Merken

# Mathematical Formulae Recognition

#### Automatisierte Medienanalyse

## Diese automatischen Videoanalysen setzt das TIB|AV-Portal ein:

**Szenenerkennung**—

**Shot Boundary Detection**segmentiert das Video anhand von Bildmerkmalen. Ein daraus erzeugtes visuelles Inhaltsverzeichnis gibt einen schnellen Überblick über den Inhalt des Videos und bietet einen zielgenauen Zugriff.

**Texterkennung**–

**Intelligent Character Recognition**erfasst, indexiert und macht geschriebene Sprache (zum Beispiel Text auf Folien) durchsuchbar.

**Spracherkennung**–

**Speech to Text**notiert die gesprochene Sprache im Video in Form eines Transkripts, das durchsuchbar ist.

**Bilderkennung**–

**Visual Concept Detection**indexiert das Bewegtbild mit fachspezifischen und fächerübergreifenden visuellen Konzepten (zum Beispiel Landschaft, Fassadendetail, technische Zeichnung, Computeranimation oder Vorlesung).

**Verschlagwortung**–

**Named Entity Recognition**beschreibt die einzelnen Videosegmente mit semantisch verknüpften Sachbegriffen. Synonyme oder Unterbegriffe von eingegebenen Suchbegriffen können dadurch automatisch mitgesucht werden, was die Treffermenge erweitert.

Erkannte Entitäten

Sprachtranskript

00:00

Chen ladies and gentlemen I like to tell you how we recognized online mathematical formula using the glamorous so here the outline of the local modulates the problems of the revolt our contribution I describe briefly How do you grammars

00:18

can be used and what is the basic idea behind as of show you something from the implementation W what is our future plan outshone the true short demonstration that the plant preseason questions don't hesitate to interrupt me during the talk openly with reporters questions so this also modulated from this your ethical point of view not from the practical use of my view of this ,comma and particle

00:48

emissions and I'm interested in the relation between statistical commissioned a structural 1 solely mathematical formulae used boy for us on which we can play and reason idea children this is the kind of people work on it so we can compare the results the others see you soon beaches theoretical background is described in our book with the promise of Schlesinger which was published the cooler he doesn't do and its application or application only two-dimensional grammar school images Wallstreet topical value grew from being these officials defended the 2005 In this work I will beauty at all only previous work which was published in the International Conference on the Internet is in recognition in September last year when we show me how to recognize the mathematical formulae indeed offline case which means if there is a scanned document and how can you from the scandal in this report I'm going to put men on the online version in which there is a public Europe tablet on which you can draw of strokes as an input and so this is what this here what is the problem of mathematical formulae recognition so let's take a simple formula like this 1 and if we know what is the but body operators which will

02:23

apply different and bosses a binary operation which divides the formula into .period problems here the 2nd 1 for that fraction this 1 is a follower which is then again divided into In a binary operation to minors be sold each formula a duration trees the corresponding generation and our aim is to create this derivation real particularly from for offline and online merchants the usual approach may offline version also an on-line is still starter is the is the symbol detection and their

03:02

recognition and having the symbols than doing the structural analysis Our case we really don't be things have basically in wanted which allows to go over the main problem which is the main problem is if the civil authority and I strongly then there is basically no good way to recover from Helen but it

03:33

doesn't make sense the spectral analysis violent environment in our case to the take several possible options and he traveled there in the structural analysis there are a lot of ambiguities in the so if you will of U.S. office scanning the skies resolve so in this case the expected

03:57

interpretation of this 1 but maybe you know that I can medical possible confusion so Bloomberg In problem this is related to the

04:18

mathematical formulae however the Google promises Schlesinger has been working on similar problem in the area of the command analysis for all the time and he was the 1st author of the book I mentioned and they tried to electric schematic diagrams successfully and they also tried to sheet music recognition using the structural recognition using structural construction so we I'm just reporting about to be formula so what is the topic of this is recognition of all the on-line strokes and

05:03

basically the approach which replied Well 40 offline formulae mentioned it briefly late morning theoretical part of was adopted what we needed to have also adopted new myself for elementary symbol detection and the representation of it there had to be designed and the passing of rhythm which creates the derivation for you have to be modified because the indication of wine formulae .period was the bounding boxes and to hear some orders structures need to be represented representative of all the strokes and the proximity in to space so the on-line cases attempt to detect all sorts of strokes that can form a symbol and to in and

