JP2892994B2 - Architectural drawing recognition method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for automatically recognizing the meaning of symbols written in drawings and the attributes of rooms from handwritten or printed architectural drawings input via an image scanner or the like. About. INDUSTRIAL APPLICABILITY The present invention can be used as automatic input means to various building-related systems for the purpose of digitizing drawings (CAD input), estimating member costs, and the like.

[0002]

2. Description of the Related Art When automatically recognizing a handwritten or printed architectural drawing, a technique of converting a line segment constituting the drawing into vector data has been used. However, there was no technology for recognizing the meaning of symbols such as walls, doors, toilets, and bathtubs, and no technology for recognizing room attributes such as toilets and bathrooms with these symbols.

For this reason, in the prior art for automatically vectorizing, after automatic vectorization, it is indispensable for the user to select each symbol again and give its meaning to all symbols. The effect of reducing input costs was insufficient.

[0004]

SUMMARY OF THE INVENTION The present invention solves the above-mentioned problem that drawing input takes a lot of time even if conventional drawing recognition technology is used. An object of the present invention is to provide a new method of recognizing architectural drawings which realizes a great reduction in input time by automatically inputting the information.

[0005]

To solve the above problems, the present invention uses the following method. (1) Based on the positional relationship between a closed area obtained from horizontal and vertical line segment information and a symbol indicating a layout, basic closed areas are integrated and the boundary position of the room is recognized.

(2) From the closed graphic information included in the room area, a drawing symbol such as a graphic symbol and a line segment symbol representing facilities attached to the room is obtained, and further, the room attribute is obtained from the information such as the drawing symbol. Determine and grant.

In the prior art, it was not possible to recognize the boundary of a room area including some closed figures from an architectural floor plan consisting of a collection of many closed figures. By recognizing the boundary position of the room from the closed area obtained from the vertical line segment information and the positional relationship between the symbols representing the layout, the room area can be correctly recognized.

Conventionally, it has not been possible to automatically recognize room functions (room attributes) such as a Japanese-style room, a kitchen, a bedroom, a bathroom, an entrance, a corridor, etc. From the included closed figure information, find the drawing symbols representing the facilities attached to the room, and recognize the room attributes using the pre-registered architectural knowledge base, etc., based on the information of the drawing symbols for each room area. It becomes possible to do.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The architectural drawing recognition method according to the present invention relates to a plan view, an elevation view, and the like showing a floor plan and an elevation view of a general house, and a room area such as a kitchen, a toilet, a bath, a toilet, a Japanese-style room, a bedroom, a living room, and the like. It can be used for automatic recognition of symbol elements inside a room, such as a kitchen set, a toilet bowl, a bathtub, and a bed.

[0010] The following describes the room area in the architectural floor plan,
An embodiment of the present invention applied to automatic recognition of a symbol element written inside a room will be described. FIG. 10 is a diagram illustrating an example of a handwritten floor plan. The room area in the architectural floor plan is separated from an adjacent room and the outside by a sign (floor sign) representing a wall or a fitting (a door, a window, a sliding door, etc.). Also, within the room area, a symbol (graphic symbol) representing the auxiliary equipment (toilet, bed, bathtub, etc.) of the room
Is written.

In the present invention, the outline information and auxiliary equipment information for each room are recognized from the handwritten floor plan or the printed drawing as shown in FIG. Kitchen, bathroom, etc.).

FIG. 11 shows an example of recognition results obtained from the handwritten floor plan of FIG. 10 by applying the present invention. FIG.
FIG. 4 is an explanatory diagram of a processing procedure of the present invention.

In FIG. 1, reference numeral 1 denotes an architectural drawing recognizing device comprising a CPU, a memory, and the like for implementing the present invention, and 2 denotes a drawing input device such as an image scanner. Room area recognition processing 10
Is a process for obtaining contour line information of a room region partitioned by walls, fittings, etc., and a drawing symbol recognition process 20 is a process for obtaining a symbol region composed of one or more figures inside the room region, and a room function recognition. The process 30 is a process of giving the attribute to the room area.

[1] Room Area Recognition Processing In the room area recognition processing 10, a basic area extraction 11 and a room area integration 12 are performed.

