用Python实现添加图片噪声

hrh123 · 发表于 2025-7-14 19:16

本帖最后由 hrh123 于 2025-7-14 19:28 编辑

最近接项目时需要实现图片噪声功能，用于突破AI重复图片检测。为此写了一个复合噪声生成函数，通过：

椒盐+高斯噪声 - 污染关键像素改变pHash值
HSV色彩扰动 - 破坏颜色统计特征
梯度对抗扰动 - 干扰特征匹配算法

该代码同时适用于：

真实场景图像退化模拟
机器学习数据增强样本生成

实现细节详见代码（需要opencv-python和numpy）

import cv2
import numpy as np

def add_noise(image: str, out_put:str,intensity=0.1):
    """
    噪声
    :param img: 输入图像路径
    :param intensity: 噪声强度 (0.0 ~ 0.5)
    :param out_put: 图像输出路径
    :return: 无
    """
    img = cv2.imread(image)

    intensity = np.clip(intensity, 0.0, 0.5)

    out = img.astype(np.float32)
    h, w = img.shape[:2]

    rnd = np.random.rand(h, w)
    salt = rnd < (intensity / 2)
    pepper = rnd > (1 - intensity / 2)
    out[salt]   = 255.0
    out[pepper] =   0.0

    gauss = np.empty_like(out)
    cv2.randn(gauss, 0.0, intensity * 50.0)
    out += gauss

    hsv = cv2.cvtColor(np.clip(out, 0, 255).astype(np.uint8),
                       cv2.COLOR_BGR2HSV).astype(np.float32)
    hue_shift = np.random.uniform(-10, 10)
    hsv[..., 0] = (hsv[..., 0] + hue_shift) % 180
    hsv[..., 1] = hsv[..., 1] * (1.0 + 0.5 * intensity)
    hsv[..., 2] = hsv[..., 2] * np.random.uniform(1.0 - intensity,
                                                  1.0 + intensity)
    out = cv2.cvtColor(np.clip(hsv, 0, 255).astype(np.uint8),
                       cv2.COLOR_HSV2BGR).astype(np.float32)

    gx = cv2.Sobel(out, cv2.CV_32F, 1, 0, ksize=3)
    gy = cv2.Sobel(out, cv2.CV_32F, 0, 1, ksize=3)
    adv = intensity * 0.1 * np.sign(gx + gy)
    out += adv

    result_img = np.clip(out, 0, 255).astype(np.uint8)

    cv2.imwrite(out_put, result_img)
    return None

有写的不好的地方还请诸位坛友多多指正

吐槽一句论坛markdown的代码高亮怎么这么丑啊{:301_997:}

wudalang123 · 发表于 2025-7-16 08:00

[HTML] 纯文本查看 复制代码

<!DOCTYPE html>
<html lang="zh-CN">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>图片噪声添加工具</title>
    <style>
        :root {
            --primary-color: #4a6fa5;
            --secondary-color: #6b8cbc;
            --accent-color: #ff6b6b;
            --light-bg: #f8f9fa;
            --dark-bg: #343a40;
            --border-radius: 8px;
            --box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1);
        }
        
        * {
            margin: 0;
            padding: 0;
            box-sizing: border-box;
            font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif;
        }
        
        body {
            background: linear-gradient(135deg, #1a2a6c, #b21f1f, #1a2a6c);
            color: #333;
            min-height: 100vh;
            padding: 20px;
            display: flex;
            justify-content: center;
            align-items: center;
        }
        
        .container {
            width: 100%;
            max-width: 1200px;
            margin: 0 auto;
        }
        
        header {
            text-align: center;
            margin-bottom: 30px;
            color: white;
            text-shadow: 0 2px 4px rgba(0, 0, 0, 0.3);
        }
        
        header h1 {
            font-size: 2.5rem;
            margin-bottom: 10px;
            background: linear-gradient(90deg, #ff8a00, #e52e71);
            -webkit-background-clip: text;
            -webkit-text-fill-color: transparent;
        }
        
        header p {
            font-size: 1.2rem;
            max-width: 800px;
            margin: 0 auto;
            color: rgba(255, 255, 255, 0.85);
        }
        
