Pillow 图像处理

pip install pillow

from PIL import Image

im = Image.open('photo.jpg')
w, h = im.size
print(f'原始尺寸: {w}x{h}')

# ── thumbnail：等比缩放（原地修改，不返回新对象）─────────
im_thumb = im.copy()
im_thumb.thumbnail((w // 2, h // 2))
print(f'缩略图尺寸: {im_thumb.size}')
im_thumb.save('thumbnail.jpg', 'JPEG', quality=85)

# ── resize：自由调整尺寸（返回新对象）────────────────────
im_resized = im.resize((800, 600), Image.LANCZOS)  # LANCZOS 高质量重采样
im_resized.save('resized.jpg', 'JPEG', quality=85)

from PIL import Image

im = Image.open('photo.jpg')

# ── 裁剪：(left, upper, right, lower) 坐标系 ──────────
# 左上角为 (0, 0)，向右向下递增
im_crop = im.crop((100, 100, 400, 400))
im_crop.save('crop.jpg', 'JPEG')

# ── 旋转 ──────────────────────────────────────────────
# 逆时针旋转 90°，expand=True 自动扩展画布
im_rotate = im.rotate(90, expand=True)
im_rotate.save('rotate90.jpg', 'JPEG')

# ── 翻转（比 rotate 更高效）────────────────────────────
im_flip_h = im.transpose(Image.FLIP_LEFT_RIGHT)   # 水平翻转
im_flip_v = im.transpose(Image.FLIP_TOP_BOTTOM)    # 垂直翻转
im_rot_90 = im.transpose(Image.ROTATE_90)          # 顺时针 90°
im_rot_180 = im.transpose(Image.ROTATE_180)        # 180°
im_rot_270 = im.transpose(Image.ROTATE_270)        # 顺时针 270°

from PIL import Image, ImageFilter, ImageEnhance

im = Image.open('photo.jpg')

# ── 预置滤镜 ──────────────────────────────────────────
im_blur = im.filter(ImageFilter.BLUR)                    # 均值模糊
im_gauss = im.filter(ImageFilter.GaussianBlur(radius=5)) # 高斯模糊（更自然）
im_sharp = im.filter(ImageFilter.SHARPEN)                # 锐化
im_contour = im.filter(ImageFilter.CONTOUR)              # 轮廓提取
im_emboss = im.filter(ImageFilter.EMBOSS)                # 浮雕
im_edges = im.filter(ImageFilter.FIND_EDGES)             # 边缘检测

# ── 自定义卷积核 ──────────────────────────────────────
# 3x3 锐化核
kernel = ImageFilter.Kernel(
    (3, 3),                    # 核尺寸
    [0, -1, 0,                # 核数据（行优先）
     -1, 5, -1,
     0, -1, 0],
    1,                         # 缩放因子
    0                          # 偏移量
)
im_custom = im.filter(kernel)

# ── 图像增强 ──────────────────────────────────────────
# 亮度：0=全黑, 1=原图, 2=两倍亮度
im_bright = ImageEnhance.Brightness(im).enhance(1.5)

# 对比度：0=全灰, 1=原图, 2=两倍对比度
im_contrast = ImageEnhance.Contrast(im).enhance(2.0)

# 色彩饱和度：0=黑白, 1=原图, 2=两倍饱和度
im_color = ImageEnhance.Color(im).enhance(1.5)

# 锐度：0=模糊, 1=原图, 2=两倍锐度
im_sharpness = ImageEnhance.Sharpness(im).enhance(2.0)

from PIL import Image

im = Image.open('photo.jpg').convert('RGB')

# ── 通道分离 ──────────────────────────────────────────
r, g, b = im.split()  # 返回三个灰度图像（'L' 模式）

# 单独处理某个通道（如增强红色通道）
from PIL import ImageEnhance
r_enhanced = ImageEnhance.Brightness(r).enhance(1.5)

# ── 通道合并 ──────────────────────────────────────────
im_new = Image.merge('RGB', (r_enhanced, g, b))
im_new.save('red_enhanced.jpg', 'JPEG')

# ── 提取 Alpha 通道 ──────────────────────────────────
im_rgba = Image.open('logo.png')  # 带 Alpha 通道的 PNG
r, g, b, a = im_rgba.split()
# a 是透明度通道，可用于遮罩操作

# ── 通道运算（ImageChops）─────────────────────────────
from PIL import ImageChops

im1 = Image.open('image1.jpg')
im2 = Image.open('image2.jpg')

