Image Processing Laboratory was founded in 1978. Professor Valery Pyatkin is the head of the laboratory since its foundation.
(E-mail: firstname.lastname@example.org ).
- - methods of automated aerospace image processing for the applied remote sensing research;
- - statistical methods of image processing;
- - computing tomography in the image processing problems;
- - Web and GIS technologies;
The automation of aerospace image processing for the applied remote sensing is one of the main directions of scientific research. Theoretical research included development and analysis of hardware-and-software concept of the aerospace image processing system, formulation and solution of a number of the urgent problems of processing and analysis of aerospace images and cartographic data. The hardware-and-software concept of the remote sensing data processing center was proposed. Many ideas of this concept were implemented in the working Center of Geoinformation Processing (CGIP) created in 1978-1984 on the base of Computing Center SB RAS. The software implemented (application packages) had been used for several years for processing of space data received from different Earth satellites and aero images to solve different problems of geophysics, geology, cartography, agriculture, and forestry. CGIP services were used by more than 30 organizations of some ministries and the departments to solve different problems of the visual data processing.
Remote sensing has become customary in investigating of Earth from space. We can enumerate more than a hundred of potential applications of remote sensing methods to study and inventory of Earth natural resources. Space information is particularly important for Earth geological structure study and mineral exploration activity. It causes the urgency of development of the methods of the automated geoindication features detection (linear and circular structures) in the aerospace images for complex study of the geological formations. The territory zoning based on lineament and circular structure characteristics carried out with the automated methods and the use of a priori information helps to estimate the availability of certain regions during exploratory works. This information can be used in the seismic mapping as it was shown by preliminary results of automated processing of the Spitak earthquake space images carried out in 1988-1989 in CGIP. The problem of circular structures detection has its own value in the study of astroblemes in Earth defense against collisions with large space bodies (like meteorites, comets). Existing methods of geoindicative features detection are not enough reliable and complete and often are subjective. It requires the creation of the new methods with use of mathematical apparatus and automated techniques. Two new approaches to the problem of automated lineament and circular structures detection, tomographic and statistical ones, were proposed. The base of the tomographic approach is the criterion of the projection optimality (informativity) from computational tomography. The mathematical basis of the approach is presented in the paper "Tomographic approach to lineament extraction in aerospace images" (А.С. Алексеев, И.Г. Казанцев, В.П. Пяткин. Томографический подход к выделению линеаментов на аэрокосмических изображениях // Issledovanie Zemli iz Kosmosa, 1988, N 5, pp. 99-103; in Russian). The base of the statistical approach is the use of nonparametric statistic criterion. This approach was experimentally tested in the processing of the aerospace images of the Eastern part of the Siberian platform (Yakutsk kimberlite province). The work was carried out together with "Yakutskgeologiya" geologists with the purpose of zoning of this territory and estimation of its availability during exploratory works. The work is described in the paper "About One Statistical Approach to the Problem of Automated Lineament Detection on Aerospace Images" (Г.И. Салов, В.П. Пяткин. Об одном статистическом подходе к задаче автоматизированного выделения линейных элементов на аэрокосмических снимках // Доклады АН СССР, 1988, т. 299, № 1 с. 76-79, in Russian). The applied investigations also included the development and analysis of a number of image processing algorithms and their software implementation in applied packages. The algorithm of lineament analysis is used in ice monitoring of Polar Region in State Research Center "Planet". The observation of Earth ice cover is one of traditional problems in gidrometeorology, climatology, and environment monitoring in general. On-line information about space distribution, drifting, type, age, and concentration of sea ice and icebergs is necessary for secure navigation, fishing, and oil and gas production in Polar Regions, as well as for ice forecast. From the climatology point of view, very important is accumulation and analysis of data about various characteristics of ice cover for many years. This data (boundaries of sea ice and shelf ice of Antarctica and Greenland, dynamics of spalling and destruction of icebergs) are indicators of regional and global climate changes.
Since 1997 Laboratory has started the works on the creation of window user interface in Microsoft Visual C++ (the first version was implemented in 1998). This product allowed creating user friendly programming environment for the development of new image processing modules and new applied remote sensing technologies. New image processing modules implementing nonparametric criteria for circular and linear structures detection were introduced in addition to the old ones. For the years 1999-2010, functions of geocoding and mapping of space images to the map by reference points on a picture and a map have been introduced to the system. Efficiency of Earth remote sensing highly depends on the methods of thematic processing of satellite date. The central thematic processing module of the system is recognition module which includes algorithms for supervised and unsupervised classification of multispectral satellite data. The system also includes modules for multidimensional data clusterization (unsupervised classification) based on K-means algorithm and multidimensional histogram analysis, and modules for supervised classification (with training) based on the use of Byes theory of maximal likelihood. Recently the modules for correlation-extremal analysis of multitemporal multispectral space images have been developed and included into software complex which are intended for the detection of space moving of observed objects: ice fields, water mass, cloud formations.
The theme “Image processing based on vertical data representation” has been developed in the last years. The approach of smooth approximation of continuous line from its discrete set of points has been developed with the help of geometrical constructions based on “field marking” in the metrical space of six-neighborhood. This approach is based on vertical data processing technique providing the transition from discrete to continuous line representation. This approach forms the basis for vectorization algorithms of raster aerospace images widely applied in GIS for cartographic purposes. The method of smooth approximation of continuous surface by discrete set of its representative points has been developed. The urgency of this problem follows from the well known in geoinformatics and cartography problem of building of digital terrain model by given relief elevation matrix.