        .app-card {
            background-color: rgba(255, 255, 255, 0.92);
            border-radius: var(--border-radius);
            box-shadow: var(--box-shadow);
            padding: 30px;
            margin-bottom: 30px;
            backdrop-filter: blur(10px);
        }
        
        .upload-section {
            text-align: center;
            padding: 30px;
            border: 2px dashed #ced4da;
            border-radius: var(--border-radius);
            transition: all 0.3s;
            margin-bottom: 30px;
            background-color: rgba(255, 255, 255, 0.7);
        }
        
        .upload-section:hover {
            border-color: var(--primary-color);
            background-color: rgba(255, 255, 255, 0.9);
        }
        
        .upload-icon {
            font-size: 4rem;
            color: #6c757d;
            margin-bottom: 20px;
        }
        
        .controls {
            display: flex;
            flex-direction: column;
            gap: 25px;
            margin-bottom: 30px;
        }
        
        .control-group {
            background-color: white;
            padding: 20px;
            border-radius: var(--border-radius);
            box-shadow: var(--box-shadow);
        }
        
        .control-group h3 {
            margin-bottom: 15px;
            color: var(--primary-color);
            display: flex;
            align-items: center;
            gap: 10px;
        }
        
        .slider-container {
            display: flex;
            flex-direction: column;
            gap: 10px;
        }
        
        .slider-label {
            display: flex;
            justify-content: space-between;
            font-weight: 500;
        }
        
        input[type="range"] {
            width: 100%;
            height: 8px;
            -webkit-appearance: none;
            background: linear-gradient(to right, #4a6fa5, #6b8cbc);
            border-radius: 4px;
            outline: none;
        }
        
        input[type="range"]::-webkit-slider-thumb {
            -webkit-appearance: none;
            width: 22px;
            height: 22px;
            border-radius: 50%;
            background: white;
            border: 2px solid var(--primary-color);
            cursor: pointer;
            box-shadow: 0 2px 4px rgba(0,0,0,0.2);
        }
        
        .preview-section {
            display: flex;
            flex-wrap: wrap;
            gap: 30px;
            margin-bottom: 30px;
        }
        
        .image-container {
            flex: 1;
            min-width: 300px;
            text-align: center;
        }
        
        .image-container h3 {
            margin-bottom: 15px;
            color: var(--dark-bg);
            padding-bottom: 10px;
            border-bottom: 2px solid var(--primary-color);
        }
        
        canvas {
            max-width: 100%;
            height: auto;
            background-color: #f0f2f5;
            border-radius: var(--border-radius);
            box-shadow: var(--box-shadow);
            display: none;
        }
        
        .placeholder {
            width: 100%;
            height: 300px;
            background: linear-gradient(45deg, #f8f9fa, #e9ecef);
            border-radius: var(--border-radius);
            display: flex;
            justify-content: center;
            align-items: center;
            color: #6c757d;
            font-size: 1.1rem;
            box-shadow: var(--box-shadow);
        }
        
        .btn-group {
            display: flex;
            justify-content: center;
            gap: 15px;
            margin-top: 20px;
            flex-wrap: wrap;
        }
        
        button {
            padding: 12px 25px;
            border: none;
            border-radius: 50px;
            font-size: 1rem;
            font-weight: 600;
            cursor: pointer;
            transition: all 0.3s;
            box-shadow: 0 4px 6px rgba(0, 0, 0, 0.1);
            display: flex;
            align-items: center;
            gap: 8px;
        }
        
        .btn-primary {
            background: linear-gradient(135deg, var(--primary-color), var(--secondary-color));
            color: white;
        }
        
        .btn-primary:hover {
            transform: translateY(-2px);
            box-shadow: 0 6px 12px rgba(0, 0, 0, 0.15);
            background: linear-gradient(135deg, #3a5a90, #5a7cac);
        }
        
        .btn-secondary {
            background: linear-gradient(135deg, #6c757d, #868e96);
            color: white;
        }
        
        .btn-accent {
            background: linear-gradient(135deg, #ff8a00, #e52e71);
            color: white;
        }
        
        .noise-types {
            display: grid;
            grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
            gap: 15px;
            margin-top: 20px;
        }
        
