EP2050022A1 - Procédé pour la fabrication de matrices d'images extensibles - Google Patents

Procédé pour la fabrication de matrices d'images extensibles

Info

Publication number
EP2050022A1
EP2050022A1 EP07786557A EP07786557A EP2050022A1 EP 2050022 A1 EP2050022 A1 EP 2050022A1 EP 07786557 A EP07786557 A EP 07786557A EP 07786557 A EP07786557 A EP 07786557A EP 2050022 A1 EP2050022 A1 EP 2050022A1
Authority
EP
European Patent Office
Prior art keywords
matrix
image
nodes
link
objects
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
EP07786557A
Other languages
German (de)
English (en)
Inventor
Maximilian Schich
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Max Planck Gesellschaft zur Foerderung der Wissenschaften eV
Original Assignee
Max Planck Gesellschaft zur Foerderung der Wissenschaften eV
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Max Planck Gesellschaft zur Foerderung der Wissenschaften eV filed Critical Max Planck Gesellschaft zur Foerderung der Wissenschaften eV
Publication of EP2050022A1 publication Critical patent/EP2050022A1/fr
Withdrawn legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/54—Browsing; Visualisation therefor

Definitions

  • the present invention relates to a method for producing image matrices.
  • the method belongs to the technical fields of image science, data processing and network science (so-called "Science of Complex Networks").
  • An image matrix is generally a two-dimensional array of images in rows and columns.
  • the position of the image (row, column) in the image matrix represents information about the relationship of the image to the contents (meanings) of the associated row and column positions and a relationship between the row and column positions.
  • Image matrix is a visualization (display) of the images, which has previously enabled the viewer to recognize relationships between images or between row and column positions.
  • image matrix is generated by the images contained (images) are specially created in the production of image matrices.
  • An initially empty table is filled with new images.
  • a conventional image matrix merely provides a systematic presentation of pre-existing ones
  • a further object of the invention is to provide a storage medium or an electronic data processing system comprising a processor and a storage medium in order to carry out the method.
  • the object is achieved by a method having the features listed in claim 1: providing a network with a set of link output nodes and a set of link destination nodes with links therebetween is followed by forming a matrix with rows and columns, with link output nodes associated with the rows and link target nodes to columns or vice versa. Finally, the searched image matrix is generated by placing visual representations of the left-hander nodes or link-destination nodes in place of the links in the matrix.
  • the invention provides an image matrix which comprises a visualization (display) of images and advantageously enables the viewer to recognize or produce relationships between images and / or to subject the images to data processing and / or data maintenance.
  • the advantage achieved by the invention is, in particular, that the invention, when dealing with larger quantities of classified objects, facilitates the investigation of phenomena in which there is a connection between the classification and the visual properties of the objects and / or the classification criteria.
  • the method facilitates the explication of direct dependencies as well as the extraction of diachronic phenomena from a given set of classified (BiId) data. Furthermore, in contrast to conventional list representations and overview tables, the method allows the simultaneous examination of data in the context of two data dimensions. Compared to the prior art, this means a considerable acceleration of the work, since cumbersome navigation in the amount of data is eliminated.
  • the method advantageously represents a completely or partially automatable tool for processing large amounts of data.
  • the image matrix according to the invention can be constructed from the data volume without prior knowledge, in particular without the user's knowledge of existing correlations between data.
  • link output nodes of the above-mentioned network are classifiable objects and link destination nodes classification criteria or vice versa.
  • objects and / or classification criteria thereby comprise a set of persons, locations, time ranges, physical objects, conceptual objects, events and periods.
  • An event is the meeting of several of the aforementioned objects in a node, ie, for example the gathering of a physical object, a location, and a time range in a residence event.
  • Periods include continuous, non-discrete expansions in one or more of the named subject dimensions, ie, for example, a style period that has both spatial and temporal extent over multiple locations and time ranges.
  • An advantageous embodiment of the specified features is given if objects and / or classification criteria are represented by individual nodes or as a group of nodes. Items and classification criteria represented by a group of nodes are multi-part.
  • Items and classification criteria represented by a group of nodes are multi-part.
  • there are multi-valued links between multipart objects and classification criteria since more than one node of the respective multipart article or classification criterion can be linked.
  • multipart objects are possibly linked only indirectly, eg via subordinate subnodes with a classification criterion.
  • edge ⁇ As the relationship between source node and destination node expressed in the value of the matrix cell thus becomes considerably more complex, it is referred to below as "edge ⁇ " for better distinction.
  • the edge between a (multi-part) object and a (multi-part) classification criterion can contain one link or several links or be empty.
  • a multi-valued link of / or to a multi-part article or a multi-part classification criterion either a detail image matrix, a one-dimensional overview table or an image montage is placed.
  • multipart, in particular hierarchically subdivided objects or classification criteria in the matrix are unfolded in an input signal into a plurality of matrix rows or matrix columns or combined into a matrix row or matrix column.
  • the image matrix can be output in the course of the method on an output device, in particular on a screen or a printer, or in a file.
  • additional information is additionally placed on the matrix elements of the image matrix, which are retrieved from a database and belong to the respective link output nodes or link destination nodes.
  • additional information may include, for example, further data relating to a visualized image.
  • a data processing can be provided, in which the data of the image matrix (images, texts and / or other information to the matrix elements) further processing, preferably a data input, an image recognition, a correlation and / or a rearrangement of the data.
  • the edited data is then stored and / or output as a processed (modified) image matrix.
  • the storage of the processed data can be done in an amount of data from which the provision of the network takes place.
