 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an interactive drawing recognition processing method which performs a vector producing process for each pattern element, which is included in image data, for example, in order to generate a data base of an
existing drawing.
2. Description of the Related Art
To construct a data base of existing drawings such as maps and diagrams of machines, there have been proposed various drawing recognition processing methods. FIG. 1 shows an example of the processing procedure of the interactive drawing
recognition processing method according to the prior art. In the interactive drawing recognition processing method, a drawing is read by a scanner to create raster data (run length data) basically in the run length form (step 52). The run length data
is displayed on a screen of a display. Next, to vectorize or transform each pattern or drawing elements into vectors, there is conducted, for example, a core or center line production process. In this case, when the operator specifies by a position
input device or a pointing device a domain or region of the drawing to be recognized on the screen, the run length data contained in the specified domain is processed such that centers of the run lengths are connected to each other to produce a center
line (step 54). Interactively conducting the core production process for the other drawing elements, there is attained vector data by performing a modification of the drawing and so forth (step 56). Thereafter, a character recognition is accomplished
for character strings included in the drawing and a structure forming process is carried out, for example, to assign attributes to the pattern and to establish relationships between patterns, thereby producing a structured data base.
Moreover, according to a conventional line type recognition method, existing drawings such as maps and diagrams of apparatuses are first inputted to the recognition system to attain binary image data and then a vectorizing process is conducted
for the binary image data. Thereafter, when a predetermined pattern is specified, a retrieval is effected with vector data for a series of patterns to be continued from the the specified pattern according to a predetermined program. A check is then
made to decide whether or not the attained series of patterns repeatedly includes the predetermined pattern, thereby recognizing the line type thereof.
Additionally, there has been conventionally proposed various map input systems in which a map is read, pattern elements thereof are converted into vector data, and patterns of houses are recognized to be assigned with codes. For example, in an
interactive map data input system of the prior art, the map is first read by a scanner to produce binary image data. Subsequently, the operator picks by a pointing device each pattern element to transform the element into vector data. Furthermore, the
operator picks each of the house patterns such that the the system recognizes the house pattern and then creates a code for the pattern. Thereafter, symbols included in the map are recognized and the structure forming process is carried out, for
example, to decide attributes of the patterns and to relate the patterns to each other, thereby attaining a structured data base.
However, in the interactive drawing recognition processing method of the prior art, when producing vectors from the respective pattern elements, the run length data is directly processed to generate center lines. Consequently, in the center line
generating process or a medial line extraction, it is required to carry out several processes such as decisions of contours and positions of run lengths. Therefore, this leads to a problem that a long period of time is required to completely accomplish
the center line generation for the pattern after the operator specifies the pattern to be recognized.
In addition, according to the conventional line type recognition method, the line type of a series of patterns is recognized at a time according to the vector data of the overall drawing and hence there exists a problem of the insufficient
processing speed in the recognition. Furthermore, since the line type is recognized according to the program, any line type other than those taken into consideration in the program cannot be coped with. This causes a problem in that only a preset range
of line types can be recognized.
Moreover, according to the interactive map input system of the prior art, in the process to recognize the house patterns, the operator specifies the house patterns one by one, leading to a problem that a considerably heavy load is imposed on the
operator and the process can be achieved only at a low speed.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide an interactive drawing recognition processing method capable of decreasing a period of work time from when a pattern to be recognized is specified to when the center line generation process is
completely finished for the pattern.
Another object of the present invention is to provide a line type recognition method for use in a drawing input apparatus in which the line recognition process can be simply carried out and the range of recognizable line types can be easily
extended.
Further another object of the present invention is to provide a house recognition method in a map input system in which the process can be accomplished at a high speed while mitigating the work load imposed on the operator.
An interactive drawing recognition processing method according to the present invention comprises the steps of: reading in advance a drawing as a processing object; generating a contour line data by obtaining a contour line for each pattern
element which is attained from a labeling process for an obtained image data, and a region data which manages in a tree structure regional information related with a region indicating in which region each line segment is contained in the overall drawing
for each line segment which constitutes a contour line. Next when a region containing a pair of line segments constituting the contour line is specified, the region data associated with the specified regions is retrieved and extracting the contour line
data corresponding to the region data which is obtained by the retrieving, and then generating a central line of the specified pair of line segments according to the extracted contour line data.
