JP2004070861A - House floor plan retrieval program and house floor plan retrieval device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a house floor plan retrieval program and a house floor plan retrieval device accurately carrying out house floor plan retrieval on the basis of similarity in shapes of floor plans by a computer. <P>SOLUTION: By defining an inherent color to each element composing a house floor plan, an image is composed that is color-coded per each element, and a characteristic amount of the image is calculated from its color distribution. By carrying out retrieval based upon similarity between images by using the characteristic amount, retrieval is carried out based upon similarity of arrangement and shape of the elements composing the floor plan. In regard to an outline shape of the whole floor plan, a binary image is composed expressing whether or not it is an area on the floor plan, a characteristic amount of the image is calculated from distribution of its luminance gradient vectors, and in the same way, retrieval is carried out based upon similarity. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a technology for searching for a house layout using a computer.
[0002]
[Prior art]
At present, housing design using CAD (Computer-Aided Design) is generally performed. Drawings created by CAD can be stored in a storage device as electronic data. Generally, the stored drawing data is managed by a relational database together with information on related houses. When the floor plan is described by CAD data, the association between the closed area in the drawing and the elements (parts) that make up the floor plan, such as the entrance, the water area, and the Japanese-style room, is defined.
[0003]
[Problems to be solved by the invention]
It is possible to search for elements that make up the floor plan of the entrance, water area, Japanese-style room, etc. (hereinafter simply referred to as “components”) with an interest in how they are arranged in the drawing. If it becomes a method that can be handled by ordinary users, the housing information service can be remarkably improved.
[0004]
When performing a search targeting the arrangement of components, that is, a search based on the similarity of layouts, currently, a state that is propositionally expressed in a form such as "entrance to the north" is an attribute value For example, a method of registering in a relational database and performing a search can be considered.
[0005]
However, it is actually impossible to comprehensively register the shapes of multiple components and their spatial arrangement, the spatial relationship between components, and the magnitude relationship between components using such a propositional expression. Impossible.
[0006]
On the other hand, the overall shape of a floor plan is also important information when considering the similarity between floor plans. Again, it is difficult to express propositionally and register it in a relational database.
[0007]
In addition, if there is no attribute that is satisfactory to the search user, the search cannot be performed naturally.
[0008]
Therefore, it is difficult to respond to various needs of users only by searching based on propositional attributes.
[0009]
An object of the present invention is to provide a house floor plan search program and a house floor plan search device that allow a computer to accurately perform a house floor plan search based on similarities in floor plan shapes.
[0010]
[Means for Solving the Problems]
The house floor plan search device of the present invention for achieving the above object converts a drawing into a color image by assigning mutually different colors depending on the type of the constituent element to a closed area for each constituent element in the floor plan, A color image is divided into a plurality of regions, a color distribution feature amount is calculated for each of the divided regions, a set of all the regions is obtained, and a house floor plan as a search key and a house floor plan with a high similarity between the color distribution feature set are determined. It is characterized by searching.
[0011]
When a color distribution histogram is used as the color distribution feature, a set of color distribution histograms of a plurality of floor plans are stored in a feature data file, and a layout similar to the house layout of the search key is searched from the file. Performed by extracting a sample.
[0012]
The color distribution histogram set is a feature amount vector, and the similarity is defined by the distance between the feature amount vectors. Then, by calculating the color distribution histogram set, information on the arrangement, size, and shape of the components in the floor plan is expressed as the color distribution of the pixel values in the color image.
[0013]
When the color distribution in the image is converted into a feature vector, a statistic such as an average or a variance of the color distribution can be used in addition to the color distribution histogram.
[0014]
According to another embodiment of the present invention, there is provided a house floor plan search apparatus for generating a binary image by specifying different colors for a house floor plan part and other parts in a floor plan, and from the binary image. A brightness gradient vector image is created by obtaining a brightness gradient vector, the brightness gradient vector image is divided into a plurality of regions, a histogram of the brightness gradient vector distribution is calculated for each of the divided regions, a set of all the regions is obtained, and a search key is obtained. It is characterized by searching for a house layout sample having a high similarity between the house layout and the luminance gradient vector distribution histogram set.
