CC BY
112. На основе теоретических исследований, а также результатов численного моделирования разработана методика расчета с использованием основных положений СП 24.13330.2011 [3] для определения несущей способности усовершенствованной анкерной конструкции при воздействии на нее выдергивающей нагрузки.
Список литературы
1. Левачев C.H., Халецкий B.C. Анкерные и якорные устройства в гидротехническом строительстве // Вестник МГСУ, 2011. № 5. С. 58-68.
2. Осмачкин А. Применение грунтовых анкеров для укрепления строительных конструкций // Инженерная защита, 2014. № 5. С. 68-75.
3. СП 24.13330.2011 Свайные фундаменты. Актуализированная редакция. СНиП 2.02.03-85. М., 2011. 83 с.
GENERAL APPROACH TO RECOGNITION OF OBJECTS ON MICRO IMAGES Mahali M.1, Iskakov K.T.2
'Mahali Maksat - Master;
2Iskakov Kazizat Takuadinovich -PhD, m.s., Professor,
DEPARTMENT OF SCIENCE ENGINEERING, L.N. GUMILYOVEURASIAN NATIONAL UNIVERSITY, ASTANA, REPUBLIC OF KAZAKHSTAN
Abstract: the article describes the main approaches associated with the processing of microscopic images. The processing steps outlined in the article are complex algorithmic procedures that are to be automated.
Keywords: microscopic images, analysis, recognition, segmentation algorithms.
Pattern recognition as one of the directions of computer graphics allows you to solve a wide range of tasks and includes the processing and analysis of data. The data may include aerospace images, signals in technical systems, medical images and many other data, the scope of which is expanding, and the processing and analysis procedures are becoming increasingly complex.
Pattern recognition systems belong to the class of intelligent systems, since provide support in decision making and are based on three basic principles:
- the principle of comparison with the standard;
- the principle of clustering;
- principle of commonality of properties.
The role of analysis and processing of microscopic images in medicine has increased most significantly. Diagnosis of diseases, and subsequently, and medical procedures are based on data obtained by methods of medical imaging. Usually, images obtained by microscopic examination have some distortions in the quality of pictures of histological objects, since they are characterized by noise, sharpness and fuzzy boundaries of objects. Addressing issues related to improving image quality requires high professionalism in their adjustment to ensure the accuracy and reliability of research results. And so the diversity of research in the field of intellectual analysis of microscopic images indicates the scale of the tasks and the high significance of the results obtained.
The largest project is the development of a cognitive system by IBM Watson, which today is the largest medical knowledge base. However, many medical and clinical
diagnostic centers need not only expert systems, but also manual automation related to laboratory diagnostics. The leading groups in the world in cell imaging research are American, European and Chinese institutions. This is evidenced by the statistics of published articles in world ranking journals for the years 2015-2017.
One of the main methods for recognizing blood patterns is segmentation. Segmentation refers to the selection of individual areas or objects in an image. Selecting objects on medical images is a very difficult task. Segmentation involves building an image model, i.e. a formal description of the statistical properties and the regular structure of the areas that make up the image [1, 2]. Segmentation algorithms make it possible to identify areas of homogeneity, such areas whose structure is well described by any of the models built in the first stage.
In the analysis of the most valuable information on the image represent the nucleus and the cell contour. For their recognition, threshold segmentation is used, in which the halftone image is converted into a binary by selecting a threshold value. During binarization, the image becomes black and white, and its pixels have values of 0 or 1 [3].
For a binary image, the following relationship holds:
r(xy)= (0,r(x,y)>L n ,yJ ll,s(x,y) < L
where r (x,y) , s (x,y) - brightness levels of image pixels, L - threshold level of brightness. Consequently, in the images r (x,y) , the pixels whose values are equal to one belong to objects, and those that are equal to zero belong to the background. In image binarization, the brightness of each pixel is compared by a threshold value. When threshold image processing one of the important tasks is the correct choice of the threshold, since errors in its meaning lead to distortion of the boundaries of regions.
One of the effective methods for selecting the threshold is the Otsu method, based on the analysis of the histogram of the distribution of the image brightness values [4]. The algorithm allows to divide the pixels of the two classes (useful and background), calculating such a threshold so that the intraclass dispersion is minimal.
a? ( t) = o! ( t) a? ( t) + o t) a? ( t) ,
where cot - these are probabilities of two classes separated by a threshold t, a? -dispersion of these classes.