An interactive drawing recognition processing method according to the present invention, further comprises, after generating the central line of the pair of line segments, the steps of: tracing another pair of line segments, which is connected to
the pair of line segments, according to the contour line data; and generating a center line of the traced another pair of line segments.
In an interactive drawing recognition processing method according to the present invention, the contour line data is undergoes a polygonal approximation process.
A line type recognizing method according to the present invention, comprises the steps of: memorizing in advance a line type pattern data in which a reference feature is registered for each constituent element which constitutes a line type
pattern of each line type; generating a contour line data by taking out a contour line for each pattern according to a binary image data obtained by reading a drawing; identifying, when a line type of a objective line to be recognized on a screen and a
position of a constituent element of the objective line, a type of the constituent element of the objective line according to the contour line data; and extracting a feature for the constituent element of the objective line, and then deciding whether or
not the objective line is the same as the specified line type by comparing the extracted feature with the reference feature corresponding to the objective line.
A house recognizing method according to the present invention, which comprises a step of reading in advance a map as a processing object and a step of recognizing a house pattern after converting an obtained binary image data to a vector data,
further comprises the steps of: extracting, when a frame is specified on a screen, in the frame as a recognition objective house pattern, a pattern which has a closed-loop contour and a size not less than a predetermined size; and indicating whether or
not the recognition objective house pattern is registered, after shaping the extracted recognition objective house pattern into a house contour and displaying the house contour on the screen.
A house recognizing method according to the present invention, further comprises, after shaping the recognition objective house pattern into the house contour, a step of performing, when an edge of the recognition objective house pattern is
shifted from an edge of an already registered house pattern, a connection process of removing the shift between the edges, the edges being inherently overlapped with each other.
According to the interactive drawing recognition method of the present invention, there are produced in advance contour line data of contour lines attained from each of the pattern element and region including information related to regions to be
managed in a tree structure, the information indicating regions in the overall drawing in which the respective line segments constituting the contour lines are included. Thanks to the provision, when a region containing a pair of line segments included
in a contour line is specified, region data corresponding to the specified region can be efficiently retrieved. Consequently, contour line data associated with the region data obtained as a result of the retrieval can be extracted at a high speed to
achieve the center line generation.
Moreover, according to the interactive drawing recognition processing method of the present invention, after the center line is created for a pair of line segments, it is possible to trace another pair of line segments linked with the pertinent
pairs of line segments according to only the contour line data. Therefore, the center line generation of the traced pair of line segments can be automatically conducted.
In accordance with the line type recognition method of the present invention, for each line type, there is stored line type pattern data in which a reference feature is registered for each constituent element of the line type pattern such that,
according to binary image data attained by reading a drawing, contour lines are obtained for each pattern of the drawing to produce contour line data. When a line type of a line to be recognized on the screen and a position of a constituent element of
the line are specified, the line type of the constituent element is identified according to the line contour data and there is extracted a feature for the constituent element. Thereafter, the extracted feature is compared with the reference feature
corresponding to the line to recognize the line type for each constituent element. When compared with the conventional case in which the line type is recognized for a series of patterns at a time according to the vector data of the overall drawing, the
line type recognition can be more simply accomplished. Additionally, for example, when a new line type is included in a map, a reference feature of each constituent element constituting a new line type pattern of the new line type can be easily added to
the existing line type pattern data. When a map having the same contents as another is represented in a different scale, the reference feature of each line type beforehand registered to the line type pattern data can be easily altered to arbitrarily
expand the range of recognizable line types.
According to the house recognition method of the present invention, when a frame is specified on a screen image, a pattern which is in a closed loop having at least a predetermined size in the frame is extracted, shaped, and displayed on the
screen. Therefore, after specifying house patterns for recognition at a time, the operator is only required to indicate whether or not each of the specified image is to be registered according to the result displayed on the screen. This consequently
reduces the work load on the operator and increases the processing speed. In addition, since the house patterns to be recognized can be specified at a time in the frame, operability is improved when compared with the conventional interactive processing
method.
Furthermore, according to the house recognition method of the present invention, after the house patterns of the recognition objects are shaped into house contours or shapes, a necessary correction or modification can be achieved for the house
patterns when a connection process is conducted for the house patterns of the recognition objects by executing a connection process for the obtained house contours.