        .noise-type {
            background: white;
            padding: 15px;
            border-radius: var(--border-radius);
            box-shadow: 0 2px 6px rgba(0,0,0,0.05);
            border: 1px solid #e9ecef;
            transition: all 0.3s;
        }
        
        .noise-type:hover {
            transform: translateY(-3px);
            box-shadow: 0 5px 15px rgba(0,0,0,0.1);
            border-color: var(--primary-color);
        }
        
        .noise-type h4 {
            display: flex;
            align-items: center;
            gap: 8px;
            margin-bottom: 10px;
            color: var(--primary-color);
        }
        
        .noise-type p {
            font-size: 0.9rem;
            color: #495057;
        }
        
        footer {
            text-align: center;
            color: rgba(255, 255, 255, 0.8);
            padding: 20px;
            font-size: 0.9rem;
        }
        
        [url=home.php?mod=space&uid=945662]@media[/url] (max-width: 768px) {
            .preview-section {
                flex-direction: column;
            }
            
            .app-card {
                padding: 20px;
            }
            
            header h1 {
                font-size: 2rem;
            }
        }
    </style>
</head>
<body>
    <div class="container">
        <header>
            <h1>图片噪声添加工具</h1>
            <p>突破AI重复图片检测 | 模拟真实场景图像退化 | 机器学习数据增强</p>
        </header>
        
        <div class="app-card">
            <div class="upload-section" id="dropZone">
                <div class="upload-icon">&#128193;</div>
                <h2>上传图片添加噪声</h2>
                <p>支持 JPG, PNG, WEBP 格式的图片文件</p>
                <p class="small">拖放图片到此处或点击下方按钮选择文件</p>
                <input type="file" id="fileInput" accept="image/*" style="display: none;">
                <button class="btn-primary" onclick="document.getElementById('fileInput').click()">
                    <span>选择图片</span>
                </button>
            </div>
            
            <div class="controls">
                <div class="control-group">
                    <h3>&#9881;&#65039; 噪声设置</h3>
                    <div class="slider-container">
                        <div class="slider-label">
                            <span>噪声强度</span>
                            <span id="intensityValue">0.30</span>
                        </div>
                        <input type="range" id="intensitySlider" min="0" max="100" value="30">
                        <div class="slider-label">
                            <span>低</span>
                            <span>高</span>
                        </div>
                    </div>
                </div>
                
                <div class="control-group">
                    <h3>&#128266; 噪声类型</h3>
                    <div class="noise-types">
                        <div class="noise-type">
                            <h4>&#129474; 椒盐噪声</h4>
                            <p>随机添加黑白像素点，模拟传感器缺陷</p>
                        </div>
                        <div class="noise-type">
                            <h4>&#128202; 高斯噪声</h4>
                            <p>添加正态分布随机噪声，模拟电子干扰</p>
                        </div>
                        <div class="noise-type">
                            <h4>&#127752; HSV扰动</h4>
                            <p>改变色调、饱和度和亮度，破坏颜色特征</p>
                        </div>
                        <div class="noise-type">
                            <h4>&#9889;&#65039; 梯度对抗</h4>
                            <p>干扰图像边缘特征，对抗AI识别</p>
                        </div>
                    </div>
                </div>
            </div>
            
            <div class="preview-section">
                <div class="image-container">
                    <h3>原始图片</h3>
                    <div class="placeholder" id="originalPlaceholder">等待图片上传...</div>
                    <canvas id="originalCanvas"></canvas>
                </div>
                <div class="image-container">
                    <h3>添加噪声后</h3>
                    <div class="placeholder" id="resultPlaceholder">处理后的图片将显示在这里</div>
                    <canvas id="resultCanvas"></canvas>
                </div>
            </div>
            
            <div class="btn-group">
                <button class="btn-accent" id="processBtn">
                    <span>添加噪声</span>
                </button>
                <button class="btn-primary" id="downloadBtn" disabled>
                    <span>下载结果</span>
                </button>
                <button class="btn-secondary" id="resetBtn">
                    <span>重置</span>
                </button>
            </div>
        </div>
        
        <footer>
            <p>基于Python噪声算法实现的HTML5应用 | 使用Canvas进行图像处理</p>
            <p>适用于：AI对抗、数据增强、图像退化模拟</p>
        </footer>
    </div>

