JPH0778803B2 - Image database search method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to image information for promptly searching and presenting desired image information using a simplified image as a key from an image database that stores a plurality of pieces of image information. It relates to the search method.

(Prior Art) Conventionally, there are the following techniques for retrieving desired image information from an image database.

(A) A unique identification number such as a screen number is given to each image information, and as shown in FIG. 12, a correspondence list of the accumulation position 2 and the identification symbol 3 in the accumulation medium 1 of each image information. 4 is prepared in advance. When the user 5 inputs the identification symbol 3 attached to the desired image information, the identification symbol is referred to on the correspondence list 4 and the image information accumulated in the corresponding accumulation position 2 is presented.

(A) A plurality of identification symbol groups are given to each image information in advance, and as shown in FIG. 13, a correspondence list 4 of the accumulation position 2 and the identification symbol group 3'of the accumulation medium 1 of each image information is prepared in advance. Have it ready. When the user 5 inputs a plurality of identification symbols that are expected to be attached to the predetermined image information, these identification symbol groups 3'are referred to on the correspondence list 4, and the identification symbol groups are associated in descending order of suitability. A method of presenting multiple pieces of image information stored in the storage position.

In the above method (a), the user 5 needs to accurately store the identification symbol 3 attached to each image information.
When the number of pieces of image information stored in the image database becomes huge, it becomes extremely difficult to search for desired image information.

Further, in the above method (a), it is general to use a word, which is called a keyword, which represents the content of image information, as an identification symbol. By applying this keyword, it is possible to search for a plurality of pieces of image information that are candidates for the desired image information, even if the information necessary for searching the desired image information is ambiguous to some extent. However, if only a symbol group such as a keyword is used as search information, there is a high possibility that image information having a layout far from what the user expected is presented. That is, individual constituent elements in the image information indicated by the keyword (here, the constituent elements are parts having the meaning of constituting the image information, for example,
12 desk image information "desk" 6, "apple" 7 and "person"
The positional relationship, size, posture, etc. of (8 is a constituent element) are information that cannot be specified by keywords, and therefore the layout of image information is completely different. In particular, if the number of pieces of image information stored in the image database becomes enormous, the number of candidate image information also becomes enormous, which makes quick search difficult.

The above-mentioned image information retrieval system will be described in more detail. It was stated that the user inputs the identification symbol or the identification symbol group of the image information as the retrieval information representing the desired image information. A simple identification number unrelated to the content of image information such as a screen number is used as the identification symbol by the method (a), and a semantic image structure is used as the identification symbol group by the method (b). Keyword 10 indicating an element is mainly used.

For example, when the desired image information is an image in which the apple 7 is placed on the desk 6 in the room 9 as shown in FIG. 14, the user can search as "10999", for example. Enter a simple identification number of 3 and keywords 10 such as "apple" and "desk". The image database is shown in Fig. 15 (a).
As shown in (b), a correspondence list of the identification number 3 or the keyword 10 and the storage position of the image information on the storage medium 1 is prepared in advance, and the search information input by the user is collated on the correspondence list. Are converted to the image information storage position 2. In the example of FIG. 15A, since the storage position 2 corresponding to the identification number 10999 is A, the image information stored in the position A of the storage medium 1 is read and presented.

However, when the search information is the identification number, the user needs to accurately check or memorize the identification number of the desired image information in advance. There is a drawback that image information far apart is presented.

Further, since the identification number itself is generally irrelevant to the image content, if the correspondence between the identification number and the content of the image information is unknown, the search is virtually impossible.

On the other hand, when the search information is a keyword, FIG. 15 (b)
A correspondence list of a plurality of words (keywords) 10 representing the contents of the image information and the storage position 2 of the image information in the storage medium 1 is prepared in advance. And
The user inputs one or more words (keywords) that are considered to represent the constituent elements of the desired image information as search information.