The interactive drawing recognition processing apparatus according to the present invention includes an image input device for reading a drawing as a processing object and thereby attaining raster data, a pointing device for indicating
coordinates of a position of a recognition objective pattern in the drawing, a central processor for executing a labeling process for the raster data produced from the image input device, and a feature extracting process of extracting a feature for each
pattern element labeled and attaining element data. A contour line extracting process of extracting edges of the raster data for each of the pattern elements and obtaining contour line data, and region data creating process of generating region data to
manage for each line segment of the contour line in a tree structure, the information being related to a circumscribed rectangle of the line segment are also provided.
The apparatus further includes an element file for storing the element data created by the central processor, contour line file for storing the contour line data generated by the central processor, a region file for for storing the region data
produced by the central processor and a vector file. The central processor references, when a recognition objective pattern is specified from the pointing device, the region file to detect the region data associated with the specified recognition
objective pattern, references the contour line file to extract the contour line data related to the detected region data, conducts a center line creation for each of the line segments of the extracted contour line to generate a center line, creates
vector data according to the resultant center line, and stores the created vector data in the vector file.
Additionally, the interactive drawing recognition processing apparatus of the present invention further includes a line type pattern file for storing line type pattern data for each line type in which a predetermined feature is registered for
each of the constituent elements obtained by dividing a line type pattern of the line type. After the vector data is stored in the vector file, the processor references, when a line type of the recognition objective line to be recognized and a position
of a constituent element of the recognition objective line are specified from the pointing device, the contour line file to identify a type of the constituent element according to the contour line data, and extracts a feature of the constituent element.
The processor then references the line type pattern file to compare the extracted feature with the reference feature corresponding to the recognition objective line, and judges to determine whether or not the recognition objective line is of the
specified line type.
Furthermore, in the interactive drawing recognition processing apparatus according to the present invention, after the vector data is stored in the vector file, the central processor extracts, when a frame is specified in the drawing by the
pointing device, as a recognition objective house pattern having a closed-loop contour and a size not less than a predetermined size in the frame specified from the pointing device and shapes the extracted recognition objective house pattern into a
predetermined house contour.
BRIEF DESCRIPTION OF THE DRAWINGS
These and other objects and advantages of the present invention will become apparent by reference to the following description and accompanying drawings wherein:
FIG. 1 is a diagram showing the procedure of the center line generation in the interactive drawing recognition processing method of the prior art;
FIG. 2 is a diagram schematically showing the construction of a computer aided design (CAD) system utilizing an embodiment of the interactive drawing recognition processing apparatus according to the present invention;
FIGS. 3A to 3D are diagrams for explaining the structure of region data in the interactive drawing recognition processing method according to the present invention;
FIG. 4 is a diagram showing the procedure of the center line generation in the interactive drawing recognition processing method according to the present invention;
FIGS. 5A to 5D are diagrams for specifically explaining the center line generation in the interactive drawing recognition processing method according to the present invention;
FIG. 6 is a diagram for explaining line types as recognition objects in an embodiment of the line type recognition method according to the present invention;
FIG. 7 is a diagram for explaining the structure of a line type pattern file of FIG. 2;
FIG. 8 is a diagram for explaining specifications of features of the respective constituent elements of the line patterns in the line type pattern file;
FIGS. 9A to 9D are diagrams for specifically explaining constituent elements of line type patterns;
FIG. 10 is flowchart for explaining the processing procedure in an embodiment of the line type recognition method according to the present invention;
FIG. 11 is a diagram showing a menu screen image displayed to achieve the line type recognition;
FIGS. 12A to 12C are diagrams for concretely explaining the contents of the line type recognition;
FIGS. 13A and 13B are diagrams for specifically explaining the contents of the line type recognition;
FIG. 14 is a diagram for explaining the processing procedure to register symbol features;
FIGS. 15A and 15B are diagrams for explaining a relative positional relationship when the symbol includes a plurality of constituent elements;
FIG. 16 is a flowchart for explaining the processing procedure to recognize a symbol;
FIG. 17 is a flowchart for explaining the processing procedure to interactively edit the result obtained by recognizing the symbol;
FIG. 18 is a flowchart for explaining the processing procedure to update the symbol feature;
FIG. 19 is a flowchart for explaining the processing procedure to update weight coefficients;
FIG. 20 is a flowchart for explaining the processing procedure to update weight coefficients;
FIGS. 21A to 21D are diagrams for specifically explaining the center line generation in an embodiment of the house recognition method according to the present invention;
FIG. 22 is a flowchart for explaining an embodiment of the house recognition method in a map input system according to the present invention; and
FIGS. 23A to 23C are diagrams for explaining the element connecting process .