According to this method, minimizing dispersion within a class is equivalent to maximizing dispersion between classes:
a? = a2- a?(t) = o 1 (t)o2 (t)(t) - n2(t)] 2,
where cot - these are probabilities for classes, nt - medium arithmetic class [5].
When image segmentation is often used the method of the watershed, based on mathematical morphology. The essence of the method lies in the fact that the image is considered as a relief, in which the watershed lines are the boundaries dividing parts of the image into segments, and the watershed pools are the corresponding image areas. The disadvantages of this method include noise sensitivity and excessive segmentation, which leads to too much selection of objects, which leads to low efficiency of image processing.
This problem is well handled by the marker watershed, which is one of the effective methods of image segmentation. The marker watershed algorithm consists of the following steps:
- calculate the image segmentation function, in which the black areas are objects;
- calculate foreground markers
- calculate background markers;
- modify the segmentation function so that its minimum is located only on the foreground and background markers;
- calculate the watershed transformation of the altered segmentation function.
Thus, the general approach to the recognition of micro images consists of determining whether the image in the image is pathological, and if so, it is necessary to proceed to
computer processing of the image. To identify the necessary objects in the image, segmentation algorithms and threshold segmentation are used, as a result of which the image is divided into two classes. The segmentation efficiency can be checked by the obtained histogram. Using marker algorithms, you can improve the representativeness of the image.
References
1. Therrien C.W., Quatieri T.F., Dudgeon D.E. Statistical model-based algorithms for image analysis. Proceeding of the IEEE. Vol. 74. Issue 4, 1986. Рp. 532-551. [Electronic resource]. URL: https://ieeexplore.ieee.org/document/1457772/ (date of acces: 14.12.2018).
2. Wirjadi O. Models and algorithms for image-based analysis of microstructures. [Electronic resource]. URL: https://d-nb.info/993579078/34/ (date of acces: 14.12.2018).
3. Panova I.A., Lysak O.Yu. Method of medical image binarization. [Electronic resource]. URL: https://storage.tusur.ru/files/8617.pdf/ (date of acces: 14.12.2018).
4. Otsu N.A. Threshold selection method from gray-level histograms. IEEE Transactions on systems, Man and Cybernetics. Vol.9, Issue 1. 1979. Pp 62-66.
5. Wikipedia. [Electronic resource]. URL: ttps://wiki2.org/ru/%D0%9C%D0%B5%D1%82%D0%BE%D0%B4_%D0%9E%D1%8 6%D1%83/ (date of acces: 14.12.2018).
ВЛИЯНИЕ МНОЖЕСТВЕННОЙ ОБРАБОТКИ ИЗОБРАЖЕНИЙ ПРИ СШИВАНИИ КАДРОВ В ФОТОПЛАН
1 2 3
Сидоркин И.И. , Маликова М.О. , Цуканов М.В.
1Сидоркин Иван Игоревич - магистрант, кафедра информационных систем, Институт приборостроения, автоматизации и информационных технологий, Орловский государственный университет им. И.С. Тургенева, младший научный сотрудник, Орловский филиал Федеральный исследовательский центр «Информатика и управление» Российская академия наук;
2Маликова Мария Олеговна - магистрант, кафедра информационных систем, Институт приборостроения, автоматизации и информационных технологий, Орловский государственный университет им. И.С. Тургенева;
3Цуканов Максим Владимирович - инженер-исследователь, Орловский филиал Федеральный исследовательский центр «Информатика и управление» Российская академия наук, г. Орёл
Сшивание кадров в фотоплан может быть использовано для ведения картографических и иных работ по составлению плана местности. Под фотопланом понимается набор совмещенных изображений. Процедура сшивания последовательности кадров в фотоплан имеет один существенный момент: сшивание осуществляется последовательным процессом склейки отдельного кадра с промежуточным результатом склейки, полученным на предыдущем этапе. Такой подход позволяет выполнять процедуру сшивания без особых сложностей, просто перебирая последовательность кадров, проверяя возможность склеивания их между собой, формирую промежуточный результат и далее уже осуществлять склейку с уже
