nfv/lib/find_checkboxes.js

[Updated to use ndv package instead dv] FormVision is a node.js library for extracting data from scanned forms

// Generated by CoffeeScript 1.12.7
var LOWER_CERTAIN_SCORE, LOWER_UNCERTAIN_SCORE, QUIETZONE_WIDTH, UPPER_CERTAIN_SCORE, UPPER_UNCERTAIN_SCORE, binarize, detectCandidates, dv, scoreCandidate, scoreToCheckState;

dv = require('ndv');

binarize = require('./filters/binarize');

QUIETZONE_WIDTH = 10;

LOWER_UNCERTAIN_SCORE = 0.02;

LOWER_CERTAIN_SCORE = 0.10;

UPPER_CERTAIN_SCORE = 1.5;

UPPER_UNCERTAIN_SCORE = 4;

detectCandidates = function(binarizedImage) {
  var candidate, candidates, i, innerCandidate, j, len, len1, ref, ref1, ref2, ref3, ref4, ref5;
  candidates = [];
  ref = binarizedImage.dilate(3, 3).connectedComponents(8);
  for (i = 0, len = ref.length; i < len; i++) {
    candidate = ref[i];
    if ((0.5 < (ref1 = candidate.width / candidate.height) && ref1 < 5) && (10 < (ref2 = candidate.width) && ref2 < (binarizedImage.width / 3))) {
      candidates.push(candidate);
    } else if (candidate.width > (binarizedImage.width / 3)) {
      ref3 = binarizedImage.crop(candidate).erode(3, 3).dilate(5, 5).connectedComponents(8);
      for (j = 0, len1 = ref3.length; j < len1; j++) {
        innerCandidate = ref3[j];
        if ((0.5 < (ref4 = innerCandidate.width / innerCandidate.height) && ref4 < 2) && (20 < (ref5 = innerCandidate.width) && ref5 < (binarizedImage.width / 3))) {
          innerCandidate.x += candidate.x;
          innerCandidate.y += candidate.y;
          candidates.push(innerCandidate);
        }
      }
    }
  }
  return candidates;
};

scoreCandidate = function(binarizedImage, candidate) {
  var candidateImage, distanceImage, imageWithGapsClosed, imageWithWeightedPixels, index, overscannedCandidate, ref, score, value;
  overscannedCandidate = {
    x: candidate.x - QUIETZONE_WIDTH,
    y: candidate.y - QUIETZONE_WIDTH,
    width: candidate.width + QUIETZONE_WIDTH * 2,
    height: candidate.height + QUIETZONE_WIDTH * 2
  };
  candidateImage = binarizedImage.crop(overscannedCandidate);
  imageWithGapsClosed = candidateImage.dilate(31, 31).erode(31, 31);
  distanceImage = imageWithGapsClosed.distanceFunction(8);
  imageWithWeightedPixels = distanceImage.and(candidateImage.invert().toGray());
  score = 0;
  ref = imageWithWeightedPixels.histogram();
  for (value in ref) {
    index = ref[value];
    score += index * value;
  }
  return score;
};

scoreToCheckState = (function(_this) {
  return function(score) {
    var checked, confidence, lerp;
    lerp = function(x, zeroAt, oneAt) {
      return (x - zeroAt) / (oneAt - zeroAt);
    };
    checked = (LOWER_UNCERTAIN_SCORE < score && score < UPPER_UNCERTAIN_SCORE);
    confidence = (function() {
      switch (false) {
        case !(score < LOWER_UNCERTAIN_SCORE):
          return lerp(score, LOWER_UNCERTAIN_SCORE, 0);
        case !((LOWER_UNCERTAIN_SCORE <= score && score < LOWER_CERTAIN_SCORE)):
          return lerp(score, LOWER_UNCERTAIN_SCORE, LOWER_CERTAIN_SCORE);
        case !((LOWER_CERTAIN_SCORE <= score && score < UPPER_CERTAIN_SCORE)):
          return 1.0;
        case !((UPPER_CERTAIN_SCORE <= score && score < UPPER_UNCERTAIN_SCORE)):
          return lerp(score, UPPER_UNCERTAIN_SCORE, UPPER_CERTAIN_SCORE);
        default:
          return 0.0;
      }
    })();
    confidence = Math.round(confidence * 92 + 5);
    return [checked, confidence];
  };
})(this);

module.exports.findCheckboxes = function(image) {
  var binarizedImage, box, candidates, checked, clearedImage, confidence, i, len, marks, ref, score;
  marks = [];
  clearedImage = new dv.Image(image);
  binarizedImage = binarize(image);
  candidates = detectCandidates(binarizedImage);
  for (i = 0, len = candidates.length; i < len; i++) {
    box = candidates[i];
    score = scoreCandidate(binarizedImage, box);
    ref = scoreToCheckState(score), checked = ref[0], confidence = ref[1];
    if (confidence > 0) {
      marks.push({
        box: box,
        checked: checked,
        confidence: confidence
      });
      clearedImage.clearBox(box);
    }
  }
  return [marks, clearedImage];
};