(1) Extraction of basic region In the extraction of basic region 11, attention is paid to line segments written in the horizontal and vertical directions in the drawing, and a closed region surrounded by the line segments is extracted. Specifically, Japanese Patent Application No. 7-302367 “Extraction Method of Drawing Closed Region” is used. In this method, a closed area is extracted from an architectural floor plan drawn by hand using a drawing sheet having grid lines described in a dropout color or the like.

This drawing preparation paper is used to determine not only the grid lines described in the dropout color, but also the distinction between the front and back sides of the paper written in different colors that are not dropped out, the positions of the paper and the grid lines, the inclination of the paper, and the like. It has at least two determination marks for detection. In addition, it is of course possible to extract a closed region from a printed architectural floor plan without using such a drawing sheet.

First, image data (raster data) is obtained from the floor plan of the building via an image scanner or the like, and based on the image data, the presence or absence of a line segment in each direction at each grid intersection is detected. , Intersection information (horizontal,
Vertical component). A state pattern of the grid intersection is determined according to the presence or absence of a line segment at each grid intersection, and a pattern map is generated.

Subsequently, from the state pattern of the grid intersections in the generated pattern map, routes which are contours of the closed region are sequentially searched based on the priority of a predetermined route search direction, and the route which has reached the start point is defined as a closed region. Extract. These closed areas are used as basic area information.

FIG. 2A is a diagram showing an example of image data of a floor plan of a building read by an image scanner. According to the above-described processing, the image data shown in FIG.
The intersection information as shown in FIG.
The basic area information (closed area) as shown in (C) is extracted.

By using such a method for extracting a closed region of a drawing, it is possible to extract a closed region without being affected by symbols such as a hinged door which becomes noise when extracting a region. Compared to this method, as shown in FIG. 3, in the method of extracting a closed region by tracing all line segments without distinguishing each other, a region considered to be a room is divided by doors and equipment symbols. It is not preferable as an extraction method.

(2) Room area integration In the room area integration 12, the basic area information obtained in the basic area extraction 11 and the floor sign information such as walls and sliding doors are extracted from the architectural floor plan, and the basic area information and the floor plan are extracted. Integration of adjacent basic areas is performed using the symbol information.

FIG. 4 is a diagram showing an example of a layout symbol to be extracted. The floor plan symbols include walls, sliding doors, general openings, hinged doors, double swing doors, folding doors, and the like. As a method for extracting a floor plan symbol from a building floor plan, for example, a method proposed in Japanese Patent Application No. 7-112729 “drawing recognition method” can be used.

In recognition of a floor plan symbol by this drawing recognition method, a visual field is moved by moving a scanning mask along a grid line of input image data, and a characteristic amount for determination is determined for each visual field. Is determined, the number of lines on the grid line and the thickness of each line are determined based on the feature amount determined for each visual field region, and the layout symbol is recognized from the determination result. This makes it possible to perform automatic recognition processing that is less susceptible to handwriting-specific line displacement, noise, and the like.

Subsequently, room areas are integrated. FIG.
It is a figure explaining the process of integration of a room area. In FIG. 5, a building floor plan (image data) is input via a scanner (). From this, the basic area information is extracted (), and the layout symbol information is extracted (). In addition, as a result of the recognition process of the layout symbol, the layout symbol is given a start point, an end point, and the attribute of the layout symbol (discrimination of a wall, a sliding door, and the like).

From the layout symbol information, all the information overlapping the outline of the basic area information are selected (). here,
The part where the outline of the basic area overlaps with the floor plan symbol is indicated by a double line, and the part where it does not overlap is indicated by a dotted line.

It is checked whether or not the outline portion (dotted line portion) of the basic area which does not overlap with the floor plan symbol exceeds a certain allowable value. As a result, when there is no outline part of the basic area that does not overlap with the floor plan symbol, or when all the outline parts of the basic area that do not overlap the floor plan sign are within the allowable value, the information of the closed area is used as the room area. sign up.

On the other hand, when the contour portion of the basic region that does not overlap with the floor plan symbol exceeds the allowable value, another adjacent closed region having the non-overlapping portion of the basic region as a common contour line, The closed area that has been checked on the contour is integrated into one closed area (,).

