有人知道在PHP中进行面部检测的好方法吗？ I came across some code here声称这样做，但我似乎无法让它正常工作。我想做这项工作（即使它会很慢），你能给我的任何帮助都会非常感激。
以下是链接中的代码：
<?php
// as published by the Free Software Foundation; either version 2
// of the License, or (at your option) any later version.
//
// This program is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License
// along with this program; if not, write to the Free Software
// Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
//
// @Author Karthik Tharavaad
// karthik_tharavaad@yahoo.com
// @Contributor Maurice Svay
// maurice@svay.Com
class Face_Detector {
protected $detection_data;
protected $canvas;
protected $face;
private $reduced_canvas;
public function __construct($detection_file = 'detection.dat') {
if (is_file($detection_file)) {
$this->detection_data = unserialize(file_get_contents($detection_file));
} else {
throw new Exception("Couldn't load detection data");
}
//$this->detection_data = json_decode(file_get_contents('data.js'));
}
public function face_detect($file) {
if (!is_file($file)) {
throw new Exception("Can not load $file");
}
$this->canvas = imagecreatefromjpeg($file);
$im_width = imagesx($this->canvas);
$im_height = imagesy($this->canvas);
//Resample before detection?
$ratio = 0;
$diff_width = 320 - $im_width;
$diff_height = 240 - $im_height;
if ($diff_width > $diff_height) {
$ratio = $im_width / 320;
} else {
$ratio = $im_height / 240;
}
if ($ratio != 0) {
$this->reduced_canvas = imagecreatetruecolor($im_width / $ratio, $im_height / $ratio);
imagecopyresampled($this->reduced_canvas, $this->canvas, 0, 0, 0, 0, $im_width / $ratio, $im_height / $ratio, $im_width, $im_height);
$stats = $this->get_img_stats($this->reduced_canvas);
$this->face = $this->do_detect_greedy_big_to_small($stats['ii'], $stats['ii2'], $stats['width'], $stats['height']);
$this->face['x'] *= $ratio;
$this->face['y'] *= $ratio;
$this->face['w'] *= $ratio;
} else {
$stats = $this->get_img_stats($this->canvas);
$this->face = $this->do_detect_greedy_big_to_small($stats['ii'], $stats['ii2'], $stats['width'], $stats['height']);
}
return ($this->face['w'] > 0);
}
public function toJpeg() {
$color = imagecolorallocate($this->canvas, 255, 0, 0); //red
imagerectangle($this->canvas, $this->face['x'], $this->face['y'], $this->face['x']+$this->face['w'], $this->face['y']+ $this->face['w'], $color);
header('Content-type: image/jpeg');
imagejpeg($this->canvas);
}
public function toJson() {
return "{'x':" . $this->face['x'] . ", 'y':" . $this->face['y'] . ", 'w':" . $this->face['w'] . "}";
}
public function getFace() {
return $this->face;
}
protected function get_img_stats($canvas){
$image_width = imagesx($canvas);
$image_height = imagesy($canvas);
$iis = $this->compute_ii($canvas, $image_width, $image_height);
return array(
'width' => $image_width,
'height' => $image_height,
'ii' => $iis['ii'],
'ii2' => $iis['ii2']
);
}
protected function compute_ii($canvas, $image_width, $image_height ){
$ii_w = $image_width+1;
$ii_h = $image_height+1;
$ii = array();
$ii2 = array();
for($i=0; $i<$ii_w; $i++ ){
$ii[$i] = 0;
$ii2[$i] = 0;
}
for($i=1; $i<$ii_w; $i++ ){
$ii[$i*$ii_w] = 0;
$ii2[$i*$ii_w] = 0;
$rowsum = 0;
$rowsum2 = 0;
for($j=1; $j<$ii_h; $j++ ){
$rgb = ImageColorAt($canvas, $j, $i);
$red = ($rgb >> 16) & 0xFF;
$green = ($rgb >> 8) & 0xFF;
$blue = $rgb & 0xFF;
$grey = ( 0.2989*$red + 0.587*$green + 0.114*$blue )>>0; // this is what matlab uses
$rowsum += $grey;
$rowsum2 += $grey*$grey;
$ii_above = ($i-1)*$ii_w + $j;
$ii_this = $i*$ii_w + $j;
$ii[$ii_this] = $ii[$ii_above] + $rowsum;
