对图像进行上采样
我有一个基本问题。
- 对图像进行上采样有哪些优点?
- 它对我的边缘检测有帮助吗?
我在互联网上没有找到太多有用的信息。
I have a basic question.
- What are the advantages of upsampling an Image?
- Does it help me in edge detection?
I have not found much useful information on the internet.
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(2)
这取决于图像。如果您的边缘非常锯齿状,它会有所帮助。最坏的情况下它什么也不做。因此,您需要为潜在的改进付出处理时间的代价。
It depends on the image. It can help if you have extremely jagged edges. At the worst it does nothing. So, you pay in processing time for a potential improvement.
为此,有两种可能的选择:
例如,您可能想要在缩小的图像上进行计算(例如分割),稍后您想要处理再次原始图像数据,因此您再次放大输出(例如分割蒙版)。
通常使用边缘检测器(例如canny,不仅仅是梯度计算)连接模糊算法。如果您在预处理中使用某种模糊蒙版,则可以通过其修改(减少或增加模糊能力)获得与图像调整大小类似的行为。
For this, there are two possible options:
As an example, you could want to do your calculations (e.g. segmentation) on a downsized Image, later on you want to work on the original Image data again, so you upsize your Output (e.g. Segmentation mask) again.
With edge detectors (e.g. canny, not only gradient computation) a blurring algorithm is usually connected. If you use some sort of blurring mask in preprocessing, it is possible that you can obtain similar behavior by its modification (decreasing, or increasing power of blurring) as in the case of image resize.