The works on “Image processing based on vertical data representation” theme caused the development of a new direction in the Laboratory, parallel and distributive computations in aerospace i mage processing (since 2000). New parallel algorithms of image processing based on vertical data processing have been designed and implemented on multicomputers. Parallel versions of algorithms for linear and circular structures detection on aerospace images have been implemented; these algorithms are based on nonparametric statistical criteria developed in the Laboratory (see above). Portable parallel image processing library PLVIP based on vertical processing has been implemented on MBC-1000/M and RM600-E30 multiprocessor systems. Program environment for PLVIP library, UNIX framework of parallel image processing, has been developed which simplifies the remote development of parallel programs.
In 1999-2010, the application package for automation of the process of forest area decoding on aerospace images has been developed. Algorithm package for texture and cluster analysis in Microsoft Visual C++ object-oriented programming system has been prepared for its embedding into Laboratory image processing system. User friendly multi-document interface has been developed for the package for image texture features study. The complex is directed to automated determination of forest age and other valuation characteristics by aero images.
The latest research in computing tomography relate mainly to the few-view tomography problems where it’s necessary to apply iteration algorithms using maximum of a priory information about the object under study. One of the most developed restoration approach for parallel geometry of scanning is Gerchberg-Papoulis algorithm using iterations in image and its Fourier transform spaces by turns. New modification of the central slice theorem has been proved which connects Fourier transform of fan-beam projections with Fourier transforms of the object. Base on this theorem, the iterative Gerchberg-Papoulis algorithm for fan-beam tomography has been developed. Practical aspect of this work is that the algorithm suggested can be used for the experimental data processing in physical few-view tomography problems as the significant part of these tomography researches deal with data obtained from fan-bean scanning geometry.Development and perspectives of new information technologies for aerospace image processing are reflected in the following research projects :
- State Support for Integration of Higher Education and Fundamental Science federal program (together with Altay State University, 1997-2000; together with SKTB Nauka of Krasnoyarsk scientific center SB RAS, 2002-2006).
- Analysis and modeling of extreme hydrological phenomena for the purposes of developing measures for unfavorable consequences prevention and damage minimization on Siberia water objects (RAS integration project No. 13-15, 2003-2005).
- Aerospace radiolocation and radiometry of the Earth surface (SB RAS integration project No. 13, 2003-2005).
- Development of the environment monitoring information technologies on the base of the distributed computational system for remote sensing data processing, Internet technologies, and multiprocessor computers (RFBR project No. 05-07-90957-в, 2005-2006).
- Design and implementation of the software for distributed parallel Earth remote sensing data processing with the use of Web technologies (RFBR project No. 07-07-00085-a, 2007-2009).
- Design and development of the multifunctional system of Earth remote sensing data processing based on modern image processing methods and algorithms, distributed parallel technologies, and Web resources (RFBR project No. 10-07-00131-a, 2010-2012).
- Research and development of algorithms and software for Earth remote sensing data processing based on modern hardware and software platforms with distributedmparallel and cloud computing (RFBR project No. 13-07-00068-a, 2013-2015).
The importance of the digital cartography investigations is emphasized by the fact that maps are commonly used in the data models presentation in the applied remote sensing. Together with Special Computer Engineering Department of SB RAS, the novel interactive system for the cartographic data control and edition was developed, paper "Interactive Processing System of Digital Information about Terrain" (А.С.Алексеев, П.А.Калантаев, В.П. Пяткин. Система диалоговой обработки цифровой информации о местности. Препринт №795, Новосибирск, ВЦ СО РАН, 1988, in Russian). This system doesn’t have analogues in our country. This complex passed the state test successfully and was introduced into practice in 1990. The efficiency of the automated cartographic system, which is the main body of any geoinfomation system, depends on its information support, i.e. its database organization. The principles of the problem-oriented cartographic database implementation were investigated, paper "Database for control and edition system of terrain digital information" (П.А. Калантаев, В.П. Пяткин и др. База данных системы контроля и редактирования цифровой информации о местности // - Исследование Земли из космоса, №3, 1986, in Russian). The investigations in digital cartography, GIS technologies, and Web cartography were carried out in the context of some projects and commercial works. We should mention here RFBR project No. 96-07-89489 "Design and creation of software tools for West-Siberian cartographic database analysis with providing information services for fundamental research" (1996-1998). In this project, Digital Cartography System was created in 1996, paper "Digital cartographic software of applied remote sensing" (А.С.Алексеев, П.А.Калантаев, В.П.Пяткин и др. Цифровое картографическое обеспечение прикладных дистанционных исследований // Труды Международной Конференции: ИНТЕРКАРТО-3, Новосибирск, 1997 г., с. 239-250, in Russian). With the help of this system, interactive digital map of Akademgorodok was created (http://loi.sscc.ru/lab/WEBLAB/Akadem/main_akaden.html.). In 1997, web page of Image Processing Laboratory of ICM&MG appeared in the Internet (second prize among all ICM&MG departments in 2001). New edition of Laboratory Web-site developed in 2016.