  • the information of the data volume can be automatically enriched and completed for further use.
  • the method can be used, for example, in the areas of Bibliometrics (explication of implicit picture quotations), art history (reception, tradition, mnemosyne), network science and copyright issues.
  • Further objects of the invention are a storage medium and / or an electronic data processing system, which comprise a processor and a storage medium, wherein the storage medium includes software that causes the processor to carry out the inventive.
  • FIG. 1 shows a flow chart with an illustration of the method according to the invention
  • Figure 3 illustrates the formation of the image matrix, with visual representations of the nodes replacing the links
  • FIGS. 4a-c image matrices with increasing information density
  • FIG. 5 Within an image matrix, the assembly of relevant sub-nodes leads to better comparability of the representations
  • FIG. 6 shows three simple steps from the matrix to the image matrix
  • FIG. 7 is a block diagram of the general procedure for producing an image matrix
  • FIG. 8 shows the raw form of the base list (adjacency list).
  • FIG. 9 shows the extraction of various record numbers (node IDs) in the base list, which permits the external answering of simultaneous local, global and metalocal questions or the reconstruction of a tree structure;
  • FIG. 10 shows a general procedure for producing a matrix (detail from FIG. 7);
  • FIG. 11 shows a detail of a matrix
  • FIG. 12 shows a section of an image matrix
  • FIG. 13 shows an illustration of an edge, which may contain a different number of links in matrix rows or matrix columns of different abstract
  • FIG. 14 shows a detail image matrix which offers a better allocation of information, while a detailed overview offers larger images on a comparable surface;
  • FIG. 15 shows a scaling or zooming of a matrix: local> metalocal> global
  • FIG. 16 shows strict node trees which are similar in zooming to the directory tree in a conventional operating system. Allow system (icons to Windows Explorer TM).
  • the main steps of a preferred embodiment of the inventive method are shown in Figure 1 and in the block diagram of Figure 7.
  • the data includes link output node data, link destination node data, and data characterizing the links between the link outbound node and the link destination node.
  • the data can generally be present as image and / or text data, wherein the data can be visualized by at least one of the link output nodes and the link destination nodes (eg also set text of a scanned book page).
  • the data is provided at step S2 in the form of a base list (BASE in Figure 7, contents eg Figure 8) which contains all the information necessary to construct the matrix and the image matrix.
  • the basic list is a data list with the structure described below, which in one Data store is stored, which may be connected to or extracted from the database.
  • Row and column positions are formed by the listing of link outbound nodes and link destination nodes (or vice versa).
  • the matrix elements ie, the cells of the matrix
  • the matrix elements comprise a zero (no information) if there is no link between the link output nodes and link destination nodes of the associated rows and columns, or a matrix element containing information about a single or multi-valued link between the associated link output nodes and Link destination node includes.
  • This information is obtained from the so-called "edge-set" information from the base list, where a function (subroutine) is placed at the respective matrix elements, with which it is queried whether the relevant combination of row and column occurs in the "edgeset". If so, the valence of the relationship (valence of the link) is queried.
  • a univalent link will do that
  • FIG. 4a Detail matrix
  • FIG. 4b overview table
  • FIG. 4c image montage
  • the desired image matrix is constructed from the matrix, in which matrix elements are replaced by visual representations of the associated link output nodes or link destination nodes.
  • a selection of the visual representation depending on the valency of the link (edge value) can take place.
  • the image matrix comprises the data of images associated with rows and columns of the image matrix, and possibly additional information. The data is available to a user who, for. B. wants to investigate relationships between images or between row and column positions.
  • the further use of the image matrix is simplified if at least a part (detail) of the image matrix is output.
  • An output of the image matrix may be made to a display device (e.g., display, printout) or to a data memory (step S5).
  • a decryption of the image matrix takes place.
  • a further data processing can be provided after step S3, S4 and / or S5, in which the images, texts and / or further information of the image matrix are subjected to further processing (step S6).
  • further information may be input from other data resources to further enrich the information represented by the image matrix.
  • An image recognition may be provided to detect and evaluate certain images (patterns) in the cells of the matrix. Between the images, if necessary after the image recognition, a correlation of specific partial images may be provided in order to establish relationships. Furthermore, a rearrangement of the data can be provided.
  • the data processing in step S6 may be performed by a user or automatically by available data processing programs, which for the respective functions, eg. B. image recognition, correlation are established.
  • step S6 the image matrix is created after a renewed passage from Sl to S3 to S6 (step S4).
  • the edited data is then saved.
  • the storage can be done in the output dataset or in a separate memory.
  • a modified image matrix can be constructed with the processed data.
  • a lot of classified items can be e.g. understand as a network of nodes and links.
  • Objects and classification criteria each form a node type; the assignment of an item to a classification criterion is done by the classification link.
  • the classification network defined in this way can be mapped as a matrix like any other network.
  • the classified objects are visually representable objects, then it is possible to correspondingly enrich the conventional matrix and to convert it into an image matrix.
  • the simple links are replaced by images of the network nodes, ie images of the objects or the classification criteria. It makes sense in many cases to pick out a part of the object that corresponds to the linked classification criterion or vice versa.
  • the method therefore appears to be particularly useful in particular if the objects in question or the classification criteria are present in a subdivided, possibly hierarchical form or in a form that can be combined into higher units.
  • the visually representable objects can assume the role of the object as well as - in special cases - those of the classification criterion.
  • Corresponding objects in the role of the object are also referred to below as image documents.
