WIXLIB

F/G 17/2.1FOREIGN TECHNOLOGY DIV WRIGHT-PATTERSON AFB OHGEOMETRIC METHODS OF PROCESSING TELEVISION IMAGES OF THE CLOUD "-ETC(U)NOV 79 K S SHIFRIN, Y S FRIDMANNLUNCLASSIFEEDF FTO-ID(RT)T-1O7-T97AD-A083 482IEl/l/E//EEl/EIlllI.580L

PHOTOGRAPH THIS SHEETLEVELINVENTORYiFTD-rp((ZS)T- I107-"/ TDOCUMENT IDENTIFICATIONDISTRIBUMIIN STATEMENT A[Approved for public release;I Distribution Unlim ited4DISTRIBUTION STATEMENTACCESSION APR 29 1980DBYDISTRIBUTION /AVAILABILITY CODESAVAIL AND/OR SPECIALDISTDATE ACCESSIONEDDISTRIBUTION STAMP8034 117DATE RECEIVED IN DTICPHOTOGRAPH THIS SHEET AND RETURN TO DTIC-DDA-2DTIC,*,0*.*FORM OCT790ADOCUMENT PROCESSING SHEET

,---.SQf.--FTD-ID(RS)T-J 407-79FOREIGN TECHNOLOGY DIVISIONCQGEOMETRIC METHODS OF PROCESSING TELEVISION IMAGESOF THE CLOUD COVER OF THE EARTHbyK. S. Shifrin, Yu. S. FridmanApproved for public release;distribution unlimited. ,LE

FTD- ID(RS)T-l'407-79UNEDITED MACHINE TRANSLATIONFTD-ID(RS)T-1407-79MICROFICHE NR.6 November 1979IFTD-79-C-001465LCS171733348GEOMETRIC METHODS OF PROCESSING TELEVISION IMAGESOF THE CLOUD COVER OF THE EARTHBy:K. S. Shifrin, Yu. S. FridmanEnglish pages:Source:44Trudy Glavnoy Geofizicheskoy Observatorii,Leningrad, Nr. 235, 1970, pp. 24-42.USSRCountry of origin:This document is a machine translation.FTD/TQTRRequester:Approved for public release; distributionunlimited.THIS TRANSLATION IS A RENDITION OF THE ORIGI.NAL FOREIGN TEXT WITHOUT ANY ANALYTICAL OREDITORIAL COMMENT. STATEMENTS OR THEORIESADVOCATED OR IMPLIED ARE THOSE OF THE SOURCEANDDO NOT NECESSARILY REFLECT THE POSITIONOR OPINION OF THE FOREIGN TECHNOLOGY DIVISION.FT - ID(i1S)T-1407-79PREPARED BY:TRANSLATION DIVISIONFOREIGN TECHNOLOGY DIVISIONWP.AFB, OHIO.Date r pjo, -97

IAUU. S.6OARD ON GEOGRAPHIC NAMES TRANSLITERATION SYSTEMblockItalicTransliterationBlockA aAaA, aP pPpB6B, bC cCCsBaV,vT TTmG,gY yyyXxXXxKh,.:hqTs,tsCh,ci-.Sh,2hF -L ei *EYe,.'WSli, zhye;E, e*43a1,1i-1RaK K)7IAY,WuyLAiK, kL)R, rT,HLUAmt2q&L,l1'Y,YbM, mMHtTransliteratic i.H,SVItalic14, nD-o0o0, 0hiSn17nP, pi' R31061pE, e;0Yu,yuYa, ya*ye initially, after vowels, and after b, b; u, elsewnere.When written as L in Russian, transliterate as y5 or e.RUSSIAN AiND LXGLISH TRIGONOIETRIC FUNCTIONSRussianEnglishRussianEnglishRussianEngl gshcbthcthschcsch

