boe.pres.k12.wv.us/boe.pres.k12.wv.us.zip/boe.pres.k12.wv.us/Utilities/Xerox Workcentre 5335/Windows Scan/32-bit_x86/Francais/cpsimage.cab/data/xipProcs/hlExtract.proc

Text File | 2009-03-16 | 7KB | 208 lines

// Load scripted support procedures
#load "xipProcs/morphology.proc";

/* @XIPHighlightExtract
  // DESCRIPTION
  This procedure extracts highlighted text regions from a color image. It can
  extract text regions for multiple highlighter colors, and return each text
  extraction as a separate image or merge the extractions into a single image.
  The resulting image(s) will be returned in the ELFLIST object "output".

  // ARGUMENTS
  XIPIMAGE input      Image to extract highlighted text regions from

  INTEGER  merge      This specifies the output format:
                        0 to keep each text extraction as a separate image

                        1 to combine each text extraction image using the AND
                          operator into a single text extraction image

  // RETURNS
  ELFLIST output      This is a list of XIPIMAGE objects representing the
                      extracted highlighted text images. The info member of
                      each XIPIMAGE is a string indicating the highlighter
                      color(s) that covered the text.
*/
PROCEDURE XIPHighlightExtract (XIPIMAGE input, INTEGER merge)
  RETURNS (ELFLIST output)
{
  XIPIMAGE   imgBase, imgText, imgHL, imgHLText;
  STRING     colorname, str, tmp;
  STRING     se5x5, se7x1;
  INTEGER    i, skipOne, nOutputs;
  INTEGER    minlevel, maxlevel;
  INTEGER    x, y, w, h;
  ELFLIST    win;
  DOUBLE     avg;

  // Import IPCore definitions
  #import "ipcore2xip.ucm";

  // Create structuring element strings
  se5x5 = XIPMakeSolidStrel (width: 5, height: 5).toString ();
  se7x1 = XIPMakeSolidStrel (width: 7, height: 1).toString ();

  // Find highlighter color regions
  str = input.hlcinfo (
          ).toString (
          ).match (str: "color:"
          );

  // Process highlighter color regions
  skipOne = 0;
  for (tmp=str.getToken (); tmp; tmp=str.getToken ())
    if (tmp == "color:") {
      if (!skipOne) {
        str.getLine ();
        skipOne++;
      } else {
        // Parse highlighter color name
        colorname = str.getToken ();

        // Parse out region window
        str.getToken ();
        x = str.getToken ();
        y = str.getToken ();
        w = str.getToken ();
        h = str.getToken ();
        win = (x, y, w, h);

        // Determine highlighter minimum and maximum levels
        if (colorname == "orange") {
          minlevel = 17;
          maxlevel = 17;
        } else if (colorname == "yellow") {
          minlevel = 34;
          maxlevel = 51;
        } else if (colorname == "spring") {
          minlevel  = 68;
          maxlevel  = 68;
        } else if (colorname == "green") {
          minlevel = 85;
          maxlevel = 85;
        } else if (colorname == "teal") {
          minlevel = 102;
          maxlevel = 102;
        } else if (colorname == "cyan") {
          minlevel = 119;
          maxlevel = 136;
        } else if (colorname == "azure") {
          minlevel = 153;
          maxlevel = 153;
        } else if (colorname == "blue") {
          minlevel = 170;
          maxlevel = 170;
        } else if (colorname == "violet") {
          minlevel = 187;
          maxlevel = 187;
        } else if (colorname == "magenta") {
          minlevel  = 204;
          maxlevel  = 238;
        } else if (colorname == "pink") {
          minlevel  = 204;
          maxlevel  = 238;
        } else if (colorname == "red") {
          minlevel = 255;
          maxlevel = 255;
        }

        imgBase = input.crop (window: win
                    ).interleave (scanline: 1
                    ).cspace (outspace: "hsv", precise: 1
                    );

        // Threshold for text (text image)
        imgText = imgBase.levels (levels: (12, 4, 4)
                    ).median (filter: "2 2 1 1 1 1"
                    ).range (lower: (0, 0, 0), upper: (255, 255, 85),
                             pixel: 1, incolor: (0, 0, 0), outcolor: (0, 0, 255)
                    ).channel (parts: "3"
                    ).convert (to: (1)
                    ).dilate_ipc (
                    ).erode_ipc (
                    );

        // Threshold for highlighter color (highlighter image)
        imgHL = imgBase.range (lower: (0, 32, 64), upper: (255, 255, 255),
                               pixel: 1, ingray: 1
                  ).levels (levels: (12, 4, 4)
                  ).median (filter: "2 2 1 1 1 1"
                  ).range (lower: (minlevel, 85, 85),
                           upper: (maxlevel, 255, 255)
                  ).channel (parts: "1"
                  ).convert (to: (1)
                  ).subsample (reduce: 4
                  ).convert (to: (1)
                  ).invert (
                  ).dilate_ipc (sestring: se5x5
                  ).erode_ipc (sestring: se5x5
                  ).erode_ipc (
                  ).dilate_ipc (
                  ).scale (awindow: (win[2], win[3])
                  ).convert (to: (1)
                  );

        // Add text image to mask channel of highlighter image
        imgHL = imgHL.addLayer (image: imgText, lindex: 0,
                                ltype: XIP_ColorMask);

        // Seed fill to get the text covered by highlighter
        imgHLText = imgHL.seedfill (iterations: 3);

        // Continue if percent average is 100% (no highlighted text in region)
        avg = getPercentAvg (inImg: imgHLText.convert (to: (8)));
        if (avg == 1.00)
          continue;

        // Label extraction image
        imgHLText.info = "Highlighter: " + colorname;

        // Add extraction image to output list
        output.insert (obj: imgHLText);
      }
    }

  // Merge extraction images if merge option is specified
  nOutputs = output.length ();
  if (merge && (nOutputs > 1)) {
    XIPIMAGE img, next;
    STRING   label;
    img = output.extract (entry: 0);

    for (i=1; i<nOutputs; i++) {
      next = output.extract (entry: 0);
      img = img.math (operator: "and", input: next);

      label = next.info;
      label.getToken ();
      img.info = img.info + ", " + label.getToken ();
    }
    output.insert (obj: img);
  }
}



/* @getPercentAvg
  // DESCRIPTION
  Given a text extraction image, return the average percent intensity. It uses
  XIPIMAGE method "avginfo" to capture image parameters of min, max, and
  average
*/
private
PROCEDURE getPercentAvg (XIPIMAGE inImg)
  RETURNS (DOUBLE percent)
{
  INTEGER  j;

  // Get average percent intensity of text extraction image
  STRING info = inImg.avginfo ().toString ();

  // Jump to "Entry 0" in "info" string then jump 13 tokens to get percent
  info.match (str: "Entry 0:");
  for (j=0; j<13; j++)
    info.getToken ();

  percent = info.getToken ();
}