# 加法（两图叠加，超过 255 截断）
added = ImageChops.add(im1, im2)

# 差值（检测两图差异，用于图像对比/变化检测）
diff = ImageChops.difference(im1, im2)

# 乘法（暗化效果）
multiplied = ImageChops.multiply(im1, im2)

# 反相
inverted = ImageChops.invert(im1)

from PIL import Image

im = Image.open('photo.jpg')

# ── 访问单个像素 ──────────────────────────────────────
pixel = im.getpixel((100, 100))  # 返回 (R, G, B) 或 (R, G, B, A)

# ── 修改单个像素 ──────────────────────────────────────
im.putpixel((100, 100), (255, 0, 0))  # 设为红色

# ── point()：批量像素变换（比逐像素快 100 倍）─────────
# 灰度反转
im_inverted = im.point(lambda p: 255 - p)

# 阈值化（二值图像）
im_bw = im.convert('L').point(lambda p: 255 if p > 128 else 0, '1')

# 亮度映射（gamma 校正）
import math
gamma = 1.5
im_gamma = im.point(lambda p: int(255 * (p / 255) ** (1 / gamma)))

# ── 使用 numpy 进行高级像素操作 ──────────────────────
import numpy as np

arr = np.array(im)                    # Image → numpy 数组 (H, W, C)
arr = arr[:, :, ::-1]                 # RGB → BGR（OpenCV 格式）
im_from_arr = Image.fromarray(arr)    # numpy 数组 → Image

from PIL import Image, ImageDraw, ImageFont

# 创建画布
im = Image.new('RGB', (800, 600), 'white')
draw = ImageDraw.Draw(im)

# ── 基础图形 ──────────────────────────────────────────
# 线段
draw.line([(0, 0), (800, 600)], fill='red', width=3)

# 矩形（outline=描边, fill=填充）
draw.rectangle([50, 50, 200, 150], outline='blue', fill='lightblue', width=2)

# 圆角矩形（Python 3.x Pillow 8.2+）
draw.rounded_rectangle([250, 50, 400, 150], radius=15, fill='green')

# 椭圆/圆
draw.ellipse([450, 50, 600, 200], outline='purple', fill='lavender', width=2)

# 多边形
draw.polygon([(100, 300), (200, 250), (300, 300), (250, 400), (150, 400)],
             fill='orange', outline='darkorange')

# 弧线
draw.arc([50, 300, 200, 450], start=0, end=180, fill='red', width=3)

# ── 文字绘制 ──────────────────────────────────────────
try:
    font = ImageFont.truetype('/System/Library/Fonts/PingFang.ttc', 36)
except IOError:
    font = ImageFont.load_default()

draw.text((300, 400), 'Hello Pillow', fill='black', font=font)

# 获取文字尺寸（用于精确定位）
bbox = draw.textbbox((0, 0), 'Hello Pillow', font=font)
text_width = bbox[2] - bbox[0]
text_height = bbox[3] - bbox[1]

# 居中绘制
x = (800 - text_width) // 2
y = (600 - text_height) // 2
draw.text((x, y), 'Centered Text', fill='navy', font=font)

im.save('drawing.png', 'PNG')

from PIL import Image
from PIL.ExifTags import TAGS, GPSTAGS

im = Image.open('photo.jpg')

# ── 读取 EXIF 数据 ────────────────────────────────────
exif_data = im.getexif()

# 解码 EXIF 标签
for tag_id, value in exif_data.items():
    tag_name = TAGS.get(tag_id, tag_id)
    print(f"{tag_name}: {value}")

# 常用 EXIF 字段
print(f"相机型号: {exif_data.get(272, 'N/A')}")       # 272 = Make
print(f"拍摄时间: {exif_data.get(36867, 'N/A')}")      # 36867 = DateTimeOriginal
print(f"曝光时间: {exif_data.get(33434, 'N/A')}")      # 33434 = ExposureTime
print(f"ISO 感光度: {exif_data.get(34855, 'N/A')}")    # 34855 = ISOSpeedRatings
print(f"焦距: {exif_data.get(37386, 'N/A')}")          # 37386 = FocalLength

# ── 根据 EXIF 方向自动旋转 ────────────────────────────
# 手机拍摄的照片经常方向信息在 EXIF 中，需要根据 Orientation 旋转
from PIL import ImageOps
im_auto = ImageOps.exif_transpose(im)  # 自动根据 EXIF Orientation 旋转