  • An image document is defined as an arbitrary, visually represented or representable object or a collection of several of the same. Typical examples of image documents are a book with illustrations, a book with scanned text pages, a hand drawing, a sketchbook, a photo, a photo collection, the photos of an Internet user or a homepage.
  • Typical examples of composite tangible classification criteria are keyword sets or tags ⁇ , which are combined into meaningful groups like Tagclustern ⁇ .
  • Typical examples of subdivided classification criteria are hierarchical systems, thesauri or ontologies. More or less complexly subdividable and, at the same time, higher units that can be summarized are sets of discrete objects such as websites, locations or physical and conceptual objects. Specially man-made objects such as ancient monuments, monuments or paintings often appear both in the role of the classification criterion and in that of the object, for example when the classification link describes the indefinite or directly demonstrable dependency of an object on other objects (reception or reception) tradition).
  • the image matrix is understood in principle in the present context as a special form of the conventional matrix.
  • the matrix therefore forms the starting point in their production. It is first enriched with the necessary information to nodes and links and then converted into a picture matrix in a simple step. The enrichment can either be taken directly from the initial data set, or it is stored in a new adjacency list and kept. This new adjacency list serves as a temporary database during editing and analysis of the image matrix. It will be referred to as a 'base list' in the further course.
  • the base list may contain the entire network of output data or only part of it, and must be recreated or updated after each major change.
  • the example network of the visualization is that of the reception.
  • the general workflow when creating an image matrix usually takes place on the basis of a few simple steps (FIG. 6): First, by sorting the matrix (permutation), as many correlating rows and columns of the matrix as possible are brought together , so that a region of particular density of filled cells (edge value greater than or equal to 1) is formed. In a further step, the unneeded rows and columns are filtered away so that only the relevant area remains visible. Finally, the filtered area is converted into a picture matrix by clicking.
  • the nodes of a network are represented as rows and columns; the edges (single or multi-valued links) as points or cells.
  • An extension of the simple matrix means weighting the links with a particular value. This is useful, for example, for a network in which the source and destination nodes of the links to groups or hierarchical structures are combined.
  • the value of the matrix cell, hereinafter referred to as the edge, here corresponds to the number of actually allocated links between the respective summaries.
  • the summary and weighting of the matrix rows and columns can be realized, as is provided in the prior art in the "block modeling" in the so-called social network analysis.
  • the corresponding value is 3 ⁇ ( Figure 2b ).
  • the weighted value of such an edge represents a detail matrix of the individual sub-nodes of the respective complexes (FIG. 2c). From the above cases, there are three possibilities for the matrix: Either instead of the respective edge, CP or '1' (according to link present or not), a value greater than 1 ⁇ appears (if the link is a summary of several links ) or a detail matrix (if the partial links are to be shown explicitly).
  • the content of the edges is replaced by the image of the linked document (part).
  • the mapping corresponding to classified (detail) nodes takes the place of the links between the document and classification nodes ( Figure 3).
  • the individual document (part) s and (sub) classifications can also be combined in the (image) matrix into higher-level (global) or intermediate (metalocal) units. Weighted edges with a value greater than 1 do not correspond in the matrix to a single link, but rather to several links between the connected nodes, possibly combined in several parts. There are basically three possibilities:
  • the second method is filling the cell with the mappings of the relevant single quadrants, without adhering to the order of the detail matrix - a procedure that makes sense especially with extensive detail matrices, otherwise the images often become too small (Figure 4b).
  • the third method involves the assembly of the partial representations contained (FIG. 4c) - an often useful application that significantly improves comparability, especially in the case of middle higher-level units.
  • the left part of FIG. 5 shows three details from a sixteenth-century hand drawing code (, Doc template ⁇ ), which are all linked by the classification link with a particular section through an antique building ("Monument *").
  • Doc template ⁇ sixteenth-century hand drawing code
  • the three parts are provided as provided by the authors of the Codex.
  • the comparison with the clearly dependent section of another collection of drawings (doc. Copy ⁇ ), which can also be seen in the matrix, is much easier thanks to the assembly in the second illustration.
  • montage possibly represents the link between an ideal parent document unit and the corresponding superordinate classification criterion - a relationship that may possibly even exist in the original data set in this form does not exist, as there are usually only the links between actually existing document (s) and possibly subordinate classification criteria are listed.
  • the image matrix therefore shows itself through the use of montages as an independent product. It is not a pure representation of the existing data, but goes beyond what has been found in the statement.
  • the already mentioned base list ⁇ (FIG. 8) or a dynamic equivalent can be present implicitly in the output data set or can be created externally.
  • the basic list of FIG. 8 is shown in three separate partial images (FIGS. 8a, 8b and 8c).
  • the base list contains information about nodes and links in the source dataset. In this case, various relationships between nodes can occur, which are shown schematically in FIG. FIG. 8 shows that the extraction of different record numbers (node IDs) in the base list permits the external answering of simultaneous local, global and metalocal questions or the reconstruction of a tree structure.
  • the base list is an admias list enriched with metainformation on nodes and edges of a network, which can serve both to produce scalable (image) matrices and to produce classical network visualizations.
  • (Picture) Matrix like network visualization tion require a so-called nodesets ⁇ (set of nodes, group of information about the nodes of the network) and one Edge sets ⁇ (edge set group information on the edges of the network). Both are contained in the base list, or can be generated dynamically from the same. Additional enrichments can serve the better sorting as well as the clearer representation of the respective final product.
  • By combining different basic lists it is also possible to combine the different network types (such as reception, traditional or tree structure) in a single visualization.