$ii2[$ii_this] = $ii2[$ii_above] + $rowsum2;
}
}
return array('ii'=>$ii, 'ii2' => $ii2);
}
protected function do_detect_greedy_big_to_small( $ii, $ii2, $width, $height ){
$s_w = $width/20.0;
$s_h = $height/20.0;
$start_scale = $s_h < $s_w ? $s_h : $s_w;
$scale_update = 1 / 1.2;
for($scale = $start_scale; $scale > 1; $scale *= $scale_update ){
$w = (20*$scale) >> 0;
$endx = $width - $w - 1;
$endy = $height - $w - 1;
$step = max( $scale, 2 ) >> 0;
$inv_area = 1 / ($w*$w);
for($y = 0; $y < $endy ; $y += $step ){
for($x = 0; $x < $endx ; $x += $step ){
$passed = $this->detect_on_sub_image( $x, $y, $scale, $ii, $ii2, $w, $width+1, $inv_area);
if( $passed ) {
return array('x'=>$x, 'y'=>$y, 'w'=>$w);
}
} // end x
} // end y
} // end scale
return null;
}
protected function detect_on_sub_image( $x, $y, $scale, $ii, $ii2, $w, $iiw, $inv_area){
$mean = ( $ii[($y+$w)*$iiw + $x + $w] + $ii[$y*$iiw+$x] - $ii[($y+$w)*$iiw+$x] - $ii[$y*$iiw+$x+$w] )*$inv_area;
$vnorm = ( $ii2[($y+$w)*$iiw + $x + $w] + $ii2[$y*$iiw+$x] - $ii2[($y+$w)*$iiw+$x] - $ii2[$y*$iiw+$x+$w] )*$inv_area - ($mean*$mean);
$vnorm = $vnorm > 1 ? sqrt($vnorm) : 1;
$passed = true;
for($i_stage = 0; $i_stage < count($this->detection_data); $i_stage++ ){
$stage = $this->detection_data[$i_stage];
$trees = $stage[0];
$stage_thresh = $stage[1];
$stage_sum = 0;
for($i_tree = 0; $i_tree < count($trees); $i_tree++ ){
$tree = $trees[$i_tree];
$current_node = $tree[0];
$tree_sum = 0;
while( $current_node != null ){
$vals = $current_node[0];
$node_thresh = $vals[0];
$leftval = $vals[1];
$rightval = $vals[2];
$leftidx = $vals[3];
$rightidx = $vals[4];
$rects = $current_node[1];
$rect_sum = 0;
for( $i_rect = 0; $i_rect < count($rects); $i_rect++ ){
$s = $scale;
$rect = $rects[$i_rect];
$rx = ($rect[0]*$s+$x)>>0;
$ry = ($rect[1]*$s+$y)>>0;
$rw = ($rect[2]*$s)>>0;
$rh = ($rect[3]*$s)>>0;
$wt = $rect[4];
$r_sum = ( $ii[($ry+$rh)*$iiw + $rx + $rw] + $ii[$ry*$iiw+$rx] - $ii[($ry+$rh)*$iiw+$rx] - $ii[$ry*$iiw+$rx+$rw] )*$wt;
$rect_sum += $r_sum;
}
$rect_sum *= $inv_area;
$current_node = null;
if( $rect_sum >= $node_thresh*$vnorm ){
if( $rightidx == -1 )
$tree_sum = $rightval;
else
$current_node = $tree[$rightidx];
} else {
if( $leftidx == -1 )
$tree_sum = $leftval;
else
$current_node = $tree[$leftidx];
}
}
$stage_sum += $tree_sum;
}
if( $stage_sum < $stage_thresh ){
return false;
}
}
return true;
}
}
用法：
$detector = new Face_Detector('detection.dat');
$detector->face_detect('maurice_svay_150.jpg');
$detector->toJpeg();
我遇到的问题似乎也出现在该页面的评论中。 “imagecolorat（）[function.imagecolorat]：320,1超出范围。”所以，我在文件的顶部添加了一个error_reporting（0）（实际上不是解决方案），它似乎有时会起作用，而有时它却什么也不做。
有什么想法吗？
答案 0 :(得分：4)
使用OpenCV执行此操作可能会更容易/更安全，这是用较低级别的代码编写的。 PHP被解释，因此在完成工作时可能会变得很慢。
希望这有帮助！
答案 1 :(得分：3)
您需要关闭错误报告
<?php
ini_set( 'display_errors', 1 );
error_reporting( E_ALL ^ E_NOTICE );
require_once('face_detector.php');
$detector = new Face_Detector('detection.dat');
$detector->face_detect('img/8.jpg');
$detector->toJpeg();
?>
答案 2 :(得分：1)
尝试从以下行中删除+1：
$ii_w = $image_width+1;
$ii_h = $image_height+1;
此代码试图检查320像素图像中位置1到320而不是0到319的颜色。
答案 3 :(得分：0)
快速修复：在compute_ii函数
替换：
$rgb = ImageColorAt($canvas, $j, $i);
使用：
$rgb = ImageColorAt($canvas, $j-1, $i-1);
答案 4 :(得分：0)
这是一个古老的主题，但这个修复程序仍然比我迄今为止看到的更好，所以它可能有助于某人
// Turn off error reporting...
$ctl = error_reporting();
error_reporting(0);
$detector = new Face_Detector('detection.dat');
$detector->face_detect('img/8.jpg');
$detector->toJpeg();
// Turn on reporting...if you wish
error_reporting($ctl);
答案 5 :(得分：0)