PIL 图像裁剪和调整大小降低质量

Question

我正在编写一个图像处理脚本。

1. 点击后台删除API
2. 检测图像中的项目边界
3. 根据边界坐标裁剪图像并将其放置在 1400 x 1000 画布上
4. 保存为PNG

我从背景去除 API 返回的图像具有完美的颜色和质量。

但是一旦我裁剪，放在画布上并保存，颜色就不那么鲜艳了，而且质量也不那么好。

下面是我的裁剪和保存功能。

def crop_and_resize_sneaker(bg_removed_image_path, bounding_box, output_dir, image_name):
    """
    Crops and resizes an image to fit a 1400 x 1000 canvas while maintaining quality.
    Uses PIL to handle images with an emphasis on preserving the original quality,
    especially during resizing and saving operations.
    """
    try:
        # Load the image with background removed
        image = Image.open(bg_removed_image_path).convert('RGBA')

        # Extract bounding box coordinates
        x, y, w, h = bounding_box

        # Crop the image based on the bounding box
        cropped_image = image.crop((x, y, x + w, y + h))

        # Define output dimensions
        output_width, output_height = 1400, 1000

        # Calculate new dimensions to maintain aspect ratio
        aspect_ratio = w / h
        if aspect_ratio > (output_width / output_height):
            new_width = output_width
            new_height = int(output_width / aspect_ratio)
        else:
            new_width = int(output_height * aspect_ratio)
            new_height = output_height

        # Resize the image using LANCZOS (high-quality)
        resized_image = cropped_image.resize((new_width, new_height), Image.LANCZOS)

        # Create a new image for the final output with a transparent background
        final_image = Image.new('RGBA', (output_width, output_height), (0, 0, 0, 0))

        # Calculate center positioning
        start_x = (output_width - new_width) // 2
        start_y = (output_height - new_height) // 2

        # Paste the resized image onto the transparent canvas
        final_image.paste(resized_image, (start_x, start_y), resized_image)

        # Save the final image as PNG with maximum quality settings
        final_img_path = os.path.join(output_dir, 'resized', f'{image_name}_sneaker_canvas.png')
        final_image.save(final_img_path, 'PNG', quality=95)  # Although 'quality' has no effect on PNGs, provided for completeness

        return final_img_path
    except Exception as e:
        logging.error(f"Error in cropping and resizing sneaker: {e}")
        return None

如何确保裁剪和调整图像大小与输入图像（删除背景）具有相同的质量？

Answer 1

问题是原始 PNG 具有嵌入的颜色配置文件，并且在保存新图像时该配置文件没有被传输。

下面的代码解决了这个问题：

profile = resized_image.info.get("icc_profile", "")
final_image.save(final_img_path, format="PNG", icc_profile=profile)