    <script>
        // DOM元素
        const fileInput = document.getElementById('fileInput');
        const intensitySlider = document.getElementById('intensitySlider');
        const intensityValue = document.getElementById('intensityValue');
        const processBtn = document.getElementById('processBtn');
        const downloadBtn = document.getElementById('downloadBtn');
        const resetBtn = document.getElementById('resetBtn');
        const originalCanvas = document.getElementById('originalCanvas');
        const resultCanvas = document.getElementById('resultCanvas');
        const originalCtx = originalCanvas.getContext('2d');
        const resultCtx = resultCanvas.getContext('2d');
        const dropZone = document.getElementById('dropZone');
        const originalPlaceholder = document.getElementById('originalPlaceholder');
        const resultPlaceholder = document.getElementById('resultPlaceholder');
        
        // 当前图片状态
        let originalImage = null;
        let processedImage = null;
        
        // 更新强度显示
        intensitySlider.addEventListener('input', () => {
            const value = intensitySlider.value / 100;
            intensityValue.textContent = value.toFixed(2);
        });
        
        // 文件上传处理
        fileInput.addEventListener('change', (e) => {
            if (e.target.files && e.target.files[0]) {
                const reader = new FileReader();
                
                reader.onload = (event) => {
                    const img = new Image();
                    img.onload = () => {
                        originalImage = img;
                        displayImage(img, originalCanvas, originalCtx, originalPlaceholder);
                        resultPlaceholder.textContent = "点击下方按钮添加噪声";
                        downloadBtn.disabled = true;
                    };
                    img.src = event.target.result;
                };
                
                reader.readAsDataURL(e.target.files[0]);
            }
        });
        
        // 拖放功能
        dropZone.addEventListener('dragover', (e) => {
            e.preventDefault();
            dropZone.style.borderColor = '#4a6fa5';
            dropZone.style.backgroundColor = 'rgba(255, 255, 255, 0.95)';
        });
        
        dropZone.addEventListener('dragleave', () => {
            dropZone.style.borderColor = '#ced4da';
            dropZone.style.backgroundColor = 'rgba(255, 255, 255, 0.7)';
        });
        
        dropZone.addEventListener('drop', (e) => {
            e.preventDefault();
            dropZone.style.borderColor = '#ced4da';
            dropZone.style.backgroundColor = 'rgba(255, 255, 255, 0.7)';
            
            if (e.dataTransfer.files && e.dataTransfer.files[0]) {
                fileInput.files = e.dataTransfer.files;
                const event = new Event('change');
                fileInput.dispatchEvent(event);
            }
        });
        
        // 显示图片到Canvas
        function displayImage(img, canvas, ctx, placeholder) {
            const maxWidth = 500;
            const maxHeight = 400;
            
            let width = img.width;
            let height = img.height;
            
            // 调整尺寸以适应容器
            if (width > maxWidth) {
                height = (maxWidth / width) * height;
                width = maxWidth;
            }
            
            if (height > maxHeight) {
                width = (maxHeight / height) * width;
                height = maxHeight;
            }
            
            canvas.width = width;
            canvas.height = height;
            ctx.clearRect(0, 0, canvas.width, canvas.height);
            ctx.drawImage(img, 0, 0, width, height);
            
            canvas.style.display = 'block';
            placeholder.style.display = 'none';
        }
        
        // 添加噪声的主函数
        function addNoise() {
            if (!originalImage) {
                alert('请先上传图片！');
                return;
            }
            
            // 显示加载状态
            processBtn.innerHTML = '<span>处理中...</span>';
            processBtn.disabled = true;
            
            // 复制原始图像到结果Canvas
            resultCanvas.width = originalCanvas.width;
            resultCanvas.height = originalCanvas.height;
            resultCtx.drawImage(originalImage, 0, 0, resultCanvas.width, resultCanvas.height);
            
            // 获取图像数据
            const imageData = resultCtx.getImageData(0, 0, resultCanvas.width, resultCanvas.height);
            const data = imageData.data;
            
            // 获取噪声强度 (0.0 ~ 0.5)
            const intensity = intensitySlider.value / 100;
            
            // 添加椒盐噪声
            addSaltPepperNoise(data, intensity);
            