For example, in this example, the user has "apple" 7 and "desk" 6
It is understood that there are three pieces of image information including both of these two keywords, and these storage positions 2 are A, B, and C, respectively.
Therefore, the three pieces of image information accumulated in the accumulation positions A, B, and C are read and presented as candidates.

Since the keyword itself expresses the content of the image information to some extent, it is possible to display some image information including the desired image information without inputting all the keywords given to the desired image information correctly. It is highly possible to present it as a candidate. That is, even if the search information input by the user is ambiguous to some extent, the image information can be searched.

However, since the search information is ambiguous, when the number of image information stored in the image database becomes enormous, the image information selected as a candidate by the keyword also becomes enormous. . Moreover, even if the keywords match, the candidates will include image information far from what the user desires. For example, the user desires the image information shown in FIG. 14 and inputs the keywords “apple” and “desk”, but image information having a completely different layout as shown in FIG. 16 is selected as a candidate. There is also. It is impossible to express the information regarding the layout of the desired image information by simply enumerating the words that represent the constituent elements of the image information, such as keywords. That is, it is difficult to sufficiently reflect the concrete image envisioned in the mind in the search only by using the keyword, and therefore it is not possible to efficiently search the image information.

On the other hand, a method is also conceivable in which the user draws a desired image information layout as a schematic figure and uses the schematic figure itself as search information. Such a method is called an image similarity search method, and compared with a method that uses an identification number or keyword as search information, it fully reflects the concrete image that the user has in mind about the desired image information. can do.

However, in such a method, it is necessary to recognize and understand what the schematic figure drawn by the user represents. Such a technique is called image understanding, but a practical image understanding technique has not been realized so far.

(Object of the Invention) In the image similarity search method of the present invention, a user semi-automatically creates a schematic figure as a search key, and by using this, a process of recognizing and understanding a schematic figure can be omitted, and a desired figure can be obtained. The purpose is to efficiently retrieve image information.

(Structure of the Invention) (Differences between Features of the Invention and Conventional Techniques) The present invention relates to a drawing / editing processor of a language input format of a schematic figure for searching image information desired by a user in an image similarity search method. The main feature is that the need for recognition and understanding seen in a handwritten figure is avoided and the user can easily create the schematic figure by using the.

Therefore, the present invention provides the above-mentioned image information with a unique identification symbol such as an image number (A),
This is different from the so-called keyword-based method (a) in which a plurality of identification symbol groups is given to each image information in advance.

(Principle of operation of the invention) In the present invention, instead of the identification number or the keyword as the search information, the constituent elements of the image information and the size, orientation, posture, positional relationship of each constituent element within the image information, etc. Use various attributes. For example, in the image information of FIG. 14 described above, the components are “desk” 6, “apple” 7, “chair” 11,
A “window” 12, each of which has attribute information such as size, orientation, posture, position, and positional relationship among the components.

The present invention uses a language input format drawing processor when the user inputs these search information. As shown in FIG. 1 (a), when the name of the component 14 is input, a simple sketch 18 showing the shape of the component is displayed. The default attribute information 15 of the constituent element is written in the search graphic edit list 14A shown in FIG.

Furthermore, as shown in FIG. 2A, the position 151, the size 152, the orientation 153, and the attitude 154 of the component are formed by the tablet 16 or the like.
By changing the etc., the schematic figure is interactively edited according to the image information to be searched.

Edit this outline figure and edit the attributes of the components (size of figure,
(Direction, posture, position, etc.) is changed, the search graphic edit list is immediately displayed as shown in FIG. 2 (b).
The attribute value of the component of 14A is also changed (here, the position is changed). By using the editing function of the drawing processor,
The user can interactively bring the displayed schematic figure 18 close to the desired image information envisioned in the head,
At the same time, the attribute information of each component can be changed on the search graphic edit list 14A.

And the finally edited search figure edit list is
This is search information for searching for desired image information. However, a search image attribute list is created in advance for each piece of image information to be stored, and as shown in FIG. 3, a correspondence table between the storage position 2 of each image information and the search image attribute list 14A (search Correspondence list) must be prepared.

