DonNTU   Masters' portal

Abstract

Contents

Introduction

With the development of computer technology has become possible to solve problems that were previously considered unfeasible on personal computers. Creating devices that perform the function of object recognition in the video stream, in most cases, will replace man – a specialized machine. It is worth noting that the quality of a person's work depends on many factors (qualification, experience, fatigue, interest). At the same time configured properly and the system will work, always providing the same quality of work. Another advantage of the replacement of human machine – the incomparable advantage of automated systems in performance.

Today, in every city, there is a center of traffic jams, which are the peak hours reach hundreds of kilometers of ground public transport significantly reduces the possibility of his carriage and becomes unattractive for passengers, snaring emergency and official vehicles, increasing the number of small accidents, traffic violations, by standing and working on the "bad" idling vehicles increased levels of air pollution.

The main reason is the transport problem in the original planning of the city, excluding the possibility of wide use of private cars. Disposition of the streets, for example, in Moscow, radial-ring and theoretically has an important advantage over rectangular (such as in New York), since due to the diagonal route that minimizes time to move around the city. But in practice, this configuration results in the roads that the city center, connecting the radial lines, is often transformed into one continuous tube.

1 Theme urgency

Manual adjustment or the use of traffic lights the problem of organizing effective movement in the big city is not solved. Therefore, the automated control of the movement of road transport – it is necessary and urgent task, as one of the modern methods of detection of problematic situations on the application of computer technology is the contour analysis.

Methods of contour analysis in the present are widely used in radio systems, communication systems, to target aircraft in the process of monitoring the movement of vehicles to detect the number of the car, the speed control, or traffic violations. Contour – the most informative structural element object. However, high-quality selection circuit – it is one of the most difficult problems processing visual information. Known methods for solving contour analysis using library OpenCV, but this package is designed to solve the problem of counting objects with visually prominent point on the image. And the contours often have disadvantages such as breaks, availability of extra lines that do not match the object under study

Therefore, the creation and implementation of software for the control of road transport will enable to control the capacity of roads and education to solve the problem of traffic jams.The system will communicate with traffic lights, when you turn on a red traffic light will take a picture of the road will be analyzed and the number of cars in the picture.

2 Goal and tasks of the research

The purpose of this work – the development of an algorithm for contouring software control system of automobile traffic.

The object of development – software that provides control of car traffic by a contour image analysis condition of the carriageway, is obtained from surveillance cameras. Subject design – modern design tools and software: MS Visio 2013, C # development environment MS Visual Studio 2012.

To achieve this goal it is necessary to solve the following problems:

1) review of existing systems;

2) to consider the basic principles of the contour analysis algorithms to analyze the evaluation of their performance;

3) selection of an optimal algorithm for the implementation of the system;

4) justification of the programming language and development environment;

5) the design of the system architecture;

6) to develop the software system.

3 Definitions and properties of loop

3.1 The concept of path

Contour analysis allows to describe, store, compare, and search for objects presented in the form of its external contours – contours. The circuit – is the border of the object, a set of dots (pixels) that separate the subject from the background. It is assumed that the circuit contains all the necessary information about the shape of the object. The interior points of the object not taken into account. This limits the range of applicability of contour analysis algorithms, but considering only the contours allows to switch from a two-dimensional image space – the space of contours, and thus reduce the computational and algorithmic complexity. Contour analysis allows to solve the basic problems of pattern recognition – moving, rotating and zooming facility.

The computer vision systems use multiple ways of encoding circuit – best known Freeman code, the two-dimensional coding, polygonal encoding. But all these methods are not used in coding the contour analysis. Instead, contour analysis circuit coding sequence consisting of complex numbers. On the fixed point of the circuit, which is called the starting point. Then, the outline costs (say – clockwise see. Figure 1 - Coding circuit), and each displacement vector is written by a complex number a + ib. Where a – displacement of the point on the axis X, and b – axis offset Y. The offset is taken with respect to the previous point.

Each vector path will be called basic vector (EV). But the very sequence of complex numbers – vector circuit (VC). Vector–circuits will be denoted by large Greek letters, and their elementary vector – small Greek letters. Thus, the vector–contour T of length k can be defined as:

Кодирование контура

Figure 1 – Coding circuit (animation: 7 frames, 15 cycles of repetition, 74 kilobytes)

Integrated coding close to the two–dimensional coding, where the contour is defined as the set of ER, represented by their two–dimensional coordinates. But the difference is that the operation of the scalar product for vectors and complex numbers – different. This circumstance also offers the advantage of the method of contour analysis.

3.2 Properties of circuits

Consider the properties of circuits:

1. The amount of the elementary vectors of the closed loop is zero. It is trivial – as elementary vectors lead to the start point, then their sum is equal to the zero vector.