It is checked again whether or not the outline of the newly added closed area and the layout symbol overlap with each other with respect to the outline of the newly added closed area in the integrated closed area. A similar determination is made as to whether the part can be a boundary of the room area. Such room area integration is repeated until all contour lines of the basic area are regarded as boundaries of the room area. The above processing is performed on all the closed areas, and the room area is extracted from the drawing ().

In the basic area extraction 11, graphic elements such as tatami mats composed of horizontal and vertical line segments are also extracted as basic areas (closed areas). Thus, one room area (here, a Japanese room area) is recognized.

FIG. 6 is a processing flowchart of the room area recognition processing. Note that the numbers in FIG. 6 correspond to those in FIG. When the architectural floor plan is input (), the basic area is extracted (), and the floor plan symbol is recognized ().

Next, the length of the overlap between the outline of the basic area and the layout symbol is checked, and it is determined whether or not the overlap length is within the allowable range. If the overlap length is within the allowable range, The basic region is determined as a room region, and if the length of the overlap is not within the allowable range, the overlap portion is set as the boundary of the room region ().

Next, it is checked whether or not the contour line of the non-overlapping portion and the contour line of the other basic area overlap. As a result, it is determined whether or not there is an overlap. If no overlap exists, it is checked whether or not the contour lines overlap with another basic area.

If there is an overlap, the two basic areas are merged (room area integration), and it is determined whether or not the newly added contour is a layout symbol. If the newly added outline is a layout symbol, the outline is regarded as a boundary of the room area, and the room area is determined (). Here, if the newly added contour is not a layout symbol, the above-mentioned processing and subsequent steps are repeated for the newly added contour.

[2] Drawing Symbol Recognition Processing The drawing symbol recognition processing 20 includes symbol candidate extraction 21 and building knowledge collation 22.

(1) Symbol Candidate Extraction In the symbol candidate extraction 21, as shown in FIG. 7, a closed figure existing in each room area is extracted. A closed figure is a closed area surrounded by line segments and an area that matches the shape of a figure registered in advance by template matching processing.

As a specific example of the template matching process, there is a method proposed in Japanese Patent Application No. 7-258366 "Drawing Recognition Method". In this method, dictionary data of a closed region and positional relationship data between the closed regions with respect to a composite region including a plurality of closed regions are registered in advance, and when the closed region is recognized, the closed region is first detected from a drawing. And
Pattern matching is performed between the detected closed area and the dictionary data of the registered closed area. When the detected closed area matches the dictionary data of the registered closed area, the registered closed area is subsequently registered. With reference to the positional relationship data between the closed regions, the detected closed region is searched for a closed region having a pattern predicted at the position where the presence is predicted from the referenced positional relationship data, and a plurality of closed regions are searched. A composite area composed of areas is automatically recognized.

(2) Architectural Knowledge Verification In the architectural knowledge verification 22, the shape of the closed figure extracted for each room area, the size such as length, width, and area, the positional relationship with other closed figures, and the like are determined. It is checked whether or not there is a kitchen set, a toilet, a bathtub, a bed, or the like that is registered in advance as a drawing symbol.

More specifically, the above-mentioned Japanese Patent Application No. 7-2583 is disclosed.
Using a method such as No. 66 “Drawing Recognition Method” or the like, drawing symbols such as stairs and storage are recognized using the shape, size, and adjacent relationship between regions of a closed figure.

In many cases, graphic symbols indicating incidental facilities of a room such as a kitchen set and a bathtub have a closed relationship with each other, and recognition using the characteristics is performed as follows.

For example, as shown in FIG. 8, a symbol representing a kitchen set is recognized as a shape that includes a closed figure representing a stove and a closed figure representing a sink in a closed figure representing an outline. This symbol is a symbol pattern used in commonly used architectural floor plans. Checks the inclusion relationship between a circular closed figure representing a stove, a square closed figure representing a sink, and a rectangular closed figure representing an outline, as well as the size of each closed figure and the position and distance between the closed figures. Thus, the kitchen set can be recognized.

[3] Room Function Recognition Processing Room function recognition processing 30 performs room function recognition 31. In the room function recognition 31, from among the drawing symbols obtained from the drawing symbol recognition processing 20 or the layout symbols, those whose function is determined by the existence of the room are selected, and the room corresponding to the room region including the symbol is selected. Assign attributes.