[0015]
The search is performed by storing a set of brightness gradient vector distribution histograms of a plurality of floor plans in a feature data file and extracting a floor plan sample similar to the house floor of the search key from the file.
[0016]
The luminance gradient vector distribution histogram set is a feature amount vector reflecting the outline shape of the house floor plan, and the similarity is defined by the distance between the feature amount vectors.
[0017]
As described above, any one of the similarity of the color distribution histogram set or the similarity of the luminance gradient vector distribution histogram set makes it possible to accurately perform a house floor plan search based on the similarity of the floor plan shape. By combining, more accurate search becomes possible.
[0018]
In addition, the house layout as a search key can be freely created by a user. Therefore, according to the present invention, selection of a propositional attribute such as using a drawing drawn by a user himself as search key data is performed. This makes it possible to specify search conditions with a high degree of freedom, which cannot be achieved by the search using.
[0019]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, a house floor plan search program and a house floor plan search device according to the present invention will be described in more detail with reference to embodiments of the invention shown in the drawings.
[0020]
FIG. 1 shows a configuration of a house floor plan search device of the present invention. In FIG. 1, reference numeral 11 denotes a program file storing a house floor plan search program 21 of the present invention, 12 denotes a CPU (central processing unit) for executing data creation and search processing based on the house plan search program 21, 2120 A data file 14 having a CAD data file 111 which is a drawing of a floor plan of a house to be searched, an image data file 130 for calculating a feature vector, and a feature data file 142 used for search; Is a memory used during processing, 15 is a display device for displaying a screen being processed and the contents of various data, 16 is an input device for inputting a floor plan of a house, and 17 is an external device. A communication connection device 18 for connecting to the device 2200 via the communication path 2300 connects these components to each other. An internal bus for, 2100 is a residential floor plan search apparatus comprising a computer configured by higher.
[0021]
The house floor plan search device 2100 is accessed from the external processing devices 2200 located at remote locations, and performs search processing. The external processing device 2200 sends data and requests from the input device 26 to the house floor plan search device 2100 in accordance with the instructions of the floor plan search program 21, and displays the search results on the display device 25. Data and requests from the input device 26 are sent from the communication connection device 27 via the communication path 2300. The house floor plan search device 2100 is a server-side device, and the external processing device 2200 is a client-side device (personal computer) used by general users.
[0022]
FIG. 2 shows a procedure of a search process executed by the computer based on the shapes and arrangements of the components according to the house floor plan search program 21. First, in the floor plan stored in the CAD data file 111, the drawing is converted into a color image by assigning different colors to the closed regions for each component in the drawing depending on the type of the component (step). S101), this is stored in the image file 130. Next, the color image is divided into a plurality of regions independently of the floor plan (step S102). Subsequently, a histogram of the color distribution is calculated for each of the divided areas, a set of all the areas is obtained (step S103), and the obtained set is stored as a feature vector in the feature data file 142. Further, a set of color distribution histograms of a plurality of house floor plans to be searched is stored in the feature data file 142 (step S104).
[0023]
Next, by operating the input device 26 of the external processing device 2200, a layout color image to be used as a search key is created using components defined by mutually different colors, and a color distribution histogram set is calculated (step S1). S105). A floor plan sample having a high similarity between the floor plan used as a search key and the color distribution histogram set is searched in the feature data file 142 (step S106), and the result is displayed on the display device 25 (step S107).
[0024]
Note that, as described above, when converting the color distribution in an image into a feature vector, in addition to the color distribution histogram, statistics such as the average and variance of the color distribution can be used. The color distribution histogram, the color distribution statistic, and the like are color feature amounts.