  • the image matrix of a reception network in FIGS. 12a-12c in superposition shows a second network in classical network visualization - the network of the tradition.
  • step S1 in FIG. 1 The starting point is a database output (step S1 in FIG. 1) which contains all relevant link relationships of a reception network.
  • step S1 in FIG. 1 the resulting simple adjacency list of the links is enriched by node information from another database selection. The procedure is the same for each selected subnetwork. For each type of link in the output data set, a separate basic list can (and should usually be) created.
  • the base list is represented as a spreadsheet, it will conveniently contain three column groups ( Figures 8a, 8b, and 8c) - one to the left exit node, one to the link destination node, and another to the resulting edges.
  • Each line represents in the list N real existing in the output data set existing link (the, self-self-edge ⁇ ).
  • the node set that is, the information about the nodes of the network, can be extracted from the first two column groups of the base list.
  • the edgeset is equivalent to or derived from the third column group.
  • the first two column groups of the base list (8a, 8b) to the nodes are in four sub-groups divided according to detail below defined summaries, self ⁇ , Parent ⁇ , Main ⁇ and, Entity2 ⁇ of the respective Linkausgangs- or link target node.
  • Each of the subgroups contains in the first place the corresponding record number ⁇ (or possibly any other node ID), in the second place the so-called Labeistring ⁇ and in third place the so-called "Occurence":
  • the first column of the four subgroups to the nodes in the base list contains the 'record number' of the source or
  • Target node or the corresponding node of the corresponding summary (see Figure 9,, RecNo ... ⁇ corresponds in Figure 8, Doc ... ⁇ , for example, DocSelf ⁇ or, Mon ... ⁇ z. MonSelf ⁇ ):
  • RecnoSelf * is the record number of the read out node itself.
  • RecnoParent ⁇ is the record number of the first parent node in the existing node hierarchy (part-of-link). It serves, for example, to display the tree structure of a document in a network visualization, in addition to reception and transmission. For the aggregation of superordinate units, it plays only an indirect role.
  • RecNoMain ⁇ is the record number of the node at the top of each node hierarchy, which coincides with the 'global' document unit. During read-out, the node hierarchy is tracked up to a marker determination for this purpose. Each node at the top of a document tree is accordingly marked as, Main ⁇ before being read out.
  • RecNoEntity2 ⁇ is the record number of a possibly existing ideoyncratic, meaningful, metalocal ⁇ unit of the document, which is marked with the help of the marker, Entity2 ⁇ .
  • the node hierarchy is also tracked upwards when reading up to the marker determination.
  • RecnoSelf ⁇ refers to a particular image in a book
  • RecnoParent ⁇ refers to the parent page directly in the book
  • RecNoMain ⁇ refers to the book itself
  • RecNoEntity2 ⁇ refers to a multi-page catalog entry in the book.
  • the second column of the four subgroups of the nodes in the base list contains the so-called Labeistring x . It serves to enrich the respective nodes in the matrix with useful information.
  • Labeistring Document In general, that is, if the source and destination nodes are of different types, it is convenient to define two different formats for the Labeistring ⁇ .
  • a meaningful Labeistring for (picture) documents and then another for the classification (here antique monuments) will be explained by way of example.
  • Type ⁇ expediently specifies the node type of the read-out entry, that is, in the case of documents, for example, whether it is a single object, a publication or a photograph.
  • LabelSelf contains only the name of the node itself. It is necessary, for example, if the tree structure of a document is to be visualized as a network without showing redundant information at the nodes of the tree.
  • 'Label' contains the complete name of the node, which may also include information from superior or, as in the case of the document location, suitably hypotactically linked nodes, eg, for individual objects, the label more or less corresponds more or less to the sequence 'place / institution / Department: Codex / Folio / Quadrant "and for publications of the series" Short Name / Position ".
  • DateName ⁇ contains, for example, the name of the (first) time range used for dating. (Of course, documents can also include the date of origin be rivaled, ie, multiple times, for example, in divergent research opinion.)
  • lstArtist ⁇ contains the first person associated with the document under the condition 'artist'. (Of course, all connected artists or even other people could stand here.).
  • 'ImgFile' contains the reference to the image file corresponding to the database entry, or in the case of documents reprinted only secondarily, the reference to the image file of the first dependent document, provided that this is a photographic copy.
  • the label string of the documents could also be enriched by further additional information - for example, by GIS information on the locality.
  • the bibliography of the classification criteria corresponds to that of the (image) documents with regard to the basic data. If the classifications are more complex structures, such as antique monuments or documents in the example case, then the corresponding Labeistring can be as rich in information as the Labeistring to the document. In this case, there are no additional enrichments for sorting included.
  • the function of the contained fields corresponds to the explanations of the label of the documents.
  • the third column of the four subgroups of the nodes in the base list (FIG. 8) contains the so-called 'occurrence' of the nodes. It indicates the relative frequency of the corresponding entry in the subgroup. It is collected by simply counting the similar, record numbers ⁇ in the first column of the subgroup. It corresponds to the output node OUT degree or the destination node IN degree.
  • the 'Occurence' must be recalculated to a subset of the output data set in the case of a limitation of the base list ⁇ . Simply reading out the total number of links to an entry from the output data set may make no sense because the restriction does not have to correspond to the existing data of the output data set.
  • the subgroups of the two column groups to the nodes in the base list can also contain information on images and sorting.
  • the fields, Image ⁇ (and, Imgext ⁇ ) in the column subgroup, DocSelf ⁇ ( Figure 8), contain the link to the respective image file or the respective section of an image file, which is important for the image matrix.
  • the record number specified therein may differ from that of the node itself, for example if the image file is from a reprographic copy - a feature that may be marked by a marker in the image matrix.
  • the 'Sort' columns in the column groups 'DocMain' and 'DoCEntity2' ( Figure 8) come from the sorting of matrices created from the base list. If necessary, the information is imported back into the basic list by means of a macro.