05:59

then rank them according to the plausibility and take a few possible options over that to speed up the process and make it simple being too you probably the take into account some of simplification assumptions and the simplification assumption is that once symbol can be In the current implementation composed only all 4 strokes and we don't consider all the strokes which are available but only those which are in the proximity so which is there is some kind of neighborhood than its neighbors the the chosen candidates .period an unrelated by DOC are In this case for OCR to the for on-line symbols and they are assigned to labels and then because of our processes optimization which minimizes some penalties to the penalties are assigned to each of the 2 nations so there is the example just briefly look at the these commerical street between here and of course this can have various interpretations this part because it's disconnected can be interpreted as if W also interpreted this as Peru ruled this along this line can be also a fractional line can be also a minus sign and this ambiguity sodden disambiguated in the passing process so we have to keep all the interpretations the beginning so now I'll give you a last the Swiss Life's sought to slide introduction alters the structure construction which is the basic theoretical

07:47

background over the approach the idea is that for each each constructed as the region in all which has grown from scratch and the it terminals in the offline case I individual pixels and in the on-line case these are strokes and some kind of area related to those it don't strokes it's still a little more complicated so are just so that the people of this and the duration of such a region is basically maiden degenerative made by applying the production rules of the appropriate grammar and to these rules have some penalties assigned so basically the duration can be assigned a panel the some of these penalties corresponding duly particle regions then summoned to leave on a global penalty later on and the final interpretive interpretation is the areas the region 2 strokes in the online case which covered all available in each and which has been minimal penalties so this is the case from the off-line because it's easier to do show so let I have region and is as this 1 which is ruled which is

09:20

the composed of all 4 regions no they noted Europe ABC and the and you can see that there a need regions in the offline considerably rectangles the combined by some kind of concatenation knowledge within the production Center if you'd like to come the because of course the problem is complicated and would even be possible single funding dramas context 3

09:57

Grandma's completely so in the case of duty dramas it's even more complicated so did here are some simplifying assumptions which allow us to dramatically reduce the complexity to make the whole approach useful practically so the region's Orrick bundles which allowed tool for adjusting the rectangles which which ultimately wanna basically interested in what is inside the EU and in such a case that to conflict-free grammar there can be constructed for which there is a passing algorithm because the generalization of famous people younger possibly on words and complexity is relatively low the polynomial you can see here and an our sizes sold the image which correspond to the so it's practical fees it practically feasible the production rules are very simple soldier on the ground basically the flow of any judgment will be the nation's 1 is

11:12

horizontal and 1 vertical combination and then that's it of these simple formulae these simple rules are enough unfortunately for amount article formally A. There is a need for extensions and this extension is supposed

11:36

to bring the mutual position all the symbols the beach the ordering Grandma prudent to take into account for instance if you like to have the power and then the notion of a size should be in taken into account and then also the relative positions of the 2 the same problem would be 40 fraction of what a fraction moreover there is some uncertainty because of the the appropriate rectangles so Indy handwritten case

12:16

wouldn't be precisely aligned children must be some kind of elasticity specific which is explicitly expressed in terms of mutual constraints between the boxes represented to some parts of the so the implementation of the who is available or something like that politician is done died in the In yellow because

12:44

reflected and to a it's it said ball just for the best thing in the verdicts of the momentum is so simple mathematical constructs like numbers and disease come ,comma Munir in binary operators Barbaro power cooperated fractions subscript superscript songs and into integral and of course the mess of things slower in circumstances that are resistant to noise which all show you In late so also my eyes lost like at the 1st World War body experiences the experiences of that

13:37

the massive lost their staple 400 formulated this is the precision and which is 88 per cent rich if the OCR terrorism they can intercom let me remind you that you'll see a lot in the public domain tool which is used in applied and if the errors made by the OCR pool are subtracted then the level of 97 per cent the approaches so fast you can see the fraction of a 2nd for a wellbore formula in a marriage beaches so those greater than in the offline case the goal of the year sequence of the stroke that given additional information which makes a which tremendous and so opinionated that's practically useful maybe it might be it might be really applies so future work in his life he tries to do take a book mathematical book and tried to beat the system to be able to call with the problems in the book Olson offline and also an online version this is a project which doesn't have any money so it's like the included project of enthusiasts now I find that people must student the lady she's been there she'll be working on it for the next several months and she would be doing nothing else so hopefully to review progress on area for for a presentation and let me will demonstrations which is the here is the tool which is the job approach

