Shulou(Shulou.com)06/02 Report--

This article mainly explains "how to use OpenCV to achieve image restoration", the content of the article is simple and clear, easy to learn and understand, the following please follow the editor's ideas slowly in depth, together to study and learn "how to use OpenCV to achieve image restoration" bar!

Where is the application of image restoration technology?

Come to think of it, we have a great photo, but because of the long time, there is no electronic version, and the paper version is very difficult to save. Therefore, long uses the scanning method to obtain the electronic version. But unfortunately, a hair fell in the scanning process, or the machine malfunctioned, which affected the photo. At this time, the problem can be solved by image restoration technology.

The powerful OpenCV library integrates two methods for image restoration

INPAINT_NS: image restoration based on Navier-Stokes

This method was proposed in 2001, and its magic is based on the theory of fluid mechanics. According to its author, the problem we need to solve can be abstracted as having a black area on a shoe image to make the best way to restore the appearance of the shoe by filling the black area.

As to how to fill this black area, it can be abstracted into a curve that separates the black area from A to B and ensures that it is blue on one side of the curve and white on the other. This curve should have the following constraints:

Maintain edge featur

Maintain color information in a smooth area

A partial differential equation is constructed to update the image intensity in the region with appeal constraints. at the same time, the Laplace operator is used to estimate the image smoothness information and propagate along the equal illuminance.

Because these equations are related to and similar to Navier-Stokes equations (equations in fluid mechanics, interested partners can do so on their own), they can be solved by methods in fluid mechanics.

As Xiaobai does not know much about hydrodynamics, it will not be explained in detail. Interested partners can read this paper for details.

Paper address: http://www.math.ucla.edu/~bertozzi/papers/cvpr01.pdf

INPAINT_TELEA: image restoration based on fast marching method

In this method, instead of using Laplace operator as the estimation of smoothness, it is described by the weighted average of the known image neighborhood of pixels. At the same time, the neighborhood pixels and the gradient are used to restore the color of the filled region pixels.

When the pixel is repaired, the boundary is updated by the fast marching method.

Paper address:

Https://pdfs.semanticscholar.org/622d/5f432e515da69f8f220fb92b17c8426d0427.pdf

Introduction to related API

Caterpillar:

Void inpaint (

Mat&src

Mat& inpaintMask

Mat& dst

Double inpaintRadius

Int flags)

Python:

Dst = cv2.inpaint (

Src

InpaintMask

InpaintRadius

Flags)

The meanings of the parameters are as follows:

Src = source image

InpaintMask = binary mask, indicating the pixels to repair.

DST = target image

InpaintRadius = the neighborhood around the pixel. In general, if the area to be repaired is thin, using smaller values will result in less blur.

Flags: INPAINT_NS (Navier-Stokes-based approach) or INPAINT_TELEA (fast-marching based approach)

Example

The first image on the left is the input image, the second image is the mask, the third image is the result of INPAINT_TELEA, and the final result is INPAINT_NS

There is a small story about this picture. Alexander Gardner took several multi-lens photos of the president at the Gardner Gallery in Washington, D.C., on Sunday, February 5, 1865. Before the end of the session, Gardner asked for a final pose for the president. He pulled the camera closer and took a picture of Lincoln's head, shoulders and chest. But the glass plate suddenly broke at this time, affecting the image. Gardner carefully took it to his dark room, made a picture and found an ominous crack in Lincoln's face. This photo, Omur118, still exists today. For years, many people thought the rift was a prophecy that Lincoln would be shot 10 weeks later.

Let's look at a more complex example, write an English word on the picture, and then fix the word through the opencv function.

Left: the original image with Scribbles. Middle: fix using fast marching method, right: fix using Navier-Stokes method.

The source code of the program and the picture links used are:

Https://github.com/spmallick/learnopencv/tree/master/Image-Inpainting

Thank you for your reading, the above is the content of "how to use OpenCV to achieve image restoration". After the study of this article, I believe you have a deeper understanding of how to use OpenCV to achieve image restoration, and the specific use needs to be verified in practice. Here is, the editor will push for you more related knowledge points of the article, welcome to follow!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.