DESCRIPTION OF THE PREFERRED EMBODIMENTS
(Interactive Drawing Recognition Processing Method and Apparatus Thereof according to the Present Invention)
As shown in FIG. 2, a computer aided design (CAD) system as an embodiment of the interactive drawing recognition processing apparatus of the present invention includes a scanner 12 as an image input device, a cathode-ray tube (CRT) display 14, a
pointing device 16, a central processor 18, an element file 22, a contour line file 24, a region file 26, a vector file 28, a structured file 32, a line type pattern file 34, a symbol knowledge base 36, a matching file 38, a link file 40, and an edited
result storage 42. Incidentally, the CAD system is employed, for example, to produce a data base of existing drawings such as maps and diagrams of machines.
The scanner 12 optically scans a drawing to produce binary image data as inputs to the system. The CRT display 14 displays thereon the inputted result.
In this system, the binary image data inputted from the scanner 12 is raster data basically in the run length format used in facsimiles or the like. The raster data is vector data of which the start point is indicated by the coordinates at which
the pixel value starts changing, for example, from "0 (white)" to "1 (black)" on a scan line and the finish point is represented by the coordinates at which the pixel value starts changing, for example, from "1" to "0" on the scan line.
The pointing device 16 conducts the following operations and includes a keyboard, a mouse, and the like.
(a) To indicate a recognition target pattern, data of coordinates indicating positions of the recognition target pattern is inputted.
(b) When achieving the line type recognition, a line type to be recognized and/or a start position of the recognition are/is specified.
(c) To register a symbol, a frame is specified on the screen of the CRT display 14 by the mouse and then the symbol is selected in the frame. (d) When conducting the house recognition, a frame is specified on the screen of the CRT display 14 by
the mouse and then patterns contained in the frame are selected at a time.
The element file 22 is used to store therein features extracted for each pattern element data (to be simply referred to as "element data" herebelow) obtained by labeling the raster data. In this case, the features of the element data include the
barycenter, area, circumscribed rectangle, moment, etc. In the contour line file 24, there is stored data of contour lines attained by extracting edges of raster data for each element data. The contour line is approximated by a polygon and includes many
line segments. One contour line forms a closed loop including broken or polygonal lines. Therefore, the contour line data is represented as data for each contour line, the data being attained by transforming line segments constituting the line into
vectors. In this regard, the element data are linked by the contour line data and file pointers or by address pointers on the memory.
Stored in the region file 26 is region data indicating in which region of the overall drawing each line segment constituting the contour line is contained. The region data is managed in a tree structure including a root, a branch node, and a
leaf. In the data management according to the tree structure, each node represents a region at an intermediate point of subdivision of a drawing and a leaf designates a region as the minimum division unit. A pointer to contour line data contained in a
region represented by a leaf is related to the leaf and is stored in a main memory 18, of the processor 18. Consequently, when a region is specified, the system can retrieve a leaf corresponding to the specified region to decide line segments of a
contour line contained in the region according to pointers related to the retrieved leaf. Description will be given of the operation related to, for example, the tree structure of region data related to a pattern having the shape of substantially a
cross. FIG. 3A shows a contour line including edges of the raster data for the cross pattern. The line segments constituting the contour line are managed for each circumscribed rectangle associated therewith (reference is to be made to FIG. 3B). The
overall drawing is split into regions each of which includes one circumscribed rectangle of each of the line segments, as shown in FIG. 3C. As can be seen from FIG. 3D, the process of subdivision is represented in the form of a region management tree
the leaves of which correspond to the minimum partition regions.