[0025]
Next, FIG. 3 shows a procedure of a search process executed by the computer based on the overall shape of the floor plan in accordance with the house floor plan search program 21. First, in the floor plan stored in the CAD data file 111, a binary image is generated by specifying different colors for the house floor plan portion and other portions in the drawing (step S111). Next, a luminance gradient vector is obtained from the binary image, and a luminance gradient vector image having the luminance gradient vector as a pixel is created (step S112), and is stored in the image file 130. Next, the luminance gradient vector image is divided into a plurality of regions independently of the floor plan (step S113). Subsequently, a histogram of the luminance gradient vector distribution is calculated for each of the divided regions, a set of all the regions is obtained (step S114), and the obtained set is stored as a feature vector in the feature data file 142. . Further, a set of brightness gradient vector distribution histograms of a plurality of sample house floor plans to be searched is stored in the feature amount data file 142 (step S115).
[0026]
Next, by operating the input device 26 of the external processing device 2200, a luminance gradient vector image is created from the floor plan binary image used as a search key, and the luminance gradient vector distribution histogram set is calculated (step S116). A floor plan sample having a high similarity between the floor plan serving as a search key and the luminance gradient vector distribution histogram set is searched in the feature data file 142 (step S117), and the result is displayed on the display device 25 (step S118).
[0027]
Instead of using the above-described binary image to grasp the overall shape, information on the shape of the boundary line included in the CAD data is used, and a feature amount vector is obtained from the boundary line information. It is also possible to perform a search by extracting a house layout.
[0028]
Here, the flow of data at the time of constructing the feature amount from the floor plan data in steps S101 to S104 will be further described with reference to FIG. The input data file 110 includes a CAD data file 111 of a floor plan and a file 112 of bibliographic attribute data of a house corresponding to the floor plan.
[0029]
The CAD data from the CAD data file 111 is converted into image data by the image converter 120 and stored in the image file 130. The image converter 120 refers to the association between the layout components and colors defined in the color table file 121 and fills a closed area in the drawing. Note that the feature values used in the present embodiment directly refer to the RGB values of the image. Therefore, as the format of the image file 130, an uncompressed or an image compression format using index colors is used.
[0030]
The image data stored in the image file 130 is registered in the similar image search server 140 as new data. The similar image search server 140 reads image data from the image file 130, calculates an image feature amount, and stores the feature amount in the feature amount data file 142. Information such as the parameter setting of the feature amount is described in the setting file 141 for the similar image search server 140.
[0031]
In the present embodiment, the feature amount reflecting the arrangement and size of the floor plan components in the drawing is calculated as follows.
[0032]
First, the entire image is divided into rectangular regions of a fixed size by defining a fixed number of divisions for each of the vertical and horizontal directions of the image. A histogram is constructed by counting pixels having a color corresponding to each layout component for each rectangular area. A set of histograms obtained by collecting histograms for each rectangular area is defined as a feature amount vector of the image.
[0033]
Next, the feature amount that reflects the overall shape of the floor plan, calculated in steps S111 to S115, is specifically calculated as follows.
[0034]
First, a binary image is formed by assigning “1” to an area in which plan layout components are defined in the drawing data and “0” to an area in which plan layout components are not defined. Here, a white and black image is formed by designating white as the assigned “1” and black as the assigned “0”. However, the present invention is not limited to this, and different colors are assigned to the assigned “1” and “0”. Can be specified.
[0035]
A monochrome gray-scale image reduced by lowering the resolution of the binary image as necessary is formed. By lowering the resolution, it is possible to reduce the influence of unnecessary data as floor plan information such as line segment data in the drawing.
[0036]
By performing a two-dimensional differentiation process on a binary image or a monochrome grayscale image, an image having a luminance gradient vector as a pixel is configured. By defining a certain number of divisions for each of the vertical and horizontal directions of the luminance gradient vector image, the whole is divided into rectangular regions of a certain size. By discretizing the direction of the luminance gradient vector and adding the vector norm of each pixel for each discretized direction for each rectangular area, the intensity distribution in the vector direction in each rectangular area is converted into histogram-like data. Express. FIG. 5 shows an example of the intensity distribution histogram in the vector direction.
