消除图像中的噪点 [英] remove noise from image

问题描述

嗨..请我需要帮助..
我在加水印后提取的图像中有一条灰色的水平线
我如何删除这些行?！?

Hi..please i need a help..
There are a gray horizontal lines in the image that i have extracted after watermarking
how can i remove these lines ?!?

推荐答案

对不起，如果它对您没有帮助.在忙碌的背景下，这很难做到使更改不引人注目，因为人眼具有空前的集成能力.

解决此问题的简单理论障碍是，原始图像中的信息部分(带有线下"的部分)将永远丢失.通常，您必须以某种方式进行弥补.

如果该水平线附近的背景严格单调，则该问题的解决方案几乎是微不足道的.您只需要用背景像素替换该行中的像素(实际上是一些较近的像素)即可.这种情况很少发生；而且我认为您是否会问这个问题.

如果背景非常平滑，则解决方案基本上可以是这样的:在水平线的每个点拾取一条垂直线，将所有像素放置在相交点的较宽范围内(即该点的下方和上方) )，然后对相交点的像素进行插值(更确切地说，是在覆盖相交点附近的一条短垂直线中进行插值.最好对替换后的像素，其值以及替换后的区域的位置和大小进行些许随机化.这里的问题是一样的:要愚弄人眼即使看到最细微的相关性也是极其困难的，另一种有效的方法是通过引入轻微的模糊来稍微降低图像的质量，但是我不是确定是否要牺牲整体质量.

最后，如果要删除的线路经过繁忙的后台，则该问题几乎是无望的.我尝试手动进行这样的思考，但并不十分成功.同时，我相对经验丰富的眼睛可以帮助我看到大多数伪造迹象.只有真正的大师才能达到几乎完美的结果.并且需要大量的耐心和丰富的经验.

有趣的是，此问题仅与明显的线条(例如划痕)有关.许多其他缺陷(例如灰尘，甚至是人类皮肤上的斑点)也很容易修复.

—SA

Sorry if it is not helping you. With a busy background, this is extremely difficult to do it the way which would make the change unnoticeable, because a human eye has unprecedented integration ability.

The simple theoretical obstacle for solving this problem is that the part of information from original image (the part with would be "under the line") is lost forever. In general case, you have to make it up somehow.

If the background in the vicinity of this horizontal line is strictly monotonous, the solution of this problem is almost trivial; you should simply replace the pixels in the line (in fact, is some close vicinity) with the background pixels. This is rarely the case; and I don''t think you would ask this question if it was the case.

If the background is very smooth, the solution is basically could be something like this: pick up a vertical line at each point of a horizontal line, take all pixel in some wide vicinity of the intersection points (that is, below and above the point) and interpolate the pixels of the intersection points (more exactly, in a short vertical line covering a close vicinity of the intersection points. It would be good to slightly randomize the replaced pixels, their values and the location and size of the replaced area. The problem here is the same: it is extremely hard to foolish a human eye which can see even the most subtle correlations. One other effective way is to slightly degrade the quality of the image by introducing of minor blur, but I''m not sure if you want to sacrifice the overall quality.

And finally, if the line to be removed goes through the busy background, the problem is nearly hopeless. I tried to do such thinks manually, and was not quite successful. At the same time, my relatively experienced eye helps me to see most signs of forgery. Only the real masters can achieve nearly flawless results; and it takes a lot of patience and enormous experience.

Interestingly, this problem is only related to distinct lines, such as scratches. Many other defects, such as dust and even the blemishes on the human skin, are pretty easy to fix.

—SA

这篇关于消除图像中的噪点的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！