当前位置：首页 > PHP

php实现相似图片

2026-02-16 22:17:52PHP

相似图片检测方法

使用感知哈希（pHash）

感知哈希通过计算图片的哈希值来比较相似度。常用方法包括平均哈希（aHash）、差异哈希（dHash）和感知哈希（pHash）。以下是使用PHP-GD库实现pHash的示例：

php实现相似图片

function calculatePHash($imagePath) {
    $size = 32;
    $smallerSize = 8;

    $image = imagecreatefromstring(file_get_contents($imagePath));
    list($width, $height) = getimagesize($imagePath);

    $resized = imagecreatetruecolor($size, $size);
    imagecopyresampled($resized, $image, 0, 0, 0, 0, $size, $size, $width, $height);
    imagedestroy($image);

    $grayscale = array();
    for ($y = 0; $y < $size; $y++) {
        for ($x = 0; $x < $size; $x++) {
            $rgb = imagecolorat($resized, $x, $y);
            $r = ($rgb >> 16) & 0xFF;
            $g = ($rgb >> 8) & 0xFF;
            $b = $rgb & 0xFF;
            $grayscale[$y][$x] = 0.299 * $r + 0.587 * $g + 0.114 * $b;
        }
    }
    imagedestroy($resized);

    $dct = array();
    for ($u = 0; $u < $smallerSize; $u++) {
        for ($v = 0; $v < $smallerSize; $v++) {
            $sum = 0;
            for ($x = 0; $x < $size; $x++) {
                for ($y = 0; $y < $size; $y++) {
                    $sum += $grayscale[$y][$x] * 
                           cos((2 * $x + 1) * $u * M_PI / (2 * $size)) * 
                           cos((2 * $y + 1) * $v * M_PI / (2 * $size));
                }
            }
            $dct[$u][$v] = $sum;
        }
    }

    $avg = array_sum(array_map('array_sum', $dct)) / ($smallerSize * $smallerSize);
    $hash = '';
    for ($u = 0; $u < $smallerSize; $u++) {
        for ($v = 0; $v < $smallerSize; $v++) {
            $hash .= ($dct[$u][$v] > $avg) ? '1' : '0';
        }
    }
    return $hash;
}

function hammingDistance($hash1, $hash2) {
    $distance = 0;
    $len = min(strlen($hash1), strlen($hash2));
    for ($i = 0; $i < $len; $i++) {
        if ($hash1[$i] != $hash2[$i]) {
            $distance++;
        }
    }
    return $distance;
}

使用颜色直方图比较

颜色直方图通过统计图片中颜色分布来比较相似度：

php实现相似图片

function compareHistograms($imagePath1, $imagePath2) {
    $bins = 8;
    $hist1 = calculateHistogram($imagePath1, $bins);
    $hist2 = calculateHistogram($imagePath2, $bins);

    $sum = 0;
    for ($i = 0; $i < $bins * $bins * $bins; $i++) {
        $sum += min($hist1[$i], $hist2[$i]);
    }
    return $sum;
}

function calculateHistogram($imagePath, $bins) {
    $image = imagecreatefromstring(file_get_contents($imagePath));
    $width = imagesx($image);
    $height = imagesy($image);

    $histogram = array_fill(0, $bins * $bins * $bins, 0);
    for ($x = 0; $x < $width; $x++) {
        for ($y = 0; $y < $height; $y++) {
            $rgb = imagecolorat($image, $x, $y);
            $r = ($rgb >> 16) & 0xFF;
            $g = ($rgb >> 8) & 0xFF;
            $b = $rgb & 0xFF;

            $rBin = floor($r / (256 / $bins));
            $gBin = floor($g / (256 / $bins));
            $bBin = floor($b / (256 / $bins));

            $index = $rBin * $bins * $bins + $gBin * $bins + $bBin;
            $histogram[$index]++;
        }
    }
    imagedestroy($image);

    $total = $width * $height;
    return array_map(function($val) use ($total) {
        return $val / $total;
    }, $histogram);
}

使用OpenCV扩展

PHP可通过OpenCV扩展实现更高级的相似度检测：

$image1 = cv\imread('image1.jpg', cv\IMREAD_COLOR);
$image2 = cv\imread('image2.jpg', cv\IMREAD_COLOR);

$orb = cv\ORB::create();
$keypoints1 = new cv\KeyPoint();
$keypoints2 = new cv\KeyPoint();
$descriptors1 = new cv\Mat();
$descriptors2 = new cv\Mat();

$orb->detectAndCompute($image1, null, $keypoints1, $descriptors1);
$orb->detectAndCompute($image2, null, $keypoints2, $descriptors2);

$matcher = cv\BFMatcher::create(cv\NORM_HAMMING, true);
$matches = new cv\DMatch();
$matcher->match($descriptors1, $descriptors2, $matches);

$similarity = 0;
foreach ($matches as $match) {
    $similarity += $match->distance;
}
$similarity = 1 - ($similarity / count($matches) / 256);

使用机器学习服务

对于大规模应用，可考虑集成云端API：

function compareWithCloudAPI($image1, $image2) {
    $client = new \GuzzleHttp\Client();
    $response = $client->post('https://api.cloudvision.com/v1/similarity', [
        'json' => [
            'image1' => base64_encode(file_get_contents($image1)),
            'image2' => base64_encode(file_get_contents($image2))
        ]
    ]);
    return json_decode($response->getBody(), true)['similarity'];
}