[0037]
A set of data obtained by collecting data for each rectangular area is defined as an image feature vector.
[0038]
Part of the numerical data provided as the bibliographic attribute data 112 in the input data is separately registered as a feature amount in the similar image search server 140 as needed. The bibliographic attribute data 112 is used in both a search process executed based on the shape and arrangement of the constituent elements and a search process executed based on the overall shape of the floor plan. At the time of search, the similar image search server 140 treats the present numerical data in exactly the same way as other image feature values.
[0039]
In the above, the image converter 120 and the similar image search server 140 are functional units formed by the CPU 12 when the CPU 12 executes processing based on the floor plan search program 21 in FIG.
[0040]
In addition, a search with high accuracy can be performed in any of the search for the shape and arrangement of the above-described constituent elements and the search for the entire shape of the floor plan. Exact search becomes possible.
[0041]
Subsequently, the flow of the processing at the time of the search in steps S105 to S107 and S116 to S118 will be further described with reference to FIG. The search device of this embodiment is the server side 2100, and the search process proceeds based on an input from the client side 2200.
[0042]
The server side 2100 on which the similar image search server 140 operates includes an image file set (data file) 2120 for all data. The image file set 2120 includes two types of data groups: actual data 2121 corresponding to an actual house, and handwritten data 2122 created by a client (user) as search data.
[0043]
When the client program 2210 issues a search request, the similar image search server 140 executes a search, reads an image corresponding to the search result, and sends the image to the program 2210 via the communication path 2300.
[0044]
When a search is performed using drawing data created on a client program 2210 to be described later as a key, the program 2210 sends the data to the similarity search server 140 as image data for new registration. The similarity search server 140 performs feature extraction from the transmitted data, stores the feature data in the feature data file 142, and stores the corresponding image data in the image file set 2120. After confirming the registration of the sending data, the program 2210 issues a search request using the data as a key. The program 2210 has a file 2220 for storing the created screen data and the like.
[0045]
FIG. 7 is a display example of a screen for creating a drawing for a search key on the client program 2210. The search is performed on the shape and arrangement of the components. The display unit 310 is a part for performing drawing. In the background grid formed by the vertical and horizontal grid lines, one square corresponds to the size of a half-tatami mat. Displayed on the display unit 320 is a list of floor plan components. After selecting the floor plan component to be designated from the list, dragging the mouse on the display unit 310 draws a rectangular area filled with the corresponding color. The size, aspect ratio, and position of the rectangular area can be changed by operating the mouse. In the display example, the common part 321 is selected, a rough shape 311 of the house is drawn, and a front entrance 312 and a water area 313 are drawn thereon.
[0046]
By pressing the button 330, a search for a floor plan similar to the drawn drawing is performed. A button 340 is used to specify a search target. In this display example, the first floor plan drawing is to be searched. The button 350 is a check box for designating whether the size of the drawn drawing is made valid as a search feature amount. In this display example, the setting is such that the difference in size is not taken into account during the search. Button 360 is a check box for designating whether to make the price of the house valid as a feature amount of the search. When this check box is checked, the price settings that can be entered to the right of it will be valid as search conditions.
[0047]
FIG. 8 is a display example of a search result displayed on the display device 25. The top 20 floor plan samples 400 are displayed in descending order of similarity to the drawn drawing.
[0048]
In the search key of this example, only the entrance and the area around the water are specified. Therefore, when calculating the similarity between the drawn drawing and the stored house floor plan, features obtained from the coincidence between the entrance and the water area and the shape of the whole floor plan are used.