DOC 79140700PAGE1Page 24.GEOMETRIC METHODS OF PROCESSING TELEVISION IMAGES OF THE CLOUD COVEROF THE BARTH.*§1. Introduction. Formulation of tAs problem.K. S. Shifrin, Yu.S. Fridman.Use of ISZ [artificial earth satellite] for meteorologicalobservations placed before applied mathematics a whole series of newproblems. one of such problems, about which will go the speech below,is the development of the mathematical method of treating the imagesof the cloud cover, obtained with the help of the television camera,situated on satellite. in spite of the large efforts/forces, spent onthe development of other methods of obtaining the information aboutatmosphere, the television photographs of the cloud cover and of theearth's surface are the up to now only form of the information,actually utilized in the meteorological practice of the series/row ofthe countries. This information is of large imterests sinace form andcloudiness, structure of the fields of cloudiness, rotation of the

DOC 79140700PAG2forms of cloudiness and their evolution can serve as basis for thediagnosis of macro-and mesoscale atmospheric processes.go are to acertain degree returned here to the idea of nephanalysis - thedetermination of a weather change in observations of cloudiness, butalready at other,much higher level: at the level of the use of these[2].meteorological satellitesIEach photograph of the cloud coverof informational units.if we take into account also an enormousnumber of photographs of cloudiness,ISZ, then itcontains a colossal quantitywill become itwhich continuously enter withis clear that here must be used themethods of machine working/treatment. According to the evaluation ofe.K.Fedorov,the discussion deals with billionsof informationdaily (1 ].ip to now in essence from this entire enormous material are usedseparate photographs for weather analysis in especially complicatedsituations, when usual synoptic analysis proves to be ambiguous. Thisis done visually, by the examination of separate photographs. Themethods of analysis are qualitative and suffer subjectivism.Althoughin a number of cases similar analysis is very useful, it is clear,however, that it cannot be comprehensive. It is necessary to studythe mass characteristics of cloud formation/educations, theirdynamics and statistics, to compare them with the ground-based data

DOC-79140700PAGE3and synoptic charts of area in question and period.Page 25.But as to fulfill a similar work with many thousands ofphotographs, which we do place?it is obvious, for this it is necessary at first to learn tosomehow preliminarily treat photographs, extracting of them basis andleaving aside secondary, unessential. Clearly also, that thesemethods of treating the photographs must be adapted to computers,which are capable of operatioznaflfulfilling this working/treatment.In the present work are presented the methods, which, as itseems to us, can be placed as the basis of the solution of this*problem.Each photograph of the cloud cover contains many different finedetails, precise transfer and analysis of which actually are notnecessary. This is the excess information (" informational noise"),which cannot be, in any came at present, rationally used. For thepractical use of photographs entire information, which is containedin photograph, must be reduced to the limited, comparatively smallnumber of integral characteristics.Itis necessary to further study

DOC 791140700PAGE14statistics and dynamics of these characteristics, to trace theirconnection/communication with meteorological processes of macro- andmesosca is.The correct selection of fundamental characteristics hasimportant value. in different problems this selection can bedifferent, hardly it is possible to propose anything universal. Themore or less reliable system of characteristics, probably, it will be m.possible to obtain after the acquisition of adequate experienceaccording to the analysis of photographs and studies of theirconnection/communication with meteorological processes. But it seemsto us doubtless that as the basis of similar analysis must be placedthe geometric characteristics of photograph-portion of the area,occupied with clouds, grain of cloud formation/educations, direction,degree of elongation and curvature of cloud bands, convexity andconnectedness of separate cloud nuclei, etc. Problem is reduced tothe development of the machine methods of detexmining the geometriccharacteristics of photographs.Let us note that a similar problem arises also during theanalysis of the radar traces of clouds and photographs.For the analysis of television photographs were proposedseries/row of the diagrams,based on the theory of random processes,

