• Two valued

The two methods , Global fixed threshold binarization and local adaptive threshold binarization

The global fixed threshold is easy to understand , That is to binarize the whole image with a unified threshold ;

The local adaptive threshold is to determine the binarization threshold of the pixel position according to the pixel value distribution of the neighborhood block of the pixel .

• Discrete Fourier transform

The high frequency part of the image represents the details of the image 、 Texture information ; The low frequency represents the contour information of the image .

• Lowpass -》 Fuzzy
• qualcomm -》 sharpening
• Corrosion and expansion are aimed at the white part （ Highlight part ） In terms of the . Dilation is to make the highlight of the image “ Domain expansion ”, The rendering has a larger highlight area than the original ; Erosion is the erosion of highlighted areas in the original image , The effect image has a smaller highlight area than the original image .
• Open operation ： Corrosion before expansion , To eliminate small objects
• Closed operation ： Expand before corrode , Used to exclude small black holes
• Morphological gradients ： Is the difference between the expansion view and the top view , Used to preserve the edge contour of an object .
• Top hat ： The difference between the original image and the operation diagram , It is used to separate patches that are lit up next to each other .
• Black hat ： The difference between the closed operation and the original image , Used to separate patches darker than adjacent points .

opencv There is a good function in getStructuringElement, We just need to pass the corresponding processing parameters to this function , Then you can carry out the corresponding operation , It is very convenient to use .

The following lists the corresponding operation macro definitions .

• Histogram equalization

After equalization, the contrast of the image becomes higher , Become brighter

• Filter processing

There are two main categories ： Linear filtering and nonlinear filtering .OpenCV There are these filtering functions in , It is very convenient to use , Now briefly introduce how to use it .

Linear filtering ： Box filtering boxFilter、 Mean filtering Blur、 Gauss filtering GaussianBlur

Nonlinear filtering ： median filtering 、 Bilateral filtering

median filtering In removing impulse noise 、 Salt and pepper noise while preserving the details of the image （ There will be no blurring of edges ）

The most important feature of bilateral filtering is edge preservation

• edge detection （canny、sobel、laplacian）

The general steps of edge detection ：

1. wave filtering —— Eliminate noise
2. enhance —— Make the boundary more obvious
3. testing —— Pick the edge

Canny Algorithm

Canny Edge detection algorithm is praised as the best edge detection algorithm by many people .

Sobel Algorithm

Laplacian Algorithm

• Hough transform

How to quickly recognize circles and lines in images ？ A very effective method is the Hough transform , It is one of the basic algorithms for recognizing various geometric shapes in images .

Hoff line transformation

Hough line transform is a method of finding straight lines in an image .OpenCV Three Hough line transformations are supported in , Namely Standard Hough transform 、 Multiscale Hough line transform 、 Cumulative probability Hough line transformation .

stay OpenCV You can call functions in HoughLines To call Standard Hough line transform and multi-scale Hough line transform .HoughLinesP The function is used to call the cumulative probability Hough line transformation .

We all know Cartesian coordinates , An equation representing a straight line on a two-dimensional coordinate axis y = a*x + b, If we want to find a straight line, we have to find out a and b Value . If expressed in polar coordinates, it is

theta Is the angle between the straight line and the horizontal line , and rho It's the radius of the circle （ It can also be understood as the distance from the origin to the straight line ）, similarly , These two parameters are also important parameters to characterize a straight line , Make sure they're both , It determines a straight line . As shown in the figure below .

• Remapping and affine transformation

Remap It is the process of placing pixels at a certain position in one image to a specified position in another image .

stay OpenCV in , It's using remap Function to realize remapping .

Affine transformation

Affine transformation refers to a linear transformation of a vector space followed by a translation , The process of transforming into another vector space .

After affine transformation of the image , It has the following characteristics ：
The relative position relationship between two-dimensional graphics remains unchanged , Parallel lines are still parallel lines , And the position order of the points on the line remains unchanged .

An arbitrary affine transformation can be expressed as multiplying by a matrix （ linear transformation ） Then add a vector （ translation ） In the form of .

Three common forms ：

• rotate ,rotation（ linear transformation ）
• translation ,translation（ Vector plus ）
• The zoom ,scale（ linear transformation ）

Affine transformation is essentially a 2* 3 Matrix M Multiply by each coordinate of the original graph , Get the coordinates of the corresponding points of the target graph .2*3 matrix M Medium 2 Representing the coordinates of the target point x And y,3 The third dimension in is the translation component . So what we need to do is find the matrix M,OpenCV Provide getAffineTransform Find the affine transformation , getRotationMatrix2D To get the rotation matrix .

Here is a brief talk about how affine transformation is done .

Now there are two images （ Here's the picture ）, Image 2 is obtained from image 1 through radioactive changes . That's the question , How can we mine the mapping relationship between two images from these two image information ？

It's simple , Just take out three points in the image （1,2,3）, Image 2 also takes out the corresponding three points （1,2,3）, Then we can find the mapping relationship between the two graphs ！

OpenCV Affine transformation is realized by the combination of two functions ：

• Use warpAffine To implement simple remapping
• Use getRotationMatrix2D To get the rotation matrix

• Image restoration technology

OpenCV That's the use of inpaint() This function is used to realize the repair function