[0049]
FIG. 9 is a panel for displaying an actually effective feature amount and changing the effective feature amount. Display unit 510 displays whether or not the similarity of the components (parts) of each floor plan is valid as a search condition. In this display example, the check boxes corresponding to the entrance and the area around the water are checked. That is, the display unit 510 selects a component.
[0050]
The button 511 is a button for validating the similarity of all the floor plan components. When this button is pressed, all the check boxes on the display unit 510 are checked. The button 512 is a button for invalidating the similarity of all the floor plan components. When this button is pressed, all the check boxes on the display unit 510 are not checked.
[0051]
The button 521 is a check box indicating whether or not the similarity of the shape of the entire floor plan is valid as a search condition. A button 522 is a check box indicating whether or not the similarity of the aspect ratio of the drawing is valid as a search condition. A button 523 is a check box indicating whether or not the similarity of the size values in the drawing is valid as a search condition. Button 524 is a check box indicating whether or not similarity of house prices is valid as a search condition.
[0052]
After the search condition is changed by changing the state of the check box on the panel in FIG. 9, when the button 530 is pressed, the changed search condition is transmitted to the similar image search server 140.
[0053]
Each component of the display unit 510 is provided with a slider 513 for designating a weight as necessary. The specified weight is used as one of the search conditions. In FIG. 9, the weight is selected in the range of 0-10. When the weight is specified, the influence on the search result changes according to the size.
[0054]
Note that a weight of 0 does not select a component, and a weight of 10 is equivalent to selecting a component. Therefore, selection and non-selection of a component on the display unit 510 can be represented by weights 0 and 10. .
[0055]
The panel in FIG. 10 is the same as the panel in FIG. 9, but the check box 610 corresponding to the Japanese room is checked, and the Japanese room is additionally selected.
[0056]
FIG. 11 shows a search result obtained by using the drawing 700 displayed at the top as key data under the conditions shown in FIG. The search key drawing itself is always displayed at the top because it has the greatest similarity. An area 710 in the drawing 700 is around water, an area 720 is a Japanese-style room, and an area 730 is a front entrance. With respect to these three floor plan components, a house floor plan sample whose layout is close is searched.
[0057]
By the way, in the housing exterior, the respective components such as the whole house layout (building), gates, trees, ponds, roads, etc. are regarded as the components in the housing layout, and the housing exterior is again regarded as the entire housing layout. Then, it becomes possible to search the exterior of the house using the same method and apparatus as used for searching the house layout.
[0058]
FIG. 12 is a display example of a search condition input screen in the exterior plan search performed from such a viewpoint. Display unit 810 is a screen for inputting an exterior plan. In FIG. 12, reference numeral 811 denotes the outline of the site, 812 denotes a building part, 813 denotes a tree, 814 denotes a pond, 815 denotes an entrance, and 816 denotes a peripheral road. As in the case shown in FIG. 7, a component to be drawn is selected from the external component list on the display unit 820, and its shape and position are input on the display unit 810. Further, in FIG. 12, 830 is a search execution button, 840 is a check box for specifying whether to enable the absolute size of the drawing as a search condition, and 850 is whether to enable the price as a search condition. This is a check box to specify whether or not to do so.
[0059]
In the above description, the search was performed by adding the road to the component parts. However, the present invention is not limited to such a search, and the search is performed on a range within the residential premises without including the road. Is also good.
[0060]
【The invention's effect】
According to the present invention, it is possible to perform a search based on the similarity of the shapes and spatial arrangements of the elements constituting the house floor plan and the similarity of the overall shape of the house floor plan, thereby enabling an accurate search of the house floor plan.
[Brief description of the drawings]
FIG. 1 is a configuration diagram for describing an embodiment of a house floor plan search device according to the present invention;
FIG. 2 is a flow chart for explaining an embodiment of a house floor plan search program according to the present invention;
FIG. 3 is a flowchart for explaining another embodiment of the house floor plan search program of the present invention.