As a method of creating a search image attribute list for each piece of image information to be stored, for example, FIGS. 4A and 4B are used.
As shown in, the image information 17 and the output graphic of the drawing processor 13 are displayed in an overlapping manner, and the schematic graphic 18 corresponding to the constituent elements of the image information is displayed on the output surface 19 of the drawing processor. Then, similarly to the case of inputting the search information, the size, the posture, the position, the orientation, etc. of the schematic figure 18 are edited using the drawing processor in accordance with the corresponding constituent elements in the image information, and the image is edited. Create a search image attribute list corresponding to the information.

At the time of search, the search pattern edit list edited by the user for the desired image information using the drawing processor is collated with the search image attribute list defined at the time of image information storage, and each component and each attribute information About one or a plurality of pieces of image information having a high degree of conformity are presented as candidate images.

In the present invention, the search graphic edit list used as the search information has various attribute information (size, posture, orientation, position, etc.) of the constituent elements of the image information, unlike the keyword, so that the search by the keyword is possible. In comparison, it is possible to search for image information that is closer to the image that the user had in mind as a candidate. From the viewpoint of image similarity search, the search graphic edit list is the same as the result of recognizing and understanding the handwritten schematic graphic drawn by the user as search information. That is, the present invention considers that the process of recognizing / understanding the approximate graphic required in the image similarity search is replaced by the drawing processor of the language input format, and thus the process of recognizing / understanding the approximate graphic is unnecessary. It can be regarded as a similar search method.

(Embodiment) FIG. 5 shows a block diagram of an embodiment of the present invention.
Is an input unit, 52 is an image component calling unit, 53 is an image component editing unit, 54 is a display unit, 55 is a candidate image information evaluation unit, 56 is a search correspondence list storage unit, 57 is an image information search control unit, Reference numeral 58 is an image information storage unit, and 59 is an editing information input unit. here,
The image information storage unit 58 stores T pieces of image information in advance, and the search correspondence list storage unit 56 stores the first image information of each of the T pieces of image information stored in the image information storage unit 58. As shown in FIG. 6, each component of the image information,
The correspondence relationship between the attribute information and the storage position in the image information storage unit 58 is stored in the form of a table, and an identification number is given to each image information for selecting a candidate image described later.

Next, an outline of operation will be described with reference to FIG. First, the name of the component of the image information desired by the user (for example, in FIG. 6)
14) is input from the input unit 51, the image component calling unit 52
Stores graphic data relating to a plurality of types of schematic figures and a default search figure edit list. The image figure element editing unit stores the schematic figure data corresponding to the names of the constituent elements and the default search figure edit list. Output to 53.

When the image component editing unit 53 reads the rough graphic data, the rough graphic data is visualized and displayed on the display unit 54. The user inputs the editing information from the editing information input section 59 while watching this. The edit information is information for moving the position of the substantially graphic being displayed on the display unit 54 and changing the size and posture, and is input from a tablet or the like.

When the image component editing unit 53 reads the editing information, the image component editing unit 53 moves the position of the schematic figure being displayed on the display unit 54, changes the size and the posture, and edits the schematic figure being displayed. Accordingly, the search graphic edit list is also changed at the same time.

In this way, the user edits each component displayed on the display unit 54 so that the position, size, posture, etc. of the corresponding component of the desired image information match.

The candidate image information evaluation unit 55 reads out the search graphic edit list created by the image component edit unit 53, and from the search correspondence list storage unit 56, the degree of matching between the search graphic edit list component and its attribute information. The image information retrieval control unit reads out the accumulation position of one or a plurality of high image information
Output to 57. The image information search control unit 57 reads out the image information corresponding to the storage position of the image information storage unit 58,
It is displayed on the display unit 54.