2. Circuit vector does not depend on the parallel transfer of the original image. Since the circuit is encoded with respect to the starting point, this encoding method is invariant initial contour shift.

3. Rotate the image at a certain angle of rotation of each EV equivalent circuit at the same angle.

4. Changing the starting point leads to the cyclic shift of the vector loop. Since the elementary vector encoded relative to the previous point, it is clear that a change in the starting point of a sequence of elementary vector will be the same, but the first basic vector is the one that begins at the starting point.

5. Changing the scale of the original image can be viewed as a vector multiplication, each elementary circuit scale factor.

Conclusion

As a result of this work were reviewed and analyzed the following tasks:

1) the basic principles of the contour analysis algorithms that allow to develop an algorithm;

2) elected an optimal algorithm for writing software that is as accurate as possible, and in a short time gives the result;

3) a review of existing systems;

4) analyze the performance evaluation algorithms contour analysis, it was found that the modern development of computer technology, contour analysis may be used as a real-time algorithm;

5) the study of the Fourier descriptors found out that this type of descriptors has all the requirements for classification: invariance under rotation, scaling and parallel shift. In this connection, the ghost contour to the object of the Fourier amplitude spectrum helps classify the object in real time.

Qualitative synthesis of digital systems, and Moore automata in particular, is one of the areas of logic design, and is not only theoretical research but also of practical interest. Development of optimal digital devices opens the way for more efficient use of capacity basis the implementation of compaction of projects, reduce material costs.

This master's work is not completed yet. Final completion: December 2015. The full text of the work and materials on the topic can be obtained from the author or his head after this date.

References

  1. Фурман Я. А. Введение в контурный анализ. - М.: ФИЗМАТЛИТ, 2003. –561 с.
  2. Детектирования автомобильных номеров на основе контурного анализа [Электронный ресурс] – Режим доступа: http://mp3car.ru/blog/ComputerVision/15.html
  3. Цифровая обработка изображений/ Интернет-ресурс. - Режим доступа: www / URL: http://rudocs.exdat.com/docs/index-87395.html - за глав. с экрана.
  4. Гонсалес Р. Цифровая обработка изображений в среде C # / Р. Гонсалес, Р. Вудс, С. Еддинс - М.: Техносфера, 2006. - 616с.
  5. Грибков И.В. Некоторые вопросы количественной оценки продуктивности детекторов границ/ И.В. Грибков, А.В. Захаров, П.П. Кольцов и др. - М.: «Программные продукты и системы» № 4, 2011. - 234с.
  6. Леухин А.Н., доц. Парсаев Н.В., асп.Рахманов Х.Э. Алгоритм локализации изображения регистрационного номерного знака автомобиля с использованием методов контурного анализа [Электронный ресурс] – Режим доступа: http://hpc.marsu.ru/Content/articles/7.pdf
  7. Алиев М.В., Панеш А.Х., Каспарьян М.С. Выделение контуров на малоконтрастных и размытых изображениях с помощью фрактальной фильтрации [Электронный ресурс] – Режим доступа: http://cyberleninka.ru/article/n/vydelenie-konturov-na-malokontrastnyh-i-razmytyh-izobrazheniyah-s-pomoschyu-fraktalnoy-filtratsii
  8. Костюкова Н.С., Коршун А.Н., Корольчук Е.А. Выделение контуров объектов в растровых изображениях http://ea.donntu.ru:8080/jspui/bitstream/123456789/15503/1/163-170.pdf
  9. Потапов А.А., Гуляев Ю.В., Никитов С.А., Пахомов А.А., Герман В.А. Новейшие методы обработки изображений - М.: ФИЗМАТЛИТ, 2008. –489 с.
  10. Гонсалес Р., Вудс Р. Цифровая обработка изображений Издательство "Мир" Москва 1982 - 312 с.
  11. Анисимов Б.В., Курганов В.Д., Злобин В.Н. Распознавание и цифровая обработка изображений - М.: ФИЗМАТЛИТ, 2003. –561 с.
  12. Хуанг Т. Обработка изображений и цифровая фильтрация Перевод с анг. к.т.н. Е.З. Сорокин, В.А.Хлебородова Издательство "Мир" Москва 1979 - 412 с.
  13. Форсайт Д. Компьютерное зрение. Современный подход. :Перевод с анг. - М.: Издательский дом "Вильямс", 2004. - 928 с.
  14. OpenCV шаг за шагом. Нахождение контуров и операции с ними [Электронный ресурс]. – Режим доступа: www/ URL: http://robocraft.ru/blog/compute revision/640.html