  • the third column group of the base list ( Figure 8c) to the edges contains up to nine subgroups (3 link starting points by 3 link targets) from the three summarization levels, Seif,, Main ⁇ and, Entity2 ⁇ . Only the two ratios, DocMain-MonMain 'and, DocEntity2-MonSelf ⁇ are shown .
  • Each subgroup contains the corresponding edge in the first column, which is created by simply concatenating the corresponding record numbers.
  • the second column of each subgroup contains the, Edgeoccurence ⁇ , which is calculated in the same way as that of the individual nodes.
  • the Edgeoccurence ⁇ can serve as an indicator of the density of various classification complexes in a large document.
  • the statement quality is of course variable, since, for example, a single good drawing can exceed many bad sketches in importance.
  • the raw form of the database output corresponds in the case of simple links between source and destination nodes, for example, the following form:
  • link home nodes and link destination nodes are represented exclusively by their ID (record number, primary key or URI ).
  • Raw Edgelist (adjacency list): Link Root Link Target RecnoDocl RecnoMonl RecnoDocl RecnoMon2 RecnoDocl RecnoMon3 RecnoDoc2 RecnoMon4 RecnoDoc3 RecnoMon2 RecnoDoc3 RecnoMon5
  • Each link exit node thus faces a single link destination node.
  • Each line thus contains a single link ratio, which explicitly exists in the database in this form. If, for example, the links in the output data set are represented as independent event nodes (or as a crosstab in the case of a relational database), the result of these events can also be read out directly. The output then immediately corresponds to the two-column form.
  • the two-column adjacency list is enriched by additional node information.
  • this allows to summarize the nodes and link relations to global and metalocal units in the matrix as well as to sort the (image) matrix according to criteria such as designation, location, date or artist of the respective nodes.
  • the enrichment of the raw adjacency list is done by simple database readings of all relevant nodes (eg documents and monuments) in the form of the above-described Labeistring ⁇ .
  • relevant nodes eg documents and monuments
  • Labeistring ⁇ it is usually not necessary to create a specially adapted result in the output data set; All documents and monuments of the output data set are simply read out.
  • a macro is then created, which replaces the record numbers in the raw adjacence list (, raw edge list y ) with the complete, read string ⁇ .
  • the selection of the relevant entries then results automatically from the record numbers existing in the raw adjacency list.
  • FIG. 10 A schematic of the basic procedure in the production of a matrix from the basic list is shown in FIG. 10 (detail from FIG. 7):
  • a node set is extracted from the base list for the sets of link output nodes and link destination nodes.
  • the classification criteria node set usually gives the columns of the matrix, the item nodeset the rows. Both nodesets are composed of the information that exists in the respective subgroup of the base list. is present. Primarily only the record number (ie the ID) of the corresponding nodes is necessary. All further information is used for later sorting of the matrix or for quick identification of the entries.
  • nodesets are extracted from the base list, existing redundancies are filtered out before insertion into the matrix, so that each classification criteria or object complex occurs only once in the corresponding nodeset. After filtering and possible pre-sorting, the nodesets are copied into an empty table (see Figure H).
  • the label ⁇ of the nodes contained in the Labeistring ⁇ may be split into different cells - accordingly
  • the edge set usually does not have to be extracted from the base list in the case of matrix production (in contrast to the classic network visualization).
  • the corresponding subgroup in the third column group (FIG. 8c) is sufficient despite its existing redundancies.
  • e denotes the corresponding edge column in the base list (eg, [baselist.xls] edges x ! $ AP: $ AP);
  • x and y are variables designating the respective output and destination nodes, respectively.
  • the link output node record number is in cell x (eg $ EU22), the link destination node record number in cell y (eg ET $ 20).
  • the spreadsheet may convert the dynamic values into fixed values, remove zeros, and assign appropriate conditional cell formatting (e.g., black background if cell content is not 0).
  • a finished matrix 5 is illustrated by way of example in FIG. 11.
  • the production of the image matrix 6 is technically divided into two sections.
  • Edgelabel ⁇ are created, which consist either of the respective link-out node, ie the document (part), or if the, edge-occurrence ⁇ has a value higher than one of several of them.
  • the second section after creating the Edgelabel, concerns the actual visualization of the image matrix.
  • the matrix is first sorted, filtered and if necessary transposed.
  • the actual image table is finally generated (FIG. 12).
  • FIG. 12 shows, by way of example, a detail of the image matrix, which in practice can be significantly larger and can comprise, for example, 200 columns and 2,000 rows.
  • Edgelabel Like the basic list, they are created only once for all possible edges. Alternatively, it would also be possible to create only the necessary edgelabel 'on the fly' at the time of visualization. The latter variant is when using the invention in a computer network, for. As on the Internet, advantage to limit the data processing effort on the processing of the currently desired information. It should be noted that the edgelabel ⁇ must be created separately for all node summaries of global, metalocal and local type. This is necessary because the edgeoccurence ⁇ of the same named edges can differ in the different summaries (see FIG.
  • the content of a cell ie an edge in the matrix, does not necessarily correspond to the direct link ratio between the respective summarized objects and classification criteria in the database. Rather, especially when global or metalent summarized Matrix rows or matrix columns, multiple edges in one edge.
  • the edge between folio and monument in FIG. 13 represents, for example, in a locally combined matrix (DocSelf matrix *) only the direct (also monovalent) link between folio and monument that also exists in the dataset.
  • DocEntity2 matrix ⁇ a metalokally aggregated matrix
  • the same edge between folio and monument represents a total of three links in the dataset: the Folio Monument link, as well as two more links between the quadrants and the monument parts.
  • the image matrix will be coded as HTML in the following.
  • the content of the, Edgelabels ⁇ is therefore defined as an HTML table cell:
  • EdgeAltText which may appear when you hover over it in the online version and a link (EdgeLink) that allows you to navigate back to the database.
  • The, EdgeLabel ⁇ corresponds to the label of the associated link output node, ie, for example, the document (part) s.
  • The, EdgeImage ⁇ corresponds to the respective image or that of the reprographic template (possibly indicated by a frame or the like).
  • EdgeAltText ⁇ contains any information from the respective Labeistring and for better control the name of the Edge (Recno $ Recno).