FIG. 7 shows a detailed view obtained by further disassembling the block diagram of FIG. 5, and the same reference numerals as those in FIG. 5 are the same. The image component calling unit 52 includes an image component search control unit 521 and an image component storage unit 522, and the image component editing unit 53 includes a drawing / editing processor 531 and a search graphic edit list storage unit. 532, and the candidate image information evaluation unit 55 includes a preliminary candidate storage unit 551, a component comparison unit 552, a basic candidate storage unit 553, a primary threshold value storage unit 554, and a primary evaluation storage unit 55.
5, a primary attribute comparison unit 556, a sorter 557, a primary candidate storage unit 558, a secondary threshold storage unit 559, a secondary evaluation storage unit 560, a secondary attribute comparison unit 561, and a sorter 562. A secondary candidate storage unit 563, a tertiary threshold storage unit 564, a tertiary evaluation storage unit 565, a tertiary attribute comparison unit 566, a sorter 567,
And a tertiary candidate storage unit 568.

Image component search control unit 52 of the image component calling unit 52
1 indicates that the image component storage unit 522 is input by the input unit 51.
It also controls the search graphic edit list storage unit 532. In the image component storage unit 522, as shown in FIG. 8, symbols representing various components, graphic data for displaying the components as a schematic graphic, and a search graphic edit list of the cargo component. The default values of K pieces of attribute information 15 are stored.

In the present embodiment, the symbols representing the constituents are given concrete names (“apple”, “desk”, etc.) and called the constituent name.

Further, as shown in FIG. 9, the figure data for displaying the schematic figure is a two-character set sequence (x ₀ , y ₀ ) (x ₁ , y ₁ ) in which the vertex coordinates of the schematic figure are arranged in the clockwise direction. ……. When representing an approximate figure as a solid in a three-dimensional space, the figure data is a three-character set sequence (x ₀ ,
y ₀ , z ₀ ) (x ₁ , y ₁ , z ₁ ) ... And topological information representing the connection of the vertex coordinates.

Further, here, the constituent element is a schematic figure in a two-dimensional space, the degree of the attribute is N = 3, and the primary attribute is “position” 151,
The secondary attribute is “posture” 154, and the tertiary attribute is “size” 152. The order of the attribute can be changed depending on the target image information.

The “position” of the primary attribute is the center of gravity of a rectangle circumscribing the schematic figure of the constituent element, and its coordinates are represented by a two-character set (x _c , y _c ). The "posture" of the secondary attribute is represented by a unit vector p in a two-dimensional space starting from "position". In the default state, p of each component is an initial position vector p ₁ (p ₁ is a unit in a two-dimensional space). Vector), and p is rotated by the rotation angle according to the rotation of the figure. The "size" of the tertiary attribute represents the enlargement / reduction ratio of a schematic figure, and the default value is 1 for each component.

Further, when treating a schematic figure as a solid in a three-dimensional space, the primary attribute is a set of three letters (x _c , y _c , z _c ) representing the coordinates of the center of gravity of a rectangular parallelepiped circumscribing the schematic figure, and the secondary attribute May be a unit vector in the three-dimensional space.

The drawing / editing processor 531 of the image component editing unit 53
It has the same function as a commercially available drawing / editing processor such as software for personal computers that generates and displays figures composed of straight lines and arcs, and edits their position, posture, size, etc. This drawing / editing processor 531
The graphic data of the constituent elements stored in the search graphic edit list storage unit 532 is read out, and the graphic unit in which the two-character set is connected by a straight line in a clockwise direction is displayed on the display unit 54.

When the figure data is a solid in a three-dimensional space, the three-dimensional coordinates indicated by the three-character set of the figure data are perspective-projected and converted into two-dimensional coordinates, and the converted vertex coordinates are indicated by the phase information. The figures connected in a straight line in order are displayed on the display unit 54.

The edit information input section 59 is a commercially available tablet or the like, and the user can move the edit information input section 59 or change the posture by converting the coordinate value of the graphic position being displayed on the display section 54. Enter edit information to change or change the size.