FIG. 9 is an explanatory diagram of room function recognition, showing examples of graphic symbol patterns used in commonly used architectural floor plans and room attributes given by the symbols.

In the present embodiment, as examples of determining room attributes from drawing symbols, kitchen set symbols indicate kitchens, bathtub symbols indicate bathrooms, toilet bowl symbols indicate toilets, vanity tables indicate washrooms, tatami symbols indicate Japanese-style rooms, The bedroom symbol is recognized from the bed symbol, and the entrance and corridor are recognized from the rising frame symbol.

The following architectural knowledge is used when recognizing an entrance or a corridor from an upward frame symbol. For example, as shown in the drawing of FIG. 11, when the room area is in contact with the rising frame and the room area is in contact with the outside area (that is, when the floor plan includes the outermost boundary line). Makes the room area the entrance,
The corridor is the area opposite the room area, which is the entrance, with the rising frame interposed.

[0045]

According to the present invention, a bedroom, a tatami room, and the like can be automatically detected from a floor plan of a bit pattern. As a result, information such as the number of rooms and the presence or absence of a bathroom is clarified. By inputting such information into a separately built integration system, it becomes possible to estimate the house from the floor plan. Can be considered.

As described above, according to the present invention, labor costs can be reduced by greatly reducing the drawing input cost, and the processing time for digitizing data (conversion to CAD data) can be shortened. Can establish an advantage.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of a processing procedure of the present invention.

FIG. 2 is a diagram illustrating basic region extraction.

FIG. 3 is a diagram showing a closed region extraction process for tracking all line segments.

FIG. 4 is a diagram illustrating an example of a floor plan symbol.

FIG. 5 is a diagram illustrating room area integration.

FIG. 6 is a processing flowchart of a room area recognition process.

FIG. 7 is a diagram illustrating symbol candidate extraction.

FIG. 8 is a diagram illustrating architectural knowledge collation.

FIG. 9 is a diagram illustrating room function recognition.

FIG. 10 is a diagram illustrating an example of a handwritten floor plan to be processed;

11 is a diagram showing a recognition result of the handwritten floor plan shown in FIG. 10;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Architectural drawing recognition apparatus 2 Drawing input apparatus 10 Room area recognition processing 11 Basic area extraction 12 Room area integration 20 Drawing symbol recognition processing 21 Symbol candidate extraction 22 Building knowledge collation 30 Room function recognition processing 31 Room function recognition

Claims (4)

(57) [Claims] 1. A method for recognizing a room area existing in a drawing from architectural drawing information input by a drawing input device, comprising: a closed area surrounded by horizontal and vertical line segments in the drawing information. A first process of recognizing a region as a basic region;
A second step of recognizing the floor plan symbol in the drawing, and detecting an overlap between the recognized closed area surrounded by the horizontal and vertical line segments and the recognized floor plan symbol. Sets the overlap as the boundary of the room area, and if there is no overlap, detects the presence or absence of a common contour line with another basic area. As a result, if a common contour line exists, And checking whether the outline of the newly added basic region is a floor plan symbol, until all of the outlines of the integrated basic region are regarded as boundaries of the room region. And a method for recognizing an architectural drawing, wherein the third step is repeated to integrate room areas. 2. A step of integrating room areas realized by the method according to claim 1, a step of recognizing a drawing symbol inside the integrated room area, and using the recognized drawing symbol as a key. A method for recognizing an architectural drawing, comprising: a step of searching for attribute data that gives a drawing symbol and an attribute of a room area corresponding to the drawing symbol; and a step of determining an attribute of the room area based on a search result. 3. A process for recognizing a drawing symbol includes determining that a plurality of closed regions having a shape conforming to a previously registered shape pattern are in a positional relationship of an inclusion relationship to each other, 3. The architectural drawing recognition method according to claim 2, wherein the step of recognizing the drawing symbol is based on the fact that the information relating to the area is within a range of values registered in advance. 4. A specific room area such as an entrance, a corridor, etc., based on a specific symbol such as a rising frame represented by a line segment such as a double line peculiar to an architectural drawing. The architectural drawing recognition method according to any one of claims 1 to 3, wherein an attribute indicating is provided.