5-1 Beautification Effect Chapter Introduction

Image filtering, noise removal

With these current effects, we can directly see these practical cases in beauty algorithms or beauty cameras.

RGB three color histogram, it has a horizontal two coordinates. Where the horizontal axis describes the color level of 0-255, a total of 256 species. Because of the color depth of the 2^8=256,8 bit. The ordinate is the probability that each color level appears on the entire picture. Note that our ordinate is described by probability, so its range is between 0-1. This is the concept of a histogram.

What is histogram equalization? If a picture on its histogram is not evenly distributed, but concentrated in a place, this time can use the histogram equalization, so that the original aggregation of the histogram as far as possible discretization, this process is called histogram equalization.

There are two kinds of histogram of grayscale picture and histogram of color picture.

It can be seen that after the image histogram equalization, the brightness of the picture, clarity and so on people's intuitive feeling will be much better.

The next effect is to beautify the effect. If the original image for some reason caused the fouling, then this way can be used to achieve image repair. In addition, when the image is patched, we also have to prepare a picture similar to the mask, which is the black picture in the lower left corner. This picture needs to be as descriptive as possible about the area to be patched, such as the white cross-intersecting line that defaced the process.

The next effect is brightness enhancement. This can be said to be a most simple sense of beautification effect. For example, add the current grayscale value to a fixed grayscale level, or multiply the current grayscale value by a fixed magnification factor of 1.1 or 1.2. This effect is not differentiated, increasing the brightness of the entire picture. So it is not a true sense of whitening.

Let's take a look at a really abrasive skin whitening effect. We are using a double-sided filter to achieve this.

Then look at a set of image filtering related landscaping effect. We will show you a variety of effects such as Gaussian filtering, median filtering, and mean filtering. Where Gaussian filtering we will use the form of the OpenCV API, median filtering we will use the form of source code for everyone to achieve.

5-1 Beautification Effect Chapter Introduction