# ── 删除 EXIF（隐私保护）───────────────────────────────
im_no_exif = im.copy()
im_no_exif.info.pop('exif', None)
im_no_exif.save('no_exif.jpg', 'JPEG', quality=95)

from PIL import Image

# ── 读取 GIF 信息 ─────────────────────────────────────
im = Image.open('animation.gif')
print(f"帧数: {im.n_frames}")
print(f"是否动画: {im.is_animated}")

# ── 逐帧处理 GIF ──────────────────────────────────────
frames = []
for i in range(im.n_frames):
    im.seek(i)  # 跳转到第 i 帧
    # 对每一帧进行处理（如缩放）
    frame = im.copy().resize((200, 200), Image.LANCZOS)
    frames.append(frame)

# ── 保存为 GIF 动画 ──────────────────────────────────
frames[0].save(
    'resized_animation.gif',
    save_all=True,           # 保存所有帧
    append_images=frames[1:], # 追加剩余帧
    duration=100,            # 每帧持续时间（ms）
    loop=0,                  # 循环次数（0=无限循环）
    optimize=True,           # 优化文件大小
)

# ── 从多张静态图创建 GIF ──────────────────────────────
images = [Image.open(f'frame_{i}.png') for i in range(10)]
images[0].save(
    'output.gif',
    save_all=True,
    append_images=images[1:],
    duration=200,
    loop=0,
)

from pathlib import Path
from PIL import Image

def batch_resize(input_dir: str, output_dir: str,
                 size: tuple = (800, 600), quality: int = 85) -> None:
    """批量缩放图片"""
    out_path = Path(output_dir)
    out_path.mkdir(parents=True, exist_ok=True)
    supported = {'.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp'}

    for img_file in Path(input_dir).iterdir():
        if img_file.suffix.lower() not in supported:
            continue
        try:
            im = Image.open(img_file)
            im.thumbnail(size, Image.LANCZOS)

            # RGBA → RGB（JPEG 不支持透明通道）
            if im.mode == 'RGBA':
                bg = Image.new('RGB', im.size, (255, 255, 255))
                bg.paste(im, mask=im.split()[3])
                im = bg

            im.save(out_path / img_file.name, quality=quality)
            print(f"✅ {img_file.name} → {im.size}")
        except Exception as e:
            print(f"❌ {img_file.name}: {e}")

batch_resize('./images', './images/resized', size=(800, 600))

from PIL import Image, ImageDraw, ImageFont

def add_text_watermark(input_path: str, output_path: str,
                       text: str = "Sample", font_size: int = 36,
                       opacity: int = 128) -> None:
    """添加文字水印"""
    im = Image.open(input_path).convert('RGBA')

    watermark = Image.new('RGBA', im.size, (0, 0, 0, 0))
    draw = ImageDraw.Draw(watermark)

    try:
        font = ImageFont.truetype('/System/Library/Fonts/PingFang.ttc', font_size)
    except IOError:
        font = ImageFont.load_default()

    # 右下角定位
    bbox = draw.textbbox((0, 0), text, font=font)
    text_w, text_h = bbox[2] - bbox[0], bbox[3] - bbox[1]
    x, y = im.width - text_w - 20, im.height - text_h - 20
    draw.text((x, y), text, font=font, fill=(255, 255, 255, opacity))

    result = Image.alpha_composite(im, watermark)
    result.convert('RGB').save(output_path, 'JPEG', quality=90)

def add_image_watermark(input_path: str, watermark_path: str,
                        output_path: str, position: str = 'bottom-right',
                        margin: int = 10, opacity: int = 128) -> None:
    """添加图片水印"""
    im = Image.open(input_path).convert('RGBA')
    logo = Image.open(watermark_path).convert('RGBA')
    logo.thumbnail((im.width // 4, im.height // 4), Image.LANCZOS)

    # 调整透明度
    alpha = logo.split()[3].point(lambda p: p * opacity // 255)
    logo.putalpha(alpha)

    positions = {
        'top-left': (margin, margin),
        'top-right': (im.width - logo.width - margin, margin),
        'bottom-left': (margin, im.height - logo.height - margin),
        'bottom-right': (im.width - logo.width - margin, im.height - logo.height - margin),
        'center': ((im.width - logo.width) // 2, (im.height - logo.height) // 2),
    }
    im.paste(logo, positions.get(position, positions['bottom-right']), logo)
    im.convert('RGB').save(output_path, 'JPEG', quality=90)