15:27

Digital implementation simple user interface and to so for instance

15:33

if you will look for formulae like that the start the industry people line ones which were drawn on the tablets stored in some form of which takes into account the the sequence of strokes and if you like to recognize the formula the formula is going to be recognized here is the outcome of the formula what you have here is the duration 3 which tells what rules are applied in both sequence for dead particulary formula and of course the Eureka and these side green boxes show the individual part of the formula and Kennedy can comedy can for instance if filed a show so this is the stroke number 6 for instance he conceded here you can look it up in the the operation created policies that which used and there is a problem because the EU production how to find then I can show you that it can even work for the drawing Inc because I don't have a problem with missile is a simple way help draw by miles long tried to and it's not fair my handwriting is playing for someone else's life but let me try to show it is for this study on so I think I'm gonna Mister Chairman few questions about this fact in a the friend so so well some problems in all of this is relatively complicated issue so it's 1 way is to assign a monumental and that's the reason we have relatively small amount of production roles and the 2nd approach is cold during the during the learning process so you be training set which is available offline too to assigned penalties and this is the problem he recalled an article would differences actually working on the but income and again it's 1 of the city's penalties are created and the and that call on a phone for the channel team not only also for symbols sofa and if if you like to see the more detailed description of how the penalties are described the look at the bigger paper paper 2 thousand 740 offline formulae and on the beach he can each have had exactly it's so complicated process and like regulated simple the there is no you it's the most common area it's more the there was a destructive sometimes for instance specially fractions and their roots in our implementation if you will not material tried to make some of the year so for instance if you if you drop tonight it doesn't it doesn't remember triggered longtime leftovers from his courage so that all may discourage them anything like that so then the authority only outcomes would it doesn't recognize my skirt or a dozen other people made history too but they could that will get out of the doesn't it is not able to in the context this symbol but if there are problems of this sort that it's the main problems which are not based on it all depending that it is the arriving water-dependent playing the stroke gets into your mind and your final .period we still a no more aggression the question is is counted on the amount of it's 1 the wrong from Uncle the policyholders and other days the

20:26

don't this is what it's all about I don't know if this is for it to be but there is no longer form at the moment it's all over the world is available this experimental school so it's not you know if you like the white the consented but you will not get anything you have at hand have now gone as you suggest they don't want you to know that nothing have if you think about this and this and that was it it but this is also of course the play which can help and it's possible that we didn't consider it at this moment but the construction of housing because if you have someone thinks and you know who you are operating on the mathematics is based physical don't wonderment that mechanics then you know what kind of formulation can expect and this this can cost constraints the possibility of nations grounded out so it's possible tool is implemented there but this is not our main problem only after fighting is much more basic you can have the house and I don't think it was back at the time of year yes something that would be yeah the main message I wanted to mediate to you is that there is something like the drama and that they are practical that's the main message and that was the reason I was going it was this is what my my interest and now it's useful know so maybe we can do about that but for a time money the commander in the garrison command papers concerning for medical conditions and also the last time using the demand I'm an assistant bishop ,comma you should use this difference and the next 50 years practical experience preparing you with the obvious experience is 1 tool which called which is called Marsh cat I think is the name of the school so this is really about paper prominent harmonization and as hasn't can 90 70 something it's totally anytime on origin the glamor to the but I don't think that I don't think so armistice was how I have you arrive here at home and that it was not so the fact that this the Irish favorite and then also said similar succeeding papers so we should clarify the difference maybe you have a son on Monday practices of his emphasis is also follows the Foshan or development in the bathroom Commission area as you might remember at the beginning of a decent and the seventies structure but recognition was very fashionable among Drexel for kinks so full books and so on and then it died out because people didn't find it was used for a practical case and the other trying some of the others will be either of the question he needed make a comparison another yesterday and the presentation of the you on the other hand ,comma the reason for this he also had harsh have met the above have I'm I have to all of those who may have run out it is apparent did the asthmatic and you have to have feeling position that take 240 sedan anything looked into the cause of the thank you very much