            // 添加高斯噪声
            addGaussianNoise(data, intensity);
            
            // 添加HSV扰动
            addHSVNoise(data, intensity);
            
            // 添加梯度对抗扰动
            addGradientNoise(data, imageData, intensity);
            
            // 将处理后的数据放回Canvas
            resultCtx.putImageData(imageData, 0, 0);
            
            // 显示结果
            resultCanvas.style.display = 'block';
            resultPlaceholder.style.display = 'none';
            downloadBtn.disabled = false;
            
            // 恢复按钮状态
            processBtn.innerHTML = '<span>添加噪声</span>';
            processBtn.disabled = false;
        }
        
        // 椒盐噪声
        function addSaltPepperNoise(data, intensity) {
            const size = data.length;
            const noiseCount = Math.floor(size / 4 * intensity * 0.5);
            
            for (let i = 0; i < noiseCount; i++) {
                // 随机位置
                const idx = Math.floor(Math.random() * size / 4) * 4;
                
                // 椒盐选择 (0 或 255)
                const value = Math.random() > 0.5 ? 255 : 0;
                
                // 设置RGB通道
                data[idx] = value;     // R
                data[idx + 1] = value; // G
                data[idx + 2] = value; // B
            }
        }
        
        // 高斯噪声
        function addGaussianNoise(data, intensity) {
            const strength = intensity * 50;
            
            for (let i = 0; i < data.length; i += 4) {
                // 对每个通道添加高斯噪声
                for (let j = 0; j < 3; j++) {
                    // 生成高斯随机数
                    const noise = strength * (Math.random() - 0.5) * 2;
                    const newValue = data[i + j] + noise;
                    
                    // 确保值在0-255之间
                    data[i + j] = Math.max(0, Math.min(255, newValue));
                }
            }
        }
        
        // HSV扰动
        function addHSVNoise(data, intensity) {
            // 随机色调偏移
            const hueShift = (Math.random() * 20 - 10);
            // 饱和度增强系数
            const satFactor = 1 + 0.5 * intensity;
            // 亮度随机变化
            const valueFactor = 1 + (Math.random() - 0.5) * intensity * 2;
            
            for (let i = 0; i < data.length; i += 4) {
                let r = data[i] / 255;
                let g = data[i + 1] / 255;
                let b = data[i + 2] / 255;
                
                // 转换为HSV
                const max = Math.max(r, g, b);
                const min = Math.min(r, g, b);
                let h, s, v = max;
                
                const d = max - min;
                s = max === 0 ? 0 : d / max;
                
                if (max === min) {
                    h = 0; // achromatic
                } else {
                    switch (max) {
                        case r: h = (g - b) / d + (g < b ? 6 : 0); break;
                        case g: h = (b - r) / d + 2; break;
                        case b: h = (r - g) / d + 4; break;
                    }
                    h /= 6;
                }
                
                // 应用HSV扰动
                h = (h * 360 + hueShift) % 360;
                if (h < 0) h += 360;
                
                s = Math.min(1, s * satFactor);
                v = Math.min(1, v * valueFactor);
                
                // 转换回RGB
                h /= 60;
                const c = v * s;
                const x = c * (1 - Math.abs(h % 2 - 1));
                const m = v - c;
                
                let r1, g1, b1;
                
                if (h >= 0 && h < 1) {
                    [r1, g1, b1] = [c, x, 0];
                } else if (h >= 1 && h < 2) {
                    [r1, g1, b1] = [x, c, 0];
                } else if (h >= 2 && h < 3) {
                    [r1, g1, b1] = [0, c, x];
                } else if (h >= 3 && h < 4) {
                    [r1, g1, b1] = [0, x, c];
                } else if (h >= 4 && h < 5) {
                    [r1, g1, b1] = [x, 0, c];
                } else {
                    [r1, g1, b1] = [c, 0, x];
                }
                
                // 设置新RGB值
                data[i] = (r1 + m) * 255;
                data[i + 1] = (g1 + m) * 255;
                data[i + 2] = (b1 + m) * 255;
            }
        }
        
        // 梯度对抗扰动（简化版）
        function addGradientNoise(data, imageData, intensity) {
            const width = resultCanvas.width;
            const height = resultCanvas.height;
            const tempData = new Uint8ClampedArray(imageData.data);
            
