| SmokeyTheBear |
11-22-2007 07:20 PM |
Code:
<?php
class OCR
{
private $im;
function Color($x, $y)
{
if (!is_resource($this->im))
{
return false;
}
extract(imagecolorsforindex($this->im, imagecolorat($this->im, $x, $y)));
$red = base_convert($red, 10, 16);
if (strlen($red) < 2)
{
$red = '0' . $red;
}
$green = base_convert($green, 10, 16);
if (strlen($green) < 2)
{
$green = '0' . $green;
}
$blue = base_convert($blue, 10, 16);
if (strlen($blue) < 2)
{
$blue = '0' . $blue;
}
return strtoupper($red . $green . $blue);
}
function Info($image, $flag = null)
{
list($width, $height, $type) = getimagesize($image);
if (!empty($flag))
{
$flag = strtolower(trim($flag));
if ($flag == 'width')
{
return $width;
}
else if ($flag == 'height')
{
return $height;
}
else if ($flag == 'type')
{
switch ($type)
{
case 1:
return 'gif';
break;
case 2:
return 'jpg';
break;
case 3:
return 'png';
break;
}
}
return false;
}
switch ($type)
{
case 1:
$type = 'gif';
break;
case 2:
$type = 'jpg';
break;
case 3:
$type = 'png';
break;
}
$result = array
(
'width' => $width,
'height' => $height,
'type' => $type
);
return $result;
}
function Read($tag, $image, $background_vector = 'FFFFFF')
{
$image_info = $this->Info($image);
$tag = strtolower($tag);
if (!is_dir('OCR/' . $tag . '/'))
{
return false;
}
$characters = array();
$handle = opendir('OCR/' . $tag . '/patterns/');
while (($file = readdir($handle)) !== false)
{
if (($file != '.') && ($file != '..'))
{
if (!is_dir('OCR/' . $tag . '/patterns/' . $file))
{
$file = str_replace('.ocr', '', $file);
$characters[$file] = $file;
}
}
}
closedir($handle);
$ignore_hash = '';
if (!empty($background_vector))
{
if (strlen($background_vector) == 6)
{
$ignore_hash = md5(str_repeat($background_vector, $image_info['height']));
}
else
{
$ignore_hash = md5($background_vector);
}
}
if ($image_info['type'] == 'gif')
{
$this->im = imagecreatefromgif($image);
}
else if ($image_info['type'] == 'jpg')
{
$this->im = imagecreatefromjpeg($image);
}
else if ($image_info['type'] == 'png')
{
$this->im = imagecreatefrompng($image);
}
$test = array();
$column = 0;
$candidates = array();
$result = '';
for ($i = 0; $i < $image_info['width']; $i++)
{
$test[$column] = '';
for ($j = 0; $j < $image_info['height']; $j++)
{
$test[$column] .= $this->Color($i, $j);
}
$test[$column] = md5($test[$column]);
if ($test[$column] == $ignore_hash)
{
$test = array();
$column = 0;
}
else
{
if (empty($candidates[$i - $column]))
{
foreach ($characters as $character)
{
$pattern = unserialize(file_get_contents('OCR/' . $tag . '/patterns/' . $character . '.ocr'));
if ($test[$column] == $pattern[$column])
{
$candidates[$i - $column][] = $character;
}
}
}
else
{
foreach ($candidates[$i - $column] as $character)
{
$pattern = unserialize(file_get_contents('OCR/' . $tag . '/patterns/' . $character . '.ocr'));
if ($test[$column] != $pattern[$column])
{
$key = array_search($character, $candidates[$i - $column]);
unset($candidates[$i - $column][$key]);
}
}
}
if (count($candidates[$i - $column]) == 1)
{
$candidates[$i - $column] = array_values($candidates[$i - $column]);
$result .= $candidates[$i - $column][0];
$character_width = $this->Info('OCR/' . $tag . '/source/' . $candidates[$i - $column][0] . '.' . $image_info['type'], 'width');
$i = $i - $column + $character_width - 1;
$column = 0;
}
else
{
$column++;
}
}
}
return $result;
}
function Train($tag, $character, $image)
{
if (!file_exists($image))
{
return false;
}
else
{
$image_info = $this->Info($image);
}
$tag = strtolower($tag);
if (!is_dir('OCR/' . $tag . '/'))
{
mkdir('OCR/' . $tag . '/');
}
if (!is_dir('OCR/' . $tag . '/source/'))
{
mkdir('OCR/' . $tag . '/source/');
}
if (file_exists('OCR/' . $tag . '/source/' . $character . '.' . $image_info['type']))
{
@unlink('OCR/' . $tag . '/source/' . $character . '.' . $image_info['type']);
}
@copy($image, 'OCR/' . $tag . '/source/' . $character . '.' . $image_info['type']);
$image = 'OCR/' . $tag . '/source/' . $character . '.' . $image_info['type'];
if ($image_info['type'] == 'gif')
{
$this->im = imagecreatefromgif($image);
}
else if ($image_info['type'] == 'jpg')
{
$this->im = imagecreatefromjpeg($image);
}
else if ($image_info['type'] == 'png')
{
$this->im = imagecreatefrompng($image);
}
$result = array();
for ($i = 0; $i < $image_info['width']; $i++)
{
$result[$i] = '';
for ($j = 0; $j < $image_info['height']; $j++)
{
$result[$i] .= $this->Color($i, $j);
}
$result[$i] = md5($result[$i]);
}
if (!is_dir('OCR/' . $tag . '/patterns/'))
{
mkdir('OCR/' . $tag . '/patterns/');
}
if (file_exists('OCR/' . $tag . '/patterns/' . $character . '.ocr'))
{
@unlink('OCR/' . $tag . '/patterns/' . $character . '.ocr');
}
$handle = @fopen('OCR/' . $tag . '/patterns/' . $character . '.ocr', 'wb');
flock($handle, LOCK_EX);
fwrite($handle, serialize($result));
flock($handle, LOCK_UN);
fclose($handle);
}
}
?>
|
