Content introduction chapter development of a complex image processing method

CHAPTER 1. DEVELOPMENT OF AN INTEGRATED IMAGE PROCESSING METHOD
1.1 The concept of image processing. Choice of software implementation environment
The input data for processing are images. The output can be not only the processed image, but also some of its characteristics. Images vary from binary, that is, those in which there are only two colors (usually black and white), to 32-bit images. Let's dwell on the most common 24-bit images of the .bmp format (from the English bitmap). This format is intended for storing raster images, that is, images that are a grid of pixels or dots of colors. In addition to it, there are many other formats, such as .jpeg (.jpg), .png, .tiff and others. They differ, as a rule, in the way the image is encoded. For example, the .jpeg format contains compressed data that was obtained using the JPEG algorithm of the same name, which is a lossy compression algorithm.
There are quite a lot of tools for working with images on a computer using high-level programming languages. The most common are: Open GL (Open Graphics Library) is an open graphics library, DirectX is a set of functions designed to solve problems related to video programming under Microsoft Windows, as well as directly Microsoft Visual Studio in which the application will be implemented. Microsoft Visual Studio is a line of Microsoft products that includes an integrated software development environment and a number of other tools. These products allow you to develop both console applications and applications with a graphical interface, including those with support for Windows Forms technology, as well as websites, web applications, etc. This programming environment
· Simple implementation of common tasks and individual approach;
Easily connect various libraries
· Fast creation of high-quality code;
· Multi-monitor support function;
· Ability to implement ideas and solutions for a wide range of platforms, including Windows OS, Windows servers, etc.
1.2 Choice of image processing and segmentation methods
The images for which filtering methods will be used will be images of banknotes. Figure 1 and Figure 2 show images of Russian banknotes of 500 and 1000 rubles, respectively, in black and white and in the “.bmp” format.
Drawing1 - original imagebanknotes of 500 rubles
Figure 2- original image banknotes of 1000 rubles
Analyzing the two image data, it can be noted that in the first image, the text information is displayed quite clearly, while the second image has significant noise.
After analyzing the image data, you can begin to define filters for segmenting text information images of both banknotes. Adobe Phoroshop CC 2015 is used as a graphical editor in which filtering methods will be determined.
Adobe Photoshop is a multifunctional graphics editor developed and distributed by Adobe Systems. It mainly works with raster images, but it has some vector tools. The product is the market leader in commercial bitmap editing tools and is Adobe's best known product.
The selection of methods for processing and segmenting the image shown in Figure 1 in the Adobe Photoshop graphics editor revealed that in order to increase the information content of the image, it is advisable to use threshold filtering, which makes it possible to determine the contours of the text information of the banknote. 
After applying the methods presented above, it was decided to use morphological filtering using image dilation and erosion. Dilatation leads to expansion, "growth" of image pixels, erosion leads to thinning, "weathering" of pixels. By combining these two operations in turn and applying them to the image, you can remove objects of a certain shape and size, as well as eliminate gaps and holes that do not exceed a given size.
The median filter is a type of digital filter that is widely used in digital signal and image processing to reduce noise.
Sample values ​​inside the filter window are sorted in ascending (descending) order; and the value in the middle of the ordered list goes to the output of the filter. In the case of an even number of samples in the window, the output value of the filter is equal to the average of the two samples in the middle of the ordered list. The window moves along the filtered signal and the calculations are repeated.
It can also be noted that median filtering is used mainly for images damaged by impulse noise, designed for an efficient signal processing procedure.
Thus, median filtering replaces the values ​​of the samples in the center of the aperture with the median value of the original samples inside the filter aperture. In practice, the filter aperture, to simplify data processing algorithms, is usually set with an odd number of samples.
Brightness transformations of digital images are often called histogram transformations, since, firstly, the image histogram changes, and secondly, the form of the transformation function (transformation parameters) is often determined adaptively, based on the previously collected histogram of the original image.
The histogram characterizes the frequency of occurrence of pixels of the same brightness in the image.
Such linear brightness conversions are also called photographic , since in traditional photography they can be set by changing the shutter speed and aperture characteristics of the lens. Other brightness display functions are possible.
The simplest method of image preparation is binarization . This conversion consists in turning the image into two-tone black and white. The main parameter of such a transformation is the threshold - the value that will be the criterion for checking the intensity of the image point. 
Binary images in the sense of subsets of pixels ("masks") are often used in digital imaging. To study the form and structure of some sets of objects of the same type, binary rasters are used in mathematical morphology.
The threshold by which pixels are divided into "black" and "white" is determined by the histogram of the image.
In this case, morphological filters mean the following algorithms:
1. Expansion (Erosion)
2. Compression (Dilation).
3. Finding the extended contour (Eroded Contour).
4. Finding a compressed contour (Dilated Contour).
All morphological filters are based on two operations - dilation and erosion . They are defined as follows. Let we have two arbitrary sets of points in a discrete two-dimensional space: and . We define functions on these sets and call them objects A and B.