When the drawing / editing processor 531 reads the editing information from the editing information input section 59, the drawing / editing processor 531 moves the position of the figure being displayed on the display section 54, changes the posture, or changes the size according to the editing information. Or

To operate this, a symbol representing a component of desired image information is input to the input unit 51. In the present embodiment, since the input symbol is a component name that specifically indicates what the component is, "Apple", "desk", etc. are input.

The image constituent element search control unit 521 reads the constituent element name input to the input unit 51, and searches the image constituent element storage unit 522 for a constituent element name that is the same as the constituent element name. If there is, the graphic data and attribute information of the component name of the image component storage unit 522 are read out and stored in the search graphic edit list storage unit 532.

The image constituent element search control unit 521 searches by a conventional searching means using a keyword or the like, but since it searches for individual constituent elements of the image information, its identification is easy.

The search graphic edit list storage unit 532 stores the component name,
Graphic data and attribute information are stored as shown in FIG.

The drawing / editing processor 531 reads the graphic data 18A of the constituent elements stored in the search graphic edit list storage unit 532, and visualizes and displays the graphic data on the display unit 54. The above procedure is repeated until all the constituent elements included in the desired image information are displayed on the display unit 54, and the name of the constituent element, the graphic data, and the attribute information are added to the search graphic edit list storage unit 532.

Next, the user inputs the edit information from the edit information input section 59. The drawing / editing processor 531 has an editing information input unit 5
When the edit information is read from 9, the figure displayed on the display unit 54 is moved, rotated, and enlarged / reduced according to the edit information, and at the same time, the constituent element of the search figure edit list storage section 532 corresponding to the figure. Also rewrite the attribute value of.

This state is shown in FIG. 11, where (a) is before instruction and (b) is after instruction. Figure "Apple" displayed on display 54
When the position of (x _c , y _c ) is changed to (x _{c '} , y _c' ), the primary attribute corresponding to the figure in the search figure edit list storage unit 532 is (x _c , y _c ) to (x _{c '} , y
_{c '} ). At the same time, the figure is rotated to change its posture, and the vector p representing the posture of the figure is changed from p ₁ to
If it is changed to p _e , the secondary attribute corresponding to the figure in the search figure edit list storage unit 532 is linked from p ₁ to p
_When it is rewritten to _e and the size of the figure is enlarged 1.3 times,
The tertiary attribute corresponding to the figure is rewritten from 1 to 1.3.

In this way, the position, orientation, and size of each constituent element of the search graphic edit list storage unit 532 are changed, and editing is performed so as to approach the position, orientation, and size of the constituent element included in the desired image information. .

When the editing is completed as described above, the component comparison unit 552
Reads the first component name stored in the search graphic edit list storage unit 532, then reads the identification numbers of all image information including the component name from the search correspondence list storage unit 56, Candidate storage unit 553 and preliminary candidate storage unit 5
Accumulate at 51.

Next, the component comparison unit 552 reads the second component name stored in the search graphic edit list storage unit 532, and then sequentially reads each identification number stored in the preliminary candidate storage unit 551, It is checked whether the constituent element name is included in the search image attribute list of the image information corresponding to the identification number of the search correspondence list storage unit 56. If it is not included, the identification number accumulated in the preliminary candidate accumulating unit 551 is deleted, and if it is included, it is left as it is.

Next, the component comparison unit 552 reads the content of the preliminary candidate storage unit 551 and writes it in the basic candidate storage unit 553. This is repeated, and if all the identification numbers of the preliminary candidate accumulating unit 551 are deleted, the component comparing unit 552 stops its operation and the primary attribute comparing unit 556 starts its operation. The primary attribute comparison unit 5
The operation of 56 will be described later.