FIG. 4 is an exemplary view for explaining a procedure for constructing floor plan data of a search target in FIG. 2;
FIG. 5 is a diagram for explaining an example of an intensity distribution histogram in a vector direction.
FIG. 6 is a configuration diagram for explaining a system that executes a search process in FIGS. 2 and 3;
FIG. 7 is a diagram for explaining creation of a drawing serving as a search key.
FIG. 8 is a diagram showing a display example of a search result.
FIG. 9 is a view for explaining setting of search conditions.
FIG. 10 is a diagram for explaining change of search conditions.
FIG. 11 is a diagram showing another display example of a search result.
FIG. 12 is a diagram for explaining creation of a drawing serving as a search key in a house exterior search according to the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 11 ... Storage device, 12 ... CPU, 17, 27 ... Communication connection device, 21 ... House layout search program, 25 ... Display device, 110 ... Input data file, 111 ... CAD data file, 112 ... Bibliographic data file, 120 ... Image converter, 130 image file, 140 similar image search server, 142 feature data file, 2100 house layout search device, 2120 data file, 2200 external processing device, 2210 client program, 2300 communication channel.

Claims (8)

A house floor plan search program for searching a house floor plan by a computer, the program comprising:
A procedure for converting the drawing into a color image by assigning different colors to the closed areas for each component in the drawing depicting the house floor plan by the type of the component,
Dividing the color image into a plurality of regions;
A procedure for calculating a color distribution feature amount for each divided area to obtain a set of all areas and a procedure for searching for a house layout having a high similarity between a house layout and a color distribution feature quantity set as a search key are executed. A house layout search program characterized by the following: Further comprising a procedure of constructing a feature vector from the shape of the boundary between the house layout part and the other part in the drawing depicting the house floor plan,
The house layout search program according to claim 1, wherein the house layout having a high similarity has a feature amount vector having a small distance from a house layout feature amount vector serving as the search key. 2. The computer readable program according to claim 1, wherein a weight is specified for the constituent element of the house layout, and the similarity of the color distribution feature quantity set is calculated in consideration of the weight. 3. A house floor plan search program for searching a house floor plan by a computer, the program comprising:
A procedure for generating a binary image by specifying different colors for the house floor plan part and the other parts in the drawing depicting the house floor plan;
A procedure for obtaining a brightness gradient vector from the binary image to create a brightness gradient vector image;
Dividing the luminance gradient vector image into a plurality of regions;
Calculating a histogram of the luminance gradient vector distribution for each of the divided regions to obtain a set of all the regions,
A house floor plan search program for executing a procedure for searching for a house floor plan with a high similarity between a house floor plan and a luminance gradient vector distribution histogram set as a search key. The house layout search program according to claim 1, wherein the house layout having a high similarity further has a high degree of matching with a bibliographic attribute of the house layout used as the search key. A house exterior search program for searching a house exterior by a computer, the program comprising:
A procedure of converting the drawing into a color image by assigning different colors depending on the type of the component to the closed area for each component in the drawing depicting the house exterior;
Dividing the color image into a plurality of regions;
A procedure for calculating a color distribution feature amount for each divided area to obtain a set of all areas, and a procedure for searching for a house exterior having high similarity between a house exterior as a search key and a color distribution feature set are described. A house exterior search program characterized by being executed. An apparatus for searching a house floor plan by a programmed computer, wherein a different color is assigned to each closed element for each component in the drawing depicting the house floor plan by the type of the constituent element to convert the drawing into a color image. Means for converting;
Means for dividing the color image into a plurality of regions;
Means for calculating a color distribution feature quantity for each divided area to obtain a set of all areas, and means for searching for a house layout having a high similarity between a house layout and a color distribution feature quantity set serving as a search key. A house layout search device characterized by the following. The house layout is displayed on a screen of a display device, and a background of a display portion of the screen for displaying the house layout is formed by vertical and horizontal grid lines in which one cell has a size of half a tatami mat. The house layout search program according to claim 1 or 4.

Images