nfv/lib/estimate_transform.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 avgMedian3, matchText;

matchText = function(word, words) {
  var item, matches;
  matches = (function() {
    var k, len, results;
    results = [];
    for (k = 0, len = words.length; k < len; k++) {
      item = words[k];
      if (item.text === word.text) {
        results.push(item);
      }
    }
    return results;
  })();
  if (matches.length === 1) {
    return matches[0];
  }
};

avgMedian3 = function(values) {
  var averagingValues, count, k, len, median, sum, value;
  if (values.length > 3) {
    values.sort();
    median = Math.floor(values.length / 2);
    averagingValues = values.slice(median - 1, +(median + 1) + 1 || 9e9);
  } else {
    averagingValues = values;
  }
  sum = 0;
  count = averagingValues.length;
  for (k = 0, len = averagingValues.length; k < len; k++) {
    value = averagingValues[k];
    sum += value;
  }
  if (count > 0) {
    return sum / count;
  }
};

module.exports.estimateTransform = function(itemsA, itemsB, fallbackScale, requiredMatchCount, findMatch) {
  var angle, byDistX, byDistY, distX, distY, expectedAngle, i, index, item, itemA1, itemA2, itemB1, itemB2, j, k, l, len, len1, m, matches, offsetX, offsetY, ref, ref1, scaleX, scaleY, transforms;
  if (requiredMatchCount == null) {
    requiredMatchCount = 7;
  }
  if (findMatch == null) {
    findMatch = matchText;
  }
  matches = new Array(itemsA.length);
  for (index = k = 0, len = itemsA.length; k < len; index = ++k) {
    item = itemsA[index];
    matches[index] = findMatch(item, itemsB);
  }
  transforms = [];
  for (i = l = 0, len1 = itemsA.length; l < len1; i = ++l) {
    itemA1 = itemsA[i];
    if (itemB1 = matches[i]) {
      for (j = m = ref = i + 1, ref1 = itemsA.length; m < ref1; j = m += 1) {
        itemA2 = itemsA[j];
        if (itemB2 = matches[j]) {
          distX = Math.abs(itemB1.box.x - itemB2.box.x);
          distY = Math.abs(itemB1.box.y - itemB2.box.y);
          angle = Math.atan2(itemB1.box.x - itemB2.box.x, itemB1.box.y - itemB2.box.y);
          expectedAngle = Math.atan2(itemA1.box.x - itemA2.box.x, itemA1.box.y - itemA2.box.y);
          if (distX + distY > 1000 && Math.abs(angle - expectedAngle) < 0.02) {
            scaleX = distX / Math.abs(itemA1.box.x - itemA2.box.x);
            scaleY = distY / Math.abs(itemA1.box.y - itemA2.box.y);
            transforms.push({
              distance: [distX, distY],
              scale: [scaleX, scaleY],
              offset: [itemB1.box.x - itemA1.box.x * scaleX, itemB1.box.y - itemA1.box.y * scaleY]
            });
          }
        }
      }
    }
  }
  if (transforms.length > requiredMatchCount) {
    byDistX = transforms.filter(function(i) {
      return i.distance[0] > 20;
    }).sort(function(a, b) {
      return a.distance[0] - b.distance[0];
    });
    scaleX = avgMedian3((function() {
      var len2, n, ref2, results;
      ref2 = byDistX.slice(-7);
      results = [];
      for (n = 0, len2 = ref2.length; n < len2; n++) {
        i = ref2[n];
        results.push(i.scale[0]);
      }
      return results;
    })());
    offsetX = avgMedian3((function() {
      var len2, n, ref2, results;
      ref2 = byDistX.slice(-7);
      results = [];
      for (n = 0, len2 = ref2.length; n < len2; n++) {
        i = ref2[n];
        results.push(i.offset[0]);
      }
      return results;
    })());
    byDistY = transforms.filter(function(i) {
      return i.distance[1] > 20;
    }).sort(function(a, b) {
      return a.distance[1] - b.distance[1];
    });
    scaleY = avgMedian3((function() {
      var len2, n, ref2, results;
      ref2 = byDistY.slice(-7);
      results = [];
      for (n = 0, len2 = ref2.length; n < len2; n++) {
        i = ref2[n];
        results.push(i.scale[1]);
      }
      return results;
    })());
    offsetY = avgMedian3((function() {
      var len2, n, ref2, results;
      ref2 = byDistY.slice(-7);
      results = [];
      for (n = 0, len2 = ref2.length; n < len2; n++) {
        i = ref2[n];
        results.push(i.offset[1]);
      }
      return results;
    })());
  } else {
    scaleX = fallbackScale;
    scaleY = fallbackScale;
    offsetX = 0;
    offsetY = 0;
  }
  return function(box) {
    return {
      x: box.x * scaleX + offsetX,
      y: box.y * scaleY + offsetY,
      width: box.width * scaleX,
      height: box.height * scaleY
    };
  };
};