            // 简化的梯度计算
            for (let y = 1; y < height - 1; y++) {
                for (let x = 1; x < width - 1; x++) {
                    const idx = (y * width + x) * 4;
                    
                    // 计算简单梯度
                    const dx = tempData[idx + 4] - tempData[idx - 4]; // 水平梯度
                    const dy = tempData[idx + width * 4] - tempData[idx - width * 4]; // 垂直梯度
                    
                    // 添加对抗扰动
                    const perturbation = intensity * 0.1 * Math.sign(dx + dy);
                    
                    // 应用扰动到RGB通道
                    for (let c = 0; c < 3; c++) {
                        const newValue = data[idx + c] + perturbation * 128;
                        data[idx + c] = Math.max(0, Math.min(255, newValue));
                    }
                }
            }
        }
        
        // 下载图片
        function downloadImage() {
            if (!resultCanvas.style.display || resultCanvas.style.display === 'none') {
                alert('请先处理图片！');
                return;
            }
            
            const link = document.createElement('a');
            link.download = 'noisy-image.png';
            link.href = resultCanvas.toDataURL('image/png');
            link.click();
        }
        
        // 重置
        function reset() {
            fileInput.value = '';
            intensitySlider.value = 30;
            intensityValue.textContent = '0.30';
            
            originalCanvas.style.display = 'none';
            resultCanvas.style.display = 'none';
            
            originalPlaceholder.style.display = 'flex';
            originalPlaceholder.textContent = '等待图片上传...';
            
            resultPlaceholder.style.display = 'flex';
            resultPlaceholder.textContent = '处理后的图片将显示在这里';
            
            downloadBtn.disabled = true;
            originalImage = null;
            processedImage = null;
        }
        
        // 绑定事件
        processBtn.addEventListener('click', addNoise);
        downloadBtn.addEventListener('click', downloadImage);
        resetBtn.addEventListener('click', reset);
    </script>
</body>
</html>

功能说明这个HTML应用实现了以下功能：

图片上传：
- 支持文件选择器上传
- 支持拖放上传
- 实时预览原始图片
噪声控制：
- 可调节的噪声强度滑块（0.0-0.5）
- 实时显示当前噪声强度值
噪声类型：
- 椒盐噪声：随机添加黑白像素点
- 高斯噪声：添加正态分布随机噪声
- HSV扰动：改变色调、饱和度和亮度
- 梯度对抗扰动：干扰图像边缘特征
结果处理：
- 实时预览添加噪声后的图片
- 下载处理后的图片
- 重置功能

使用说明

点击"选择图片"按钮或拖放图片到上传区域
调整噪声强度滑块（推荐0.1-0.3）
点击"添加噪声"按钮进行处理
查看右侧预览效果
满意后点击"下载结果"保存图片

这个工具适用于：

突破AI重复图片检测
机器学习数据增强
图像退化模拟
图像处理研究

提示：对于大尺寸图片，处理可能需要几秒钟时间，请耐心等待。

chimun666 · 发表于 2025-7-15 09:40

試試看看

我也有用AI寫過用1張相搜尋電腦內過萬張相中有沒有相同不是搜電腦內有多少不同喔

yangyanghq · 发表于 2025-7-14 20:55

虽然不会，但还是要支持一下贴主的用心

hlovet · 发表于 2025-7-15 00:03

虽然不会，但还是要支持一下贴主的用心

滴水菩提 · 发表于 2025-7-15 05:56

魔高一尺道高一丈

yxf515321 · 发表于 2025-7-15 07:31

应该有一个适用范围吧

quintar · 发表于 2025-7-15 08:16

太专业了~

anning666 · 发表于 2025-7-15 09:31

大佬~~~那个AI重复图片检测的东东,能发粗来观摩学习一下吗

79887220 · 发表于 2025-7-15 10:52

厉害了，我的哥.......................

4leafcolver · 发表于 2025-7-15 11:34

学习一下，也许下次P图用得到，哈哈

帐号		自动登录	找回密码
密码			注册[Register]

[Python 原创] 用Python实现添加图片噪声

免费评分

本帖被以下淘专辑推荐:

免费评分