image_duplicate.module

<?php

include_once 'inc/phash/phash.inc';
include_once 'inc/surf/surf.inc';
include_once 'inc/surf/fast_hessian.inc';
include_once 'inc/surf/integral_image.inc';
include_once 'inc/surf/response_layer.inc';
include_once 'inc/surf/gaussian_constants.inc';
include_once 'inc/surf/surf_interest_point.inc';
include_once 'inc/kmeans/kmeans.inc';

function image_duplicate_block_info() {
  $blocks = array();
  $blocks['image_duplicate_search'] = array(
    'info' => t('Image duplicate Search'),
  );
  return $blocks;
}

function image_duplicate_file_insert($file) {
  $image = image_load($file->uri);
  if ($image) {
    $type = substr($image->info['mime_type'], strpos($image->info['mime_type'], "/") + 1);
    $url = file_create_url($file->uri);
    $image_gd = call_user_func('imagecreatefrom' . $type, $url);

    $board = new Surf($image);
    $points = $board->getFreeOrientedInterestPoints();

    // Print points
    foreach ($points as $point) {
      imageellipse($image_gd, $point['x'], $point['y'], 5, 5, 0);
    }
    call_user_func_array('image' . $type, array($image_gd, 'points.' . $image->info['extension']));

    $kmeans = new KMeans();
    $kmeans->setData($points)
      ->setXKey('x')
      ->setYKey('y')
      ->setClusterCount(10)
      ->solve();

    $clusters = $kmeans->getClusters();
    $phash = array();

    $phash[] = array(
      'fid' => $file->fid,
      'full' => 1,
      'phash' => phash_dct_imagehash($image->resource, $image->info['extension'], $image->info['width'], $image->info['height'])
    );

    foreach ($clusters as $cluster) {
      $data = $cluster->getData();
      if (!empty($data)) {
        $bounds = $cluster->getBounds();
        $width = $bounds['maxX'] - $bounds['minX'];
        $height = $bounds['maxY'] - $bounds['minY'];

        $image_part = phash_create_tmp($image->resource, $image->info['extension'], $width, $height);
        imagecopy($image_part, $image->resource, 0, 0, $bounds['minX'], $bounds['minY'], $width, $height);

        $phash[] = array(
          'fid' => $file->fid,
          'full' => 0,
          'phash' => phash_dct_imagehash($image_part, $image->info['extension'], $image->info['width'], $image->info['height'])
        );

        // Add cluster
        imagerectangle($image_gd, $bounds['minX'], $bounds['minY'], $bounds['maxX'], $bounds['maxY'], 0);
      }
    }

    // Print clusters
    call_user_func_array('image' . $type, array($image_gd, 'clusters.' . $image->info['extension']));

    $query = db_insert('image_duplicate')->fields(array('fid', 'full', 'phash'));
    foreach ($phash as $record) {
      $query->values($record);
    }
    $query->execute();

  }
}

function image_duplicate_block_view($delta = '') {
  $block = array();
  switch ($delta) {
    case 'image_duplicate_search':
      $block['subject'] = '';
      $block['content'] = drupal_get_form('image_duplicate_serach_form');
      break;
  }
  return $block;
}

function image_duplicate_serach_form() {
  $form = array();
  $form['image'] = array(
    '#title' => t('Image'),
    '#type' => 'managed_file',
    '#description' => 'Upload an image and click "Search"',
    '#upload_location' => 'public://',
  );

  $form['submit'] = array(
    '#type' => 'submit',
    '#value' => t('Search'),
  );
  return $form;
}

function image_duplicate_serach_form_submit($form, &$form_state) {
  if (isset($form_state['values']['image'])) {
    $patterns = db_select('image_duplicate', 'i')
      ->fields('i')
      ->condition('fid', $form_state['values']['image'], '=')
      ->execute()
      ->fetchAll();

    $haystack = db_select('image_duplicate', 'i')
      ->fields('i')
      ->condition('fid', $form_state['values']['image'], '!=')
      ->execute()
      ->fetchAll();

    foreach ($patterns as $pattern) {
      $pattern->phash = hex2bin($pattern->phash);
    }

    $candidate_list = array();

    foreach ($haystack as $candidate) {
      $candidate->phash = hex2bin($candidate->phash);
      foreach ($patterns as $pattern) {
        $diff = phash_hamming_distance();
        if ($diff <= 10) {
          $candidate_list[] = $candidate;
        }
      }
    }

    // Make file permanent
    $file = file_load($form_state['values']['image']);
    $file->status = FILE_STATUS_PERMANENT;
    file_save($file);
  }
}