Shulou(Shulou.com)06/01 Report--

This article mainly introduces C++ OpenCV how to achieve blurred image, has a certain reference value, interested friends can refer to, I hope you can learn a lot after reading this article, the following let the editor take you to understand it.

Blurred image

Image blur is the most commonly used and simple operation in image processing. One of the reasons for using this operation is to lower the voice during image preprocessing.

Convolution

It is superimposed. The important physical meaning of convolution is the weighted superposition of one function (such as unit response) on another function (such as input signal).

It is popular to say:

At each position of the input signal, a unit response is superimposed and the output signal is obtained.

This is why the unit response is so important.

Application of convolution

Convolution is carried out with a template and an image. For a point on the image, the origin of the template coincides with the point, then the point on the template is multiplied by the corresponding point on the image, and then the product of each point is added to get the convolution value of that point. This is done for every point on the image. Because most templates are symmetrical, the templates do not rotate. Convolution is an integral operation that is used to calculate the area of overlap between two curves. It can be regarded as weighted summation, which can be used to eliminate noise and feature enhancement.

Replace the pixel value of a point with the weighted average of the pixel value of the point around it.

Convolution is a kind of linear operation. The common mask operation in image processing is convolution, which is widely used in image filtering.

One of the most important cases of convolution relations is the convolution theorem in signal and linear systems or digital signal processing. By using this theorem, the convolution operation in time domain or space domain can be equivalent to the multiplication operation in frequency domain, so that fast algorithms such as FFT can be used to realize effective calculation and save operation cost.

Use pictures to illustrate

Four kinds of fuzzy algorithms are included in OpenCV:

Blur (mean ambiguity)

GaussianBlur (Gaussian blur)

MedianBlur (median blur)

BilateralFilter (bilateral filtering)

New project

Create a new project opencv-0009, configure properties (VS2017 configure OpenCV common properties), and then write # include and main methods in the source file. And load the picture that we use to show it.

Mean ambiguity

Add a function blur ()

Then run the effect.

Gaussian blur

We add a method GaussianBlur ()

We change the original mean fuzzy size to 9: 9, Gaussian blur also uses 9: 9, and then show the effect.

The last edge is Gaussian blur, the picture is not very clear, in fact, if you look carefully to see the contrast, the Gaussian blur outline on the right can be more obvious.

Median fuzzy

The median is the filter noise of statistical sorting.

The median has a good inhibitory effect on salt and pepper noise.

Median fuzzy API

MedianBlur (Mat src,Mat dst,ksize)

Ksize size must be greater than 1 and must be odd

We add a method medianBlur ()

We set ksize to 9, and then we can see the effect.

We can see that on the far right is our median fuzzy, and the overall blur is relatively average.

Bilateral ambiguity

The characteristics of bilateral ambiguity:

Mean ambiguity can not overcome the loss of edge pixel information because mean filtering is based on average weight.

Gaussian blur part overcomes this defect, but it can not be avoided completely, because the difference of pixel value is not taken into account.

Gaussian bilateral blur is an edge-preserving filtering method, which avoids the loss of edge information and keeps the image outline unchanged.

Bilateral fuzzy API

BilateralFilter (src,dst,d=15,150,3)

15 color-the calculated radius and the number of images within the radius will be included in the calculation. If-1 is provided, the value will be taken according to the parameters of sigma space.

150--sigma color, which determines how many pixels within the difference will be calculated.

3--sigma space, if the value of d is greater than 0, it is declared invalid, otherwise the value of d is calculated according to it.

We add a method bilateralFilter ()

Because the color difference of our original image is not big, so we adjust the value a little bit, and then we'll see the effect.

You can see that the basic appearance has not changed, but the place of the red frame has been blurred, and the outline is still very obvious.

Zoom in a little bit and take a look.

Thank you for reading this article carefully. I hope the article "how to achieve blurred images on C++ OpenCV" shared by the editor will be helpful to everyone. At the same time, I also hope that you will support and follow the industry information channel. More related knowledge is waiting for you to learn!

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.