00:00

Punkt

Ideal <Mathematik>

Explosion <Stochastik>

Modul

Computeranimation

Ausdruck <Logik>

00:47

Physikalischer Effekt

Resultante

Nichtlinearer Operator

Bruchrechnung

Mathematik

Einfach zusammenhängender Raum

Polarkoordinaten

Statistische Hypothese

Kartesische Koordinaten

Endlich erzeugte Gruppe

Derivation <Algebra>

Gerade

Physikalische Theorie

Computeranimation

Inverser Limes

Ausdruck <Logik>

Topologie

Freie Gruppe

Leistung <Physik>

Reelle Zahl

Explosion <Stochastik>

Dimension 2

03:01

Thermodynamisches System

Graphische Statik

Derivation <Algebra>

Punktspektrum

Stochastische Abhängigkeit

Analysis

Computeranimation

Analysis

03:56

Diagramm

Flächeninhalt

Gruppe <Mathematik>

Vorlesung/Konferenz

Analysis

Analysis

Ausdruck <Logik>

05:02

Nachbarschaft <Mathematik>

Prozess <Physik>

Quader

Minimierung

Vorzeichen <Mathematik>

Derivation <Algebra>

Grundrechenart

Frequenz

Physikalische Theorie

Raum-Zeit

Computeranimation

Ausdruck <Logik>

Algebraische Struktur

Gruppendarstellung

Vorzeichen <Mathematik>

Sortierte Logik

Modulform

Mereologie

Quadratzahl

Ordnungsstruktur

Indexberechnung

Abstand

Gerade

07:45

Dämpfung

Flächeninhalt

Rechteck

Radikal <Mathematik>

Derivation <Algebra>

Schlussregel

Überlagerung <Mathematik>

Biprodukt

Computeranimation

09:56

Erweiterung

Rechteck

Relation <Mathematik>

Schlussregel

Kombinator

Biprodukt

Massestrom

Komplex <Algebra>

Arithmetischer Ausdruck

Computeranimation

Ausdruck <Logik>

Freie Gruppe

Polynom

Faserbündel

11:35

Umwandlungsenthalpie

Ortsoperator

Bruchrechnung

Nebenbedingung

Ortsoperator

Quader

Rechteck

Ikosaeder

Erweiterung

Term

Arithmetischer Ausdruck

Computeranimation

Freie Gruppe

Mereologie

Punkt

Elastische Deformation

Term

Leistung <Physik>

12:44

Impuls

Folge <Mathematik>

Prozess <Physik>

t-Test

Geräusch

Zahlenbereich

Nichtlinearer Operator

Stichprobenfehler

Computeranimation

Ausdruck <Logik>

Übergang

Arithmetische Folge

Leistung <Physik>

Bruchrechnung

Nichtlinearer Operator

Mathematik

Binärbaum

Physikalisches System

Binärbaum

Variable

Integral

Integral

Flächeninhalt

Leistung <Physik>

Projektive Ebene

Quadratzahl

15:24

Subtraktion

Folge <Mathematik>

Wellenpaket

Prozess <Physik>

Quader

Fächer <Mathematik>

Zahlenbereich

Bilinearform

Extrempunkt

Computeranimation

Eins

Ausdruck <Logik>

Deskriptive Statistik

Wurzel <Mathematik>

Gerade

Regulator <Mathematik>

Gammafunktion

Bruchrechnung

Nichtlinearer Operator

Green-Funktion

Schlussregel

Biprodukt

Ausgleichsrechnung

Frequenz

Sinusfunktion

Menge

Flächeninhalt

Sortierte Logik

Mereologie

20:23

Harmonische Analyse

Nebenbedingung

Subtraktion

Mathematik

Momentenproblem

Ortsoperator

Physikalischer Effekt

Physikalismus

Bilinearform

Paarvergleich

Kombinatorische Gruppentheorie

Computeranimation

Sinusfunktion

Algebraische Struktur

Flächeninhalt

Mechanismus-Design-Theorie

Gammafunktion

### Metadaten

#### Formale Metadaten

Titel | Mathematical Formulae Recognition |

Serientitel | DML 2008 workshhop - Towards Digital Mathematics Library. |

Teil | 03 |

Anzahl der Teile | 14 |

Autor | Průša, Daniel |

Lizenz |
CC-Namensnennung 3.0 Unported: Sie dürfen das Werk bzw. den Inhalt zu jedem legalen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen. |

DOI | 10.5446/21271 |

Herausgeber | River Valley TV |

Erscheinungsjahr | 2012 |

Sprache | Englisch |

#### Technische Metadaten

Dauer | 25:47 |

#### Inhaltliche Metadaten

Fachgebiet | Mathematik |