  • the 'EdgeLink' opens the link home node, ie document (part) in the database. Exceeds the Edge-Occurrence ⁇ the value 1, so a detailed matrix, which in turn contains the relevant link output node takes the place of the single link output node in the table cell ideally.
  • ⁇ the value 1
  • the overview image replaces the single image in the 'edgeimg' of the table cell. It must be specially created for all edges of multiple occurrences, for which purpose a concordance is created from the base list in which all record numbers of the link output nodes, as well as their image reference, are collected at the multiple occurring edge. From the concordance, one HTML file is then created for each edge. It carries the name of the edge and allows navigation from the individual contained nodes back to the output data set. Finally, the HTML version of the overview image is converted to an image file using a special tool (e.g., Html2jpg TM) for insertion into the HTML version of the image matrix.
  • a special tool e.g., Html2jpg TM
  • the finished 'Edgelabel' for values greater than 1 contains as name the original name of the edge in the form, recno $ recno ⁇ as, edgeimg ⁇ the overview image as well as, edgealttext ⁇ the value of, Edge-Occurence ⁇ of the edge and the, label ⁇ of the parent document complex.
  • the edge link ⁇ expediently does not directly refer to the output data set, since the corresponding edge does not always represent a real existing relationship but rather a summary of several such relationships. The link therefore expediently opens the interactive HTML version of the overview image, from which individual quadrants one can then navigate into the output data set. 6.3. Generation of the image matrix
  • the matrices are enriched with the Edgelabels with the help of several macros.
  • the first two macros replace the name of the edge in the matrix cell with the HTML table cell.
  • the third macro generates the HTML overview panels and the fourth generates the corresponding image files.
  • the enriched matrix may subsequently be extracted from the spreadsheet, e.g. with a good HTML editor (such as Adobe Dreamweaver TM) are imported into an HTML file and displayed in the browser as a picture matrix.
  • a good HTML editor such as Adobe Dreamweaver TM
  • the matrix can also be easily processed in an enriched form and quickly converted into an image matrix.
  • a matrix in which the documents are mapped only locally would, in the case of a dataset with 10 000 classification links, comprise up to 10 ⁇ 000 document lines and would therefore not be useful for direct human interaction. Moreover, in such a matrix, it would be impossible to create areas of meaningful density for an image matrix because a majority of the lines would typically contain only a single or very few filled cells. On the other hand, a matrix in which the documents are mapped globally prevents numerous detailed questions, since so many links are combined in many cells, that a meaningful comparison is prevented by too much density.
  • metalocal units i.e., for example, book / chapter / place instead of book / place.
  • the metalocal unit and thus the multi-level hierarchical subdivision of documents in humanities databases as a whole, finds its primary purpose here.
  • the metalocal unit encounters in the image matrix (as in any other form of display of the data) both the over-globalization, as well as the excessive local fragmentation.
  • the new metalocal units become accessible in the following generation of matrices (refresh).
  • Simultaneously visible image information in the image matrix allows the recognition and production of meaningful sorts or groupings (permutations) of several individual nodes and node complexes. It is possible to play with the more or less existing subjectivity of the subdivision of the classification criteria or the objects themselves. The roots of the possibly existing strict node trees are virtually cut off for this purpose. Information can be found in the
  • Sequence can be sorted differently and combined into alternative meaningful units. This creates new meaningful groupings of nodes, which are not necessarily oriented to the usual physical distribution of the represented objects (eg drawings of an artist from different collections, an assumed reconstruction project or the like).
  • the found groupings are first compiled by permutation in the (image) matrix.
  • the visual properties of the image matrix have an advantageous effect in the context of this process, especially in the case of manual permutation, since the sorting criterion is always in view.
  • Boundary lines mark the groupings together directly in the (image) matrix. Alternatively, however, they can also be firmly integrated into the output data set as cognitive concepts ⁇ (ie, for example, as virtual objects).
  • The, cognitive concepts ⁇ are deposited as a set of linked alias nodes and represent the newly found groupings as further use as virtual (image) documents. So they offer an alternative to the given physical order without destroying them.
  • Independent, cognitive concepts ⁇ according to this definition can also serve to deposit the assemblies discussed above in the output data set.
  • Randomness can be more easily excluded the stronger the correlation of the similar visual objects. All other causes are usually much harder to distinguish. Due to the contained image information, due to the two-dimensional matrix order and due to its permeability, the image matrix proves to be an extremely useful tool. Detected direct dependencies (historical events) or other more precisely specifiable relationship between two mappings can be stored in the image matrix, for example by drawing left-hand arrows (eg by 'clicking and dragging' with the mouse). With appropriate implementation, the image matrix can thus serve as a convenient user interface for processing the output data quantity.
  • object as in the example discussed above, are post-antique (picture) documents
  • classification criteria are anti-monuments
  • Art history is devoted to the time-bound phenomena of (image) documents based on the classification criteria
  • Classical archeology is dedicated to the time-bound phenomena of the classification criteria on the basis of the (picture) documents.
  • the image matrix serves the following purposes:
  • the image matrix serves to exclude dependent representations, since Here only the respective first representation of a series of copied representations of relevance is.
  • the exclusion criteria are direct dependencies, which only become apparent in the image matrix (see advantage 4).
  • Vagically dated nodes e.g., dated, 17th century *, or terminus ante or post
  • Vagically dated nodes may be more accurately ranked based on the image information present in the overview.
  • the image matrix facilitates data analysis and revision, for example, duplicate but differently named as well Unidentified objects can be detected by inspection and possibly geraergt or otherwise related. If the classification criteria are of a visual nature, corresponding candidates automatically accumulate in the same matrix columns or rows.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

