by 4 points:

  • IMPORTANT: If you have several raster images in your document, select the one to calibrate by four points.
    See section Raster Image Selection for detailed information.
  • Choose 4 point correction from the Image menu.
  • Enter the frame size in the Width and Height fields of the 4 point correction dialog.
    You can also choose a standard frame size from the list of Frame Size or find the closest by pressing Find Closest paper.
    The list of standard formats is set in the Properties dialog. See information in section Customizing Program Parameters.
  • Press Find Frame  - the program finds and highlights the image frame.
    If the frame is defined correctly:
    • Press OK.
    Alternatively:
    • Press Measure frame  button, and then specify four points of the frame on the screen.
    • Press OK.
         

Autocorrecting Image

Autocorrect runs a set of predefined operations.

How to apply autocorrection:

  • Choose Preferences from the Tools menu.
  • Specify a set of commands to apply with autocorrection using the Preferences dialog. Close Preferences.
  • If you have several raster images in your document, select the one(s) to apply autocorrection to.
    See section Raster Image selection for detailed information.
  • Choose AutoCorrect from the Image menu.

NOTE: The parameters of the commands included in the autocorrection list (except Rotate) are automatically defined. For example, deskewing with auto definition of rotation angle, removing speckles with auto definition of speckle size, etc. If you want to specify the parameters of executed commands precisely, use the batch process.
See section Batch and Script for detailed information.

Editing raster text

You can edit raster texts -original text under the defined rectangle is erased and replaced with new text.

  • From the Draw menu, choose Raster, and then Edit Raster text.
  • Draw a frame over the text to be modified.
  • While drawing, you can specify the frame angle in the corresponding field of the Inspector.
  • Enter new text in the Text field of the Inspector.
    You can set text parameters in the corresponding fields of the Inspector.
Original raster text
Selected text area
After editing

Restoring Images of Poor Quality

How to work with a raster image of very poor quality:

There isn't much choice - either redraw it in the document with the corresponding program, or restore it with a hybrid editor (vectorization of such document produces poor results).

The purpose of this section is not to discuss the plus and minus points of these choices, but to provide you with some methods to restore scanned documents of poor quality.

  • Separating an image by size - can be used for blueprints, images on dark backgrounds, with lots of speckles.
    Detailed information is provided in section Separating Monochrome Images by Size.
  • Editing selected raster objects (modifying their properties, geometry using «grips» and others).
    Detailed information is provided in section Editing data, selected with object selection methods.
  • Tracing in smooth mode.
  • Tracing in vector producing mode. Editing and rasterizing the obtained vector object.
    Detailed information is provided in section Tracing Modes and Commands.
  • Drawing missing raster lines in Raster drawing mode.
    Detailed information is provided in section Raster Drawing Mode.
  • Editing raster texts.
    Detailed information is provided in section Editing raster text.

Example of restoring an image

Original image fragment
Choosing and editing the axis line
Result of editing the axis line
Tracing the curve in smooth mode
Hatch on the original raster image
Tracing the hatch in Create Vector and Erase Raster mode
Rasterizing the vector hatch

Then you can apply the following actions:
Edited raster image
Select the whole raster image (or fragment)
Delete the raster fragments you need from the selection
Remove the selected speckles

 

 [ welcome ]   [ hybrid graphics ]   [ quick start ]   [ tutorial ]   [ reference ]   [ about ] 
 
 [ Welcome ]   [ WiseImage for Windows ]   [ WiseImage for AutoCAD ] 

 
top