The symbol knowledge base 36 is provided to store therein symbol knowledge representing features of symbols to be registered. Stored in the matching file 38 are any symbols (or constituent elements) matching an isolated raster. Additionally,
the link file 40 is disposed to store therein link data for isolated raster data, the link data denoting a grid point nearest thereto when the screen is subdivided according to the grid unit. In the edited result storage 42, there is stored such
information items attained when the operator edits the recognition result, as symbols for which the recognition has been attempted, symbols successfully recognized, symbols for which the recognition has been failed, and symbols erroneously recognized.
The information items stored in the edited result storage 42 are utilized in the update of features and the like.
Particularly, the embodiment employs a region multi-dimensional (R-MD) tree to manage region data. The R-MD tree is a multi-dimensional tree in which information related to regions is converted into coordinates to increase the data retrieving
speed. Concretely, according to the R-MD tree, a two-dimensional circumscribed rectangle is expressed by a central position thereof (x.sub.c,y.sub.c) and lengths x.sub.w and y.sub.w thereof in x and y directions, respectively. The set of four values
(x.sub.c,y.sub.c,x.sub.w,y.sub.w) characteristic to the circumscribed rectangle are assumed to indicate a point in a four-dimensional coordinate system. A predetermined coordinate conversion is further conducted for the obtained point such that the
resultant point is managed in the form of a multi-dimensional tree in the new four-dimensional coordinate system.
The central processor 18 executes predetermined processes according to an interactive drawing input program 18.sub.2 stored therein. For example, a labeling process is conducted as a pre-processing according to the input binary image data
(raster data); a feature extraction process and a contour extraction process are carried out to create element data, contour line data, and region data. Moreover, a center line generation is interactively accomplished for a recognition target pattern to
produce vectors according to raster data. The vector file 28 is arranged to store therein vector data resultant from the vector producing process. Additionally, the central processor 18 executes a structure decision process including an attribute
assigning process to assign meanings and attributes to patterns and a relationship establishing process to define semantic relationships between the patterns. The structured data is stored in the structured file 32, which is used as a CAD data base.
In addition, the processor 18 executes a symbol recognition process and a process in which symbols resultant from the recognition are interactively edited on the screen and features and the like of the registered symbols are updated according to
the edited results. The update process is executed by depressing "Update" button displayed in the screen of the CRT display 14.
Furthermore, the processor 18 performs a house recognition process after the vector generation to recognize house patterns to produce codes associated therewith. In this process, a shaping process and a connection process may also be effected.
The shaping process is conducted for the following reason. In a stage in which the house patterns are simply transformed into vectors, there may exist a case in which one edge thereof is represented with a plurality of polygonal lines, which are to be
shaped into one line. Moreover, a corner of the house patterns may possibly have been rounded or chamfered. The rounded portion is to be shaped into the original corner. The connection process is executed in the following case. Between a house
pattern as the recognition target and a house pattern beforehand registered to the system, when a side of the former is shifted from the associated side of the latter, the positional shift is to be removed by the process.
Subsequently, the interactive drawing recognition processing method of the embodiment will be described with reference to FIG. 4 and FIGS. 5A to 5D. FIG. 4 is a flowchart for explaining a procedure of producing center lines for a recognition
target pattern according to the method, whereas FIGS. 5A to 5D are diagrams for specifically explaining the center line generation.
First, the drawing is read by the scanner 12 to attain raster data (step 2 of FIG. 4). There is obtained, for example, raster data of a cross-shaped pattern as shown in FIG. 5A. The central processor 18 executes a labeling process for the
raster data, extracts therefrom features of the respective pattern elements thus labeled to produce element data, and stores the data in the element file 22. Additionally, the processor 18 obtains edges from raster data of each pattern element to store
the data of a contour line in the contour line file 24 (step 4). Next, according to the contour line data, the processor 18 creates region data for each line segment of the contour line to manage information related to the circumscribed rectangle in the
tree structure (step 6). In this connection, the processor 18 automatically executes the operations up to this point after the drawing is read by the scanner 12. The contour line and circumscribed rectangle may be visible or invisible to the user.