La présente invention concerne un procédé pour la fabrication d'une matrice d'image (1), comprenant les étapes de mise à disposition d'un réseau avec une quantité de nœuds sources de liens (2) et une quantité de nœuds cibles de liens (3) avec des liens intercalés (4), la formation d'une matrice (5) avec des lignes et des colonnes, les nœuds sources de liens étant associés aux lignes et les nœuds cibles de liens aux colonnes ou inversement, et placement de représentations visuelles (6) des nœuds sources de liens (2) ou des nœuds cibles de liens (3) à la place des liens dans la matrice de sorte qu'il en résulte la matrice d'image (1).
EP07786557A 2006-08-07 2007-08-03 Procédé pour la fabrication de matrices d'images extensibles Withdrawn EP2050022A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE102006036826 2006-08-07
PCT/EP2007/006900 WO2008017430A1 (fr) 2006-08-07 2007-08-03 Procédé pour la fabrication de matrices d'images extensibles

Publications (1)

Publication Number Publication Date
EP2050022A1 true EP2050022A1 (fr) 2009-04-22

Family

ID=38596892

Family Applications (1)

Application Number Title Priority Date Filing Date
EP07786557A Withdrawn EP2050022A1 (fr) 2006-08-07 2007-08-03 Procédé pour la fabrication de matrices d'images extensibles

Country Status (3)

Country Link
US (1) US20100194756A1 (fr)
EP (1) EP2050022A1 (fr)
WO (1) WO2008017430A1 (fr)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10437999B1 (en) * 2016-08-31 2019-10-08 Symantec Corporation Runtime malware detection
CN109614613B (zh) * 2018-11-30 2020-07-31 北京市商汤科技开发有限公司 图像的描述语句定位方法及装置、电子设备和存储介质

