getWideChunks method

List<Artifact> getWideChunks()

Identifies artifacts that are significantly wider than average.

This method finds artifacts whose width exceeds twice the average width of all artifacts in the band, which often indicates merged characters.

Special cases:

• If there are only 1-2 artifacts of similar width, they are not considered wide
• Width comparison uses a dynamic threshold based on the number of artifacts

Returns a list of artifacts that are candidates for splitting.

Implementation

List<Artifact> getWideChunks() {
  final List<Artifact> listToInspect = [];

  // If we have 0 or 1 artifacts, there's nothing to inspect
  if (artifacts.isEmpty || artifacts.length == 1) {
    return listToInspect;
  }

  // Special case: If we have exactly 2 artifacts with similar widths,
  // don't consider either of them as wide chunks
  if (artifacts.length == 2) {
    final double widthRatio = artifacts[0].cols / artifacts[1].cols;
    // If the width ratio is between 0.7 and 1.3, they're similar enough
    if (widthRatio >= 0.7 && widthRatio <= 1.3) {
      return listToInspect; // Return empty list
    }
  }

  // Calculate threshold based on number of artifacts
  // With fewer artifacts, we need a higher threshold to avoid false positives
  double thresholdMultiplier = 2.5;
  if (artifacts.length <= 3) {
    thresholdMultiplier = 1.5; // More conservative for small sets
  }

  final double thresholdWidth = averageWidth * thresholdMultiplier;

  for (final Artifact artifact in artifacts) {
    artifact.needsInspection = artifact.cols > thresholdWidth;
    if (artifact.needsInspection) {
      listToInspect.add(artifact);
    }
  }
  return listToInspect;
}