In this way, the component comparison unit 552 has the image information having the component name in the search correspondence list storage unit 56 for all the component names stored in the search graphic edit list storage unit 532. To find out. With such a function, when the search is completed for all the constituent element names, the contents of the basic candidate accumulating unit 553 are images containing the largest number of constituent element names accumulated in the search graphic edit list accumulating unit 532. The identification number of information is stored.

The primary attribute comparison unit 556 includes a search graphic edit list storage unit 532.
The primary attribute of the first constituent element stored in the first candidate is read, then the identification number of the first candidate stored in the basic candidate storage unit 553 is read, and the image information of the identification number in the search correspondence list storage unit 56 is read. Read the primary attribute of the component having the same name as the component from the search image attribute list,
The distance d between the two is calculated and stored in the primary evaluation value storage unit 555.

In this embodiment, two primary attributes (x _c , y _c ) and (x _{c '} , y
_{c 'distance} d1 (x _{c of), y c, x c'} , y c ') the _{d1 (x c, y c,} x c', y c ') = (x c -x c') 2 + ( y _c −
y _{c ′} ) ² .

If there is no component with the same name, d1 = 0.
Similarly, the primary attribute comparison unit 556, regarding the primary attributes of all the constituent elements stored in the search graphic edit list storage unit 532, corresponds to the primary attributes of the corresponding constituent elements of the search image attribute list of the image information. And the distance is calculated and added to and accumulated in the primary evaluation value accumulating unit 555.

Then, when the distance evaluation is completed for all the constituent elements, the contents of the primary evaluation value accumulating section 555 are calculated by the number of constituent elements existing in both the search graphic edit list accumulating section 532 and the search image attribute list of the image information. Break. Since such a function is provided, when the distance evaluation is completed for the primary attributes of all the constituent elements, the contents of the primary evaluation value accumulating unit 555 are the constituent elements of the first candidate image information and the search corresponding to them. This is the average value of the distances from the primary attributes of the constituent elements accumulated in the graphic editing list accumulating section 532, and the image information of the first candidate accumulated in the basic candidate accumulating section 553 and the schematic figure edited / created by the user. And the conformity of the primary attribute (positional information in the case of the present embodiment) of each component.

Next, the primary attribute comparison unit 556 compares the content of the primary threshold value storage unit 554 with the content of the primary evaluation value storage unit 555. In addition,
The content of the primary threshold value storage unit 554 is a threshold value for selecting a candidate based on the evaluation value of the primary attribute. If the evaluation value is smaller than the threshold value, the identification number of the image information and the evaluation value are stored in the primary candidate storage unit 558, and the contents of the primary evaluation value storage unit 555 are rewritten to 0.

Next, the primary attribute comparison unit 556 uses the second candidate of the basic candidate storage unit 553.
Regarding the primary attributes of all the constituent elements of the image information of the candidate image information, the distances from the primary attributes of the respective constituent elements of the search graphic edit list accumulating unit 532 are calculated as in the case of the first candidate image information, The average value is stored in the primary evaluation value storage unit 555.

Next, the primary attribute comparison unit 556 compares the content of the primary threshold value storage unit 554 with the content of the primary evaluation value storage unit 555, and if the evaluation value is smaller than the threshold value, the Two
The identification number and the evaluation value of the candidate image information are accumulated.

Each time a new candidate is added to the primary candidate accumulating unit 558, the sorter 557 compares the evaluation values of the candidates accumulated in the primary candidate accumulating unit 558, and the primary candidate accumulating unit in descending order of evaluation value.
Rewrite the contents of 558. If the number of candidates stored in the primary candidate storage unit 558 exceeds the primary candidate constant S1, S1 +
The identification numbers and evaluation values of the first and subsequent candidates are deleted.

In this way, the primary candidate comparison unit 556 evaluates all the image information in the basic candidate storage unit 553 for conformity with the primary attributes of the constituent elements stored in the search graphic edit list storage unit 532, and sorts them. 557 is S from smallest evaluation value to S
One candidate is stored in the primary candidate storage unit 558.

