如何裁剪UIImageView中UIImage的可见区域用于显著性检测？

• 问题根源：你之前用imageView.frame作为裁剪矩形完全错误——这个矩形是视图在屏幕上的位置，而cgImage.cropping(to:)需要的是原图的像素坐标，而且没考虑ScaleAspectFill的缩放逻辑：这种模式下图片会按比例放大/缩小到填满视图，超出部分被裁剪，必须先算出视图里显示的那部分对应原图的具体区域。

• 正确实现步骤：

  1. 计算图片相对于UIImageView的缩放比例：取宽、高缩放系数中的较大值（因为ScaleAspectFill要填充满视图）。
  2. 算出缩放后图片的尺寸，再计算居中显示时的偏移量（超出视图的部分就是被裁剪掉的区域）。
  3. 将视图的可见区域转换为原图的像素坐标，同时处理图片方向问题（避免裁剪结果错位）。

完整代码实现 #

// 计算UIImageView可见区域对应的原图CGRect
func visibleRectInOriginalImage(for imageView: UIImageView) -> CGRect? {
    guard let image = imageView.image else { return nil }
    
    let viewSize = imageView.bounds.size
    let imageSize = image.size
    
    // ScaleAspectFill模式下取宽高缩放比例的较大值
    let scaleWidth = viewSize.width / imageSize.width
    let scaleHeight = viewSize.height / imageSize.height
    let scale = max(scaleWidth, scaleHeight)
    
    // 缩放后的图片尺寸
    let scaledImageWidth = imageSize.width * scale
    let scaledImageHeight = imageSize.height * scale
    
    // 居中显示时的偏移量（超出视图的部分会被裁剪）
    let offsetX = (scaledImageWidth - viewSize.width) / 2
    let offsetY = (scaledImageHeight - viewSize.height) / 2
    
    // 转换为原图的坐标
    let originalX = offsetX / scale
    let originalY = offsetY / scale
    let originalWidth = viewSize.width / scale
    let originalHeight = viewSize.height / scale
    
    // 处理图片方向，修正坐标（UIImage的imageOrientation会影响CGImage的实际布局）
    var rect = CGRect(x: originalX, y: originalY, width: originalWidth, height: originalHeight)
    switch image.imageOrientation {
    case .left:
        rect = CGRect(x: originalY, y: imageSize.width - originalX - originalWidth, width: originalHeight, height: originalWidth)
    case .right:
        rect = CGRect(x: imageSize.height - originalY - originalHeight, y: originalX, width: originalHeight, height: originalWidth)
    case .down:
        rect = CGRect(x: imageSize.width - originalX - originalWidth, y: imageSize.height - originalY - originalHeight, width: originalWidth, height: originalHeight)
    case .leftMirrored:
        rect = CGRect(x: imageSize.height - originalY - originalHeight, y: originalX, width: originalHeight, height: originalWidth)
    case .rightMirrored:
        rect = CGRect(x: originalY, y: imageSize.width - originalX - originalWidth, width: originalHeight, height: originalWidth)
    case .downMirrored:
        rect = CGRect(x: originalX, y: originalY, width: originalWidth, height: originalHeight)
    case .upMirrored:
        rect = CGRect(x: imageSize.width - originalX - originalWidth, y: imageSize.height - originalY - originalHeight, width: originalWidth, height: originalHeight)
    default:
        break
    }
    
    return rect
}

// 针对原图坐标的裁剪函数，修正CGImage与UIImage的坐标差异
func cropImage(image: UIImage?, toRect rect: CGRect) -> UIImage? {
    guard let image = image, let cgImage = image.cgImage else { return nil }
    
    // CGImage以左下角为原点，UIImage以左上角为原点，转换y坐标
    let cgRect = CGRect(
        x: rect.origin.x,
        y: image.size.height - rect.origin.y - rect.size.height,
        width: rect.size.width,
        height: rect.size.height
    )
    
    guard let croppedCGImage = cgImage.cropping(to: cgRect) else { return nil }
    // 保留原图的scale和orientation，避免图片变形或方向错误
    return UIImage(cgImage: croppedCGImage, scale: image.scale, orientation: image.imageOrientation)
}

使用示例 #

// 获取UIImageView可见区域对应的原图矩形
if let visibleRect = visibleRectInOriginalImage(for: yourImageView),
   let croppedImage = cropImage(image: yourImageView.image, toRect: visibleRect) {
    // 用裁剪后的图片发起显著性检测请求
    sendAttentionBasedSaliencyRequest(image: croppedImage)
}

内容的提问来源于stack exchange，提问作者Karina