Next, the center line creation is interactively carried out for the target pattern. The center line creation in this case means a process to determine a line, for example, connecting the centers of two circles respectively inscribed on two
lines. First, the operator picks by the pointing device 16 a predetermined region (the aperture region of a cursor) in the screen of the CRT display 14 to specify a pair of line segments for the recognition thereof (step 8). In this situation, as can
be seen from FIG. 5A, the operator picks a predetermined region R.sub.1 and then designates two substantially parallel lines S.sub.1 and S.sub.2 on the left-hand side of the cross-shaped pattern. In response thereto, there is conducted a search through
the region management tree to detect a leaf corresponding to the region R.sub.1 designated in screen (step 12 of FIG. 5C). According to a pointer stored in association with the leaf, the process extracts from the contour line file 24 contour line data
corresponding to line segments S.sub.1 and S.sub.2 (step 14). Next, as shown in FIG. 5D, the process conducts an operation to generate a center line for the extracted two segments S.sub.1 and S.sub.2 to attain the center line (step 16). After the
center line is produced for the line segments according to the specified region, the system achieves a trace process for the two line segments to continuously generate center lines. In short, since the contour line itself is one closed line, the system
directly accesses the contour line file 24 to automatically attain line segments linked to the line segments S.sub.1 and S.sub.2 previously undergone the center line creation, the segments S.sub.1 and S.sub.2 being extended in a predetermined direction,
e.g., toward the right in FIG. 5A. The obtained center lines form a polygonal line (step 18). In this process, when the target point reaches, for example, a crosspoint of the cross-shaped pattern, the center line cannot be generated and hence the
center line creation is stopped. Thereafter, when the operator specifies by the pointing device 16 a pair of line segments as another recognition object, e.g., two line segments S3 and S4 substantially parallel to each other on the upper portion of the
cross-shaped pattern, the center line creation is similarly carried out for these segments.
In the operation, when the center line creation is stopped, the operator may freely edit or modify the center lines generated up to the point. For example, during the center line creation for a hand-written drawing, the operator may correct a
position of a line, transform a curved line into a straight line, or modify a line generated by mistake. Incidentally, in the modification, only the vector data attained by the center line creation is modified. Namely, the original raster data is kept
unchanged.
After the center line creation is performed for the contour line of a desired pattern element as above, a predetermined process is executed for the data resultant from the center line creation. For example, a plurality of polygonal lines are
approximately transformed into a line. In a case where the image includes, for example, level lines related to data of a map, the lines are smoothed in the process. The vector data attained according to the data resultant from the center line creation
is stored in the vector file 28 (step 22). Thereafter, a symbol recognition process is conducted for symbols contained in the drawing. In addition, a structure decision process is interactively carried out such that the attained structure data is
stored in the structured file 32. Incidentally, although the symbol recognition process requires the element data and contour line data, the contour line data is linked with the element data. Consequently, the contour line data is obtained via the
element data in the process.
According to the interactive drawing recognition processing method of the embodiment, there are produced contour line data by obtaining edges for the respective pattern elements according to the raster data thereof and region data in which
information related to a circumscribed rectangle of line segments constituting the contour line is managed according to a tree structure such that the contour line data is supervised by the region data. Thanks to this provision, when a region including
a pair of line segments is specified as the recognition target, the system can efficiently retrieve region data associated with the designated region to extract contour line data corresponding to the retrieved region data at a high speed so as to achieve
the center line creation for the specified region. Therefore, the period of time lapsed from when a pair of line segments are specified for the recognition target to when the center line generation is finished for the line segments can be minimized. In
addition, after the center line is produced for the pair of line segments, another pair of line segments linked with the line segments previously processed can be traced at a high speed only according to the contour line data without using the region
data. This makes it possible to automatically execute the center line creation for the traced line segments.
Although the tree structure is a region multi-dimensional (R-MD) tree in this embodiment, there may be used a k-d tree, or the like.
According to the interactive drawing recognition processing method described above, there are created in advance the contour line data by attaining a contour line for each image element and the region data including, for each line segment
constituting the contour line, information indicating a region in the overall drawing in which the line segment exists, the information being managed according to a tree structure. Due to the contour line data and the region data, when a region
including a pair of line segments constituting a contour line is specified, region data associated with the designated region can be effectively retrieved and hence contour line data corresponding to the retrieved region data can be obtained at a high
speed to achieve the center line creation for the specified region. Therefore, the period of time necessary for the operation can be minimized.