Similarly, the secondary attribute comparison unit 561 has the primary candidate storage unit 558.
Search image edit list storage section for all image information of
The average distance from the secondary attributes of the constituent elements accumulated in 532 is obtained as an evaluation value and accumulated in the secondary evaluation value accumulating unit 560,
If the final contents of the secondary evaluation value storage unit 560 are smaller than the contents of the secondary threshold value storage unit 559, the secondary candidate storage unit 5
The identification number and evaluation value of the image information are stored in 63.

The content of the secondary threshold value storage unit 559 is a threshold value for selecting a candidate image based on the evaluation value of the secondary attribute. Then, the sorter 562 accumulates S2 candidates in the descending order of evaluation value in the secondary candidate accumulating unit 563. However, the distance d2 (p, p ') between the two secondary attributes ("posture") p and p'is set to d2 (p, p') =-p.p '.

Similarly, the tertiary attribute comparing unit 566 also uses, as an evaluation value, the average distance between all the image information in the secondary candidate accumulating unit 563 and the tertiary attribute of the constituent elements accumulated in the search graphic edit list accumulating unit 532. Obtained and stored in the tertiary evaluation value storage unit 565, and if the final content of the tertiary evaluation value storage unit 565 is smaller than the content of the tertiary threshold value storage unit 564, the tertiary candidate storage unit 568 identifies the image information. Store the number and rating.

The content of the tertiary threshold value storage unit 564 is a threshold value for selecting a candidate image based on the evaluation value of the tertiary attribute. Then, the sorter 567 accumulates S3 candidates in the descending order of evaluation value in the tertiary candidate accumulating unit 568. However, the distance d3 (f, f ') between the two tertiary attributes ("size") f and f'is set to d3 (f, f') = (f-f ') ² .

The image information search control unit 57 reads the identification numbers of the S3 candidate images stored in the tertiary candidate storage unit 568, and reads the S3 image information corresponding to the identification numbers from the image information storage unit 58, It is displayed on the display unit 54.

It should be noted that the schematic figure data used in the present embodiment is solid figure data in a three-dimensional space, and the primary attribute (“position”) is the center of gravity of the rectangular parallelepiped circumscribing the solid figure and the secondary attribute (“posture”) p.
If is the unit vector in the three-dimensional space, it becomes possible to edit the posture of each component of the image information in the three-dimensional space, and use it as information when retrieving the three-dimensional posture of the component. be able to.

Furthermore, the position (x _c , y _c ) of the primary attribute is set as the position coordinate (x _c , y _c , z _c ) in the three-dimensional space, and the image component is a basic unit that constitutes a natural image such as an apple or a desk. Not as
It is clear that if the basic building blocks (spheres, triangular prisms, cylinders, cubes, etc.) that make up the objects themselves such as apples and desks are used, the search method will be a database that stores individual objects as a single unit. . Further, although the present embodiment has been described with a schematic figure as a constituent element, it is clear that a similar effect can be obtained by laying out the words themselves on the screen instead of the schematic figure, although the search accuracy is lowered. .

(Effect of the invention) As described above, according to the present invention, the search graphic edit list used as the search information is different from the keyword, and various attribute information of the components of the image information, such as size, posture,
Since it has an orientation and a position, it is possible to search for image information that is closer to the image drawn by the user as a candidate, as compared with the conventional search using only a keyword. Therefore, it is possible to quickly search for the image information that matches the image of the user.

Further, the search pattern edit list is the same as the result of recognizing and understanding the handwritten schematic pattern drawn by the user as search information when viewed from the image similarity search. This is a similar image search method that does not require recognition and understanding of approximate figures, by considering that the process of recognizing and understanding approximate shapes required for image similarity search is replaced by a language input format drawing processor. Can be regarded as

In addition, the number of image information as seen in the conventional methods (a) and (b) does not become enormous, and the user can quickly and quickly without memorizing the correspondence between the identification number and the image information. It is searchable and has excellent economical efficiency and operability.