DOC-79140700PAGN5information theories and identification of forms (491 Formally theproblem of pattern recognition can be realized both by the determinedway on the basis of certain of the sufficiently complete system ofthe parameters and with the kelp of the algorithms of recognitionwith stochastic instruction. In this article we propose the strictlyformalized, or determined methods of the interpretation ofphotographs.4in view of the complexity of the forms of satellite photographsduring the recognition of cloud systems without preliminaryparametrization the requirements for the memory will considerablyexceed the possibility of contemporary ZTsVff [digital computer].Furthermore, the determined instruction always is more reliable thanstochastic.The values of our parameters can be considered as the componentsof the vector function, determined in the space of data of thebrightness of surface elements, moreover if this space hasdimensionality on the order of thousands, then the dimensionality ofthe space of vector function does not usually exceed 10-15.Page 26.For machine working/treatment to photographs it is necessary to

DOC 791410700PAGE6give digital form. Image brightness at discrete/digital points can beobtained either by the method of photographic photometry or itisdirect by TV signals from satellite. So that working/treatment ofphotographs in digital form would prove to be virtually appropriate,itis necessary to create the device/equipment, converting the analogTV signals, obtained directly from weather satellite, intodiscrete/digital signals in code representation for directinput/introduction into machine.At present there is no generally accepted system of theclassification of cloud images with ISZ, similarly, for example,accepted during ground observations. The comparison of thesesatellite and grouzid observations of cloud cover showed that theresults of determining the cloudiness, as a rule, coincidesatisfactorily, whereas the identifications of the forms ofcloudiness frequently diverge. Satellite photographs, for example,usually do not make itpossible to discover the scattered cumulusclouds and the thin translucent clouds of upper and average of tiers(2].This question was in detail examined by J. Conover [10). Theyproposed the detailed diagram of classification telev--'ion of thephotographs of clouds with ISZ (Fig. 1).According to this diagramthe clouds are divided into two groups: the cloud of cumulus and

DOC 79140700PAGE7noncusulus forms, in turn, the clouds of each group are subdividedinto banded and nonbanded. Then occurs the more detailed distributionof clouds in dependence on straightness and camber of bands,convexity and concavity of cloud nuclei for the clouds of cumulusforms, fibrousness and noafibrousness of the clouds of noncumulusforms.In this article are proposed the methods of the automaticdetermination of the parameters, which characterize the properties ofthe tele-photographs of cloudiness indicated. Of course the diagramof Conover is not universal. In work (3],for example, is proposedanother classification of the photographs of the cloud systems,obtained by Soviet by the cosmonauts: clouds are subdivided intothree basic groups.The first group includes the clouds, which develop in uniform inits properties air mass, thus, for instance air-mass clouds. Thesecond group includes the clouds, connected with the existence ofcyclones. Into third group enter the clouds of jet streams, whichappear with stable in the direction high winds.Va accepted the diagram of Conover, since it is sufficientlydetailed and constructed on material of the television photographs,obtained from satellites. Let us emphasize that the described below

MOW-DOC 791410700PIGS8methodology is applied to any classifications based on theexamination of the geometric special features/peculiarities of thephotographs of the cloud cover of the Earth.

DOC79140700PAGE9mmuu06,qAAKy'leoopauu ,QKY OI0(amHOAleq rB lteA ,Aft BAlaSlmlAft He-Alb Kmai.CI 8fleyusuepm.g.PC06 a mCMWIN7182 HeoqIIBle baoumm-(3IBib Hes., ,e(7)manFig.1. Diagram of the classification of cloud photographs.Key:(1).Cloud formation/educations. (2).Noncumulus. (4).Banded. (5).Donconvex nuclei.slightly .(3).Convex nuclei.Nonfibrous.(10).(7).Simple orWinding or bent.Page 27.§2. Characteristics of photograph and the methods of theirdetermination by ETsVM.Let us consider the consecutively/serially differentcharacteristics of photograph and let us point out the methods oftheir determination by BTsfy.In this case we will consider that the