Moreover, after the center line is produced for the pair of line segments, another pair of line segments connected to the line segments previously processed can be traced at a high speed only according to the contour line data. Therefore, it is
possible to automatically execute the center line creation for the traced line segments. (Line Type Recognition Method of the Present Invention)
Description will be next given of an embodiment of the line type recognition method according to the present invention. The line type recognition method of the embodiment is employed to recognize types (line types) of lines drawn in a map and
can be achieved in a CAD system shown in FIG. 2.
According to "Kokudo Kihonzu Zushiki Do Tekiyo Kitei (Fundamental Land Diagrams and Application Rules Thereof)" published from the Geographical Survey Institute of the Ministry of Construction of Japan, there are stipulated many line types to be
adopted in maps. In the description of the present invention, the line types of recognition targets are particularly limited to the railways, boundaries, level lines, etc. as shown in FIG. 6. The line types representing railways include the Japan
Railways (JR), private railway, subway, cableway, and special railroad. The line types expressing boundaries include the boundary for To, Fu, and prefectures, boundary for branch offices of Hokkaido, boundary for district, city, and ward of Tokyo,
boundary for vegetation, boundary for arable land, boundary for sections, and boundary for towns, villages, and designated towns. In addition, the boundaries standing for level lines include the auxiliary curved line, special auxiliary curved line,
ground depression, and the like. Moreover, the line types as recognition objects include the footpath, boundary, steep slope/relative height, and precipice (small). Each of these line types repetitiously includes a predetermined pattern, which includes
a combination of constituent elements such as a line segment, a node or branch point, a crosspoint, and an isolated point. Therefore, in this embodiment, the line type is recognized by achieving comparison of predetermined features for each constituent
element.
In the line type pattern file 34 shown in FIG. 2 is stored line type pattern data obtained for each line type by subdividing a line type pattern thereof into a set of elements, the data including a predetermined reference feature of each of the
elements. In this embodiment, there are five kinds of constituent elements used, such as the line segment (ordinary line segment), bold line segment, node, crosspoint, and isolated point. The bold line segment is included for the recognition of JR.
The features, respectively, of the line segment and bold line segment are represented by the length of line segment. Additionally, an isolated point has an area in an actual map and hence the diameter thereof is used as the feature thereof. The feature
of a node is represented by the number of branches from the node, the branches being associated with a line entering the node. For a crosspoint, the number of lines extending or outgoing therefrom (number of outgoing lines) is used as the feature.
The layout of the line pattern file 34 will be described by referring to FIG. 7. In this diagram, the first field "linefontmember" is disposed to store therein the number of line types defined in the file 34. In this example, the number of
definitions of line types is set to 20 as shown in FIG. 6. Thereafter, the features of each line type pattern are described in fields enclosed by [LINEFONT] and [END]. In the fields representing the features, the name of line type is set to a field
"name", the number of elements constituting the pattern is described in a field "count", and a feature of each element is specified in a field "feature". Two numeric values are set to each of the fields "feature" in which the first value express the
type of the element and the second value specifies the feature. The values designating element types are as shown in FIG. 8 in which "0" to "4" are used to represent the ordinary line segment, bold line segment, isolated point, node or branch point, and
crosspoint, respectively. Additionally, the values specifying the features are set to express those related to the elements, respectively.
Description will now be specifically given of the contents of the line pattern file 34 in association with several line types. In the first example, line types to represent boundaries for To (of Tokyo To), Fu (e.g., Kyoto Fu), and prefectures.
As can be seen from FIG. 9A, the line type pattern includes combinations of such four elements as isolation point A.sub.1, node A.sub.2, line segment A.sub.3, and node A.sub.4. Isolation point A.sub.1 has a diameter of 0.3 mm, line segment A.sub.3
possesses a length of 2.5 mm, and nodes A.sub.2 and A.sub.4 each have two branches. Therefore, a value "4" is set to the "count" field. Two values "2,0.3" are set to the first "feature" field as features of isolated point A.sub.1. Set to the second
"feature" field are two values "3,2" indicating features of node A.sub.2. In the third "feature" field, there are set two values "0,2.5" as features of line segment A.sub.3. Features of node A.sub.4 are set as values "3,2" to the "fourth" feature
field.
Description will next be given of a second example related to line types representing dikes. The line type pattern is formed by combining two elements including line segment B.sub.1 and crosspoint B.sub.2 as shown in FIG. 9B. The | | |