from pathlib import Path
from PIL import Image

def batch_convert(input_dir: str, output_dir: str,
                  target_format: str = 'WEBP', quality: int = 85) -> None:
    """批量转换图片格式（推荐转 WEBP，体积减少 30-50%）"""
    out_path = Path(output_dir)
    out_path.mkdir(parents=True, exist_ok=True)
    supported = {'.jpg', '.jpeg', '.png', '.bmp', '.gif', '.webp', '.tiff'}

    for img_file in Path(input_dir).iterdir():
        if img_file.suffix.lower() not in supported:
            continue
        try:
            im = Image.open(img_file)
            if target_format.upper() in ('JPEG', 'JPG') and im.mode in ('RGBA', 'P'):
                im = im.convert('RGB')

            out_file = out_path / f"{img_file.stem}.{target_format.lower()}"
            save_kwargs = {}
            if target_format.upper() in ('JPEG', 'JPG', 'WEBP'):
                save_kwargs['quality'] = quality
            im.save(out_file, target_format, **save_kwargs)
            print(f"✅ {img_file.name} → {out_file.name}")
        except Exception as e:
            print(f"❌ {img_file.name}: {e}")

# WebP 格式：同等质量下比 JPEG 小 25-35%，比 PNG 小 80%+
batch_convert('./images', './images/webp', target_format='WEBP', quality=80)

from PIL import Image, ImageDraw, ImageFont, ImageFilter
import random
import string

def generate_captcha(width: int = 240, height: int = 60,
                     length: int = 4, font_size: int = 36) -> tuple[str, Image.Image]:
    """生成字母验证码图片，返回 (验证码文字, 图像对象)"""
    # 随机验证码文字
    chars = ''.join(random.choices(string.ascii_uppercase + string.digits, k=length))

    image = Image.new('RGB', (width, height), (255, 255, 255))
    draw = ImageDraw.Draw(image)

    try:
        font = ImageFont.truetype('/System/Library/Fonts/Menlo.ttc', font_size)
    except IOError:
        font = ImageFont.load_default()

    # 填充背景噪点
    for x in range(width):
        for y in range(height):
            if random.random() < 0.3:
                draw.point((x, y), fill=(
                    random.randint(64, 255),
                    random.randint(64, 255),
                    random.randint(64, 255),
                ))

    # 绘制文字（每个字符随机偏移和旋转）
    char_width = width // length
    for i, ch in enumerate(chars):
        x = char_width * i + random.randint(5, 15)
        y = random.randint(5, 15)
        color = (random.randint(32, 127), random.randint(32, 127), random.randint(32, 127))
        draw.text((x, y), ch, font=font, fill=color)

    # 添加干扰线
    for _ in range(3):
        x1, y1 = random.randint(0, width), random.randint(0, height)
        x2, y2 = random.randint(0, width), random.randint(0, height)
        draw.line([(x1, y1), (x2, y2)], fill=(
            random.randint(32, 127), random.randint(32, 127), random.randint(32, 127)
        ), width=1)

    # 轻微模糊增加识别难度
    image = image.filter(ImageFilter.GaussianBlur(0.5))

    return chars, image

# 使用
code, img = generate_captcha()
img.save('captcha.jpg', 'JPEG')
print(f"验证码: {code}")

from PIL import Image
from pathlib import Path

def create_long_image(image_paths: list[str], direction: str = 'vertical',
                      gap: int = 0, bg_color: str = 'white') -> Image.Image:
    """
    将多张图片拼接为长图。

    Args:
        image_paths: 图片路径列表
        direction: 'vertical'（竖向拼接）或 'horizontal'（横向拼接）
        gap: 图片间距（像素）
        bg_color: 背景色
    """
    images = [Image.open(p) for p in image_paths]

    if direction == 'vertical':
        max_width = max(im.width for im in images)
        total_height = sum(im.height for im in images) + gap * (len(images) - 1)
        result = Image.new('RGB', (max_width, total_height), bg_color)

        y_offset = 0
        for im in images:
            # 居中对齐
            x = (max_width - im.width) // 2
            result.paste(im, (x, y_offset))
            y_offset += im.height + gap
    else:
        max_height = max(im.height for im in images)
        total_width = sum(im.width for im in images) + gap * (len(images) - 1)
        result = Image.new('RGB', (total_width, max_height), bg_color)

        x_offset = 0
        for im in images:
            y = (max_height - im.height) // 2
            result.paste(im, (x_offset, y))
            x_offset += im.width + gap

    return result

# 使用：竖向拼接聊天截图
paths = sorted(Path('./screenshots').glob('screen_*.png'))
long_img = create_long_image([str(p) for p in paths], direction='vertical', gap=10)
long_img.save('long_screenshot.png', 'PNG')