[Brief description of drawings]

1 to 4 are explanatory views of the operation principle of the present invention,
FIG. 1 shows the display state of a schematic figure by the drawing processor,
The figure shows a state in which attribute values of constituent elements of the search graphic edit list are changed, FIG. 3 shows a correspondence table (search correspondence list) between the storage positions of image information and the search image attribute list, and FIG. 4 shows image information. Editing example (search image attribute list) by superimposing with a schematic figure, FIG. 5 is a block diagram of an apparatus of one embodiment of the present invention, and FIG. 6 is a storage position in the image information storage unit 58 of FIG. FIG. 7 is a detailed block diagram of FIG. 5, and FIG. 8 is a block diagram for displaying the constituent elements and their schematic shapes. Example of attribute information such as graphic data, FIG. 9 is graphic data for displaying a schematic graphic, FIG. 10 is a storage example of the search graphic edit list storage unit 532, and FIG. 11 is a drawing / editing processor by user input. Fig. 12: One example before and after instructing the edit list for search to be created Through FIG. 16 shows an example of retrieving desired image information from a conventional image database,
Fig. 12 shows an example of image information retrieval using identification symbols, Fig. 13 shows an example of image information retrieval using identification symbols (keywords), Fig. 14
Figures and 16 are examples of image information, Figure 15 is an identification number,
It is an explanatory view of the relationship with the storage position of the storage medium by the keyword. 1 ... storage medium, 2 ... storage position, 3 ... identification code,
3 '... identification symbol group, 4 ... correspondence list, 5 ... user input, 6 ... desk, 7 ... apple, 8 ... person, 9 ...
Room, 10 ... keyword, 11 ... chair, 12 ... window, 13 ...
… Drawing processor, 14 …… Components, 14A …… Search graphic edit list, 15 …… Default attribute information, 151 ……
Position, 152 …… size, 153 …… direction, 154 …… posture, 16
…… Tablet, 17 …… Image information, 18 …… Abbreviated figure, 18A
...... Graphic data, 19 ...... Display surface of graphic data, 51 ...... Input section, 52 ...... Image component calling section, 521 ...... Image component search control section, 522 ...... Image component storage section, 53 ... … Image component editor, 531 …… Drawing / editing processor, 532
...... Search pattern edit list storage section, 54 display section, 55
... Candidate image information evaluation unit, 551 ... Preliminary candidate storage unit, 552 ...
… Component comparison section, 553 …… Basic candidate storage section, 554 …… Primary threshold value storage section, 555 …… Primary evaluation value storage section, 556 ……
Primary attribute comparison unit, 557 …… Sorter, 558 …… Primary candidate storage unit, 559 …… Secondary threshold value storage unit, 560 …… Secondary evaluation value storage unit, 561 …… Secondary attribute comparison unit, 562 … Sorter, 56
3 …… Secondary candidate storage, 564 …… Third threshold storage, 56
5 ... Tertiary evaluation storage unit, 566 ... Tertiary attribute comparison unit, 567 ...
… Sorter, 568 …… Third candidate storage unit, 56 …… Search correspondence list storage unit, 57 …… Image information search control unit, 58 …… Image information storage unit, 59 …… Edit information input unit.

Claims (1)

[Claims] 1. A first search graphic edit list is created in advance for each image in an image database, and a second search graphic edit list is created from image information to be searched, and the first search graphic edit list is created. In the image database search method for presenting search candidate images based on the matching result between the graphic editing list for search and the second graphic editing list for search, the first and second search graphic editing lists are images. Each of the constituent elements, which is a part having the meaning of constructing, has attribute information regarding the layout of the schematic figure representing the shape of the constituent element in the image. The list is created by inputting the name of the component in the drawing editing processor and displaying the schematic figure and editing the layout of the schematic figure. Attribute information on out, the image database retrieval method which comprises position in the image of the components, size, information about the orientation and attitude.