Family Cites Families (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5659626A (en) * 1994-10-20 1997-08-19 Calspan Corporation Fingerprint identification system
US6222939B1 (en) * 1996-06-25 2001-04-24 Eyematic Interfaces, Inc. Labeled bunch graphs for image analysis
US7065253B2 (en) * 1999-09-03 2006-06-20 Intel Corporation Wavelet zerotree coding of ordered bits
US6304898B1 (en) * 1999-10-13 2001-10-16 Datahouse, Inc. Method and system for creating and sending graphical email
WO2001031502A1 (fr) * 1999-10-27 2001-05-03 Fujitsu Limited Dispositif et procede de classement et de rangement d'informations multimedia
US6675158B1 (en) * 1999-11-30 2004-01-06 Sony Corporation Method and apparatus for organizing data pertaining to audiovisual content
US7127087B2 (en) * 2000-03-27 2006-10-24 Microsoft Corporation Pose-invariant face recognition system and process
US7418470B2 (en) * 2000-06-26 2008-08-26 Massively Parallel Technologies, Inc. Parallel processing systems and method
KR20040041082A (ko) * 2000-07-24 2004-05-13 비브콤 인코포레이티드 멀티미디어 북마크와 비디오의 가상 편집을 위한 시스템및 방법
US7224991B1 (en) * 2000-09-12 2007-05-29 At&T Corp. Method and system for handwritten electronic messaging
US7003308B1 (en) * 2000-09-12 2006-02-21 At&T Corp. Method and system for handwritten electronic messaging
US7176921B2 (en) * 2000-10-20 2007-02-13 Sony Corporation Graphical rewriting system for multimedia descriptions
TW569148B (en) * 2002-04-09 2004-01-01 Ind Tech Res Inst Method for locating facial features in an image
US8321470B2 (en) * 2003-06-20 2012-11-27 International Business Machines Corporation Heterogeneous multi-level extendable indexing for general purpose annotation systems
US7711584B2 (en) * 2003-09-04 2010-05-04 Hartford Fire Insurance Company System for reducing the risk associated with an insured building structure through the incorporation of selected technologies
US8503825B2 (en) * 2004-05-06 2013-08-06 Christopher L. Kavanau Method and system for aligning and classifying images
US20050278656A1 (en) * 2004-06-10 2005-12-15 Microsoft Corporation User control for dynamically adjusting the scope of a data set
US8878851B2 (en) * 2004-11-12 2014-11-04 Synchronica Plc Method and system for streaming documents, e-mail attachments and maps to wireless devices
US20060233461A1 (en) * 2005-04-19 2006-10-19 Honeywell International Inc. Systems and methods for transforming 2d image domain data into a 3d dense range map
US20070153731A1 (en) * 2006-01-05 2007-07-05 Nadav Fine Varying size coefficients in a wireless local area network return channel
US8243022B2 (en) * 2006-11-27 2012-08-14 Microsoft Corporation Federated virtual graffiti
US8019167B2 (en) * 2007-01-03 2011-09-13 Human Monitoring Ltd. Compressing high resolution images in a low resolution video
US10235429B2 (en) * 2007-08-20 2019-03-19 Stephen W. Meehan System and method for organizing data in a dynamic user-customizable interface for search and display

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See references of WO2008017430A1 *

Also Published As

Publication number Publication date
WO2008017430A1 (fr) 2008-02-14
US20100194756A1 (en) 2010-08-05

Similar Documents

Publication Publication Date Title
DE3650417T2 (de) Informationsaufzeichnungs- und Wiederauffindungssystem.
EP0910829B1 (fr) Systeme de banque de donnees
DE69418474T2 (de) Semantisches objektmodellierungssystem und verfahren um relationelle datenbankschemata herzustellen
DE69622350T2 (de) Ein Geschichtsanzeigegerät
DE69432575T2 (de) Dokumentenerkennungssystem mit verbesserter Wirksamkeit der Dokumentenerkennung
EP1311989B1 (fr) Procede de recherche automatique
DE112010000947T5 (de) Verfahren zur völlig modifizierbaren Framework-Datenverteilung im Data-Warehouse unter Berücksichtigung der vorläufigen etymologischen Separation der genannten Daten
DE112005003157T5 (de) Domainspezifisches Datenelement-Mappingverfahren- und System
DE112020001874T5 (de) Datenextraktionssystem
DE202011110873U1 (de) Skalierbare Wiedergabe von großen räumlichen Datenbanken
DE60310881T2 (de) Methode und Benutzerschnittstelle für das Bilden einer Darstellung von Daten mit Meta-morphing
DE112018001165T5 (de) Automatisierte dokumentenanalyse für unterschiedliche natürliche sprachen
DE69632835T2 (de) Verfahren zur automatischen verarbeitung von information über benutzerdaten
DE112018002626T5 (de) Verfahren und Systeme zur optimierten visuellen Zusammenfassung von Sequenzen mit zeitbezogenen Ereignisdaten
WO2008017430A1 (fr) Procédé pour la fabrication de matrices d'images extensibles
DE19908204A1 (de) Fraktales Netz n-ter Ordnung zum Behandeln komplexer Strukturen
WO2011044865A1 (fr) Procédé de détermination d'une similarité d'objets
DE602005006086T2 (de) Verfahren zum Ausführen einer Modularisierung eines Hypertextes
Rehbein et al. Codicology and Palaeography in the Digital Age
DE102009042659A1 (de) Verfahren zur automatisierten Katalogisierung von digitalen Rasterdaten mit räumlichem Bezug
DE10057634A1 (de) Verfahren zur Verarbeitung von Text in einer Rechnereinheit und Rechnereinheit
DE69122324T2 (de) Verfahren und gerät zur graphischen befragung einer datenbank
DE19729911A1 (de) System zur Verbesserung der Organisation von Daten einer Dokumentation
WO2013075745A1 (fr) Procédé et système d'élaboration de modèles d'utilisateurs
Eckl et al. Ten years of Soziale Passagen: An empirical analysis of the journal’s topics

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20090112

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU LV MC MT NL PL PT RO SE SI SK TR

AX Request for extension of the european patent

Extension state: AL BA HR MK RS

17Q First examination report despatched

Effective date: 20100305

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20100916