JPH0573627A - Picture retrieval device - Google Patents

Abstract

(57) [Summary] [Purpose] To quickly search for images that match the search conditions from among multiple images that have been accumulated in advance. [Structure] A plurality of images are input in advance from the image input means 1, the image feature description is extracted by the image feature extraction means 8, and the image feature description is input to the image structure interpretation means 10 to create an image structure description, The image structure description is stored together with the image input to the image structure storage means 11. on the other hand,
An image feature description is extracted by the image feature extraction unit 3 from the image of the search condition input from the search condition input unit 2, and the image feature description is input to the image structure interpretation unit 5 to create an image structure description. This image structure description and the image structure description accumulated in the image structure accumulating means 11 are used as the image structure collating means 6
The search result is displayed on the search result display means 7 by collating and searching. [Effect] The image search speed is increased.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image retrieving apparatus for inputting and accumulating image information such as documents and drawings and retrieving an appropriate image in response to a request.

[0002]

2. Description of the Prior Art FIG. 7 shows, for example, Japanese Patent Application No. 1-189768.
FIG. 8 is a block diagram showing a configuration of a conventional image retrieval device disclosed in Japanese Patent Publication No. JP-A-2003-264, in which 1 is an image input means for inputting an image to be stored, and 14 is an image feature extracted from the input image. , Image feature storage means for storing together with this image, 2 is a search condition input means for inputting an image which is an example of an image satisfying a search condition, and 3 is for extracting image features from the image input by this search condition input means 2. Image feature extraction means for creating an image feature description, 4 is an interpretation rule setting means for setting an interpretation rule in advance, and 5 is based on the interpretation rule and the image feature description created by the image feature extraction means 3. , An image structure interpretation means for creating an image structure description corresponding to the search condition, and 6 collates the image structure description with the image feature description stored by the image feature storage means 14. Image structure collating means for retrieving images matching the search-condition Te, retrieval result display means for displaying an image retrieved by the image structure collating means 6 7, 13
Is an image structure creating means for creating an image structure of an image to be searched.

Next, the operation of the apparatus shown in FIG. 7 will be described. For example, an image to be stored as shown in FIGS. 2A to 2D described later is input by the image input unit 1 and stored as image data. Next, the image feature storage means 14 analyzes the stored image data, extracts image features as shown in FIG. 3 described later, and associates this with image data to store as image feature description. The feature accumulated here is the symbol S included in each image.
The number of each of S1, S2 and S3 and the connection relation between the symbols are represented by a symbol string.

On the other hand, the user has a search request, for example, "I want to search for images in which the same type of symbol is connected." Create an image like this. The search condition input means 2 reads this image and stores it internally as image data. Next, the image feature extraction means 3 analyzes the contents of this image data, recognizes the types of symbols and connection relationships, and creates an image feature description as shown in FIG. 8, for example. Here, as an image satisfying the above-mentioned search condition, FIG.
The image number (a) and the image number (b) are stored, but the image feature description shown in FIG. 3 corresponds only to the image number (a).

Next, the image structure interpretation means 5 converts the image feature description shown in FIG. 8 to create an image structure description as shown in FIG. This image structure description exactly corresponds to the above search conditions. The conversion from the image feature description to the image structure description is performed using the interpretation rule set by the interpretation rule setting means 4. Next, the image structure matching means 6
Searches for an image satisfying the search condition by matching the image structure description with the image feature description stored in advance. For example, when the image structure description shown in FIG. 6 is compared with the image feature description shown in FIG. 3, the image number (a) satisfies the condition if X = S1, and the image number (b) satisfies the condition if X = S2. Since the conditions are satisfied, the images of (a) and (b) are searched. Finally, the retrieved images are
It is displayed on the CRT by the search result display means 7.

[0006]

Since the conventional image retrieving apparatus is constructed as described above, all the image feature descriptions stored in the image feature storing means 14 by the image structure collating means 6 are searched. Since it was converted into the image structure description every time, there was a problem that the search time of the image to be searched was long.

The present invention has been made to solve the above problems, and an object of the present invention is to provide an image search apparatus capable of performing a search in a shorter time than the conventional search time.

[0008]

An image search apparatus according to the first aspect of the present invention sets an image feature extraction unit 8 for extracting an image feature description from an image of an image input unit 1 and a predetermined image feature interpretation rule. Interpreting rule setting means 9, an image structure interpreting means 10 for creating an image structure description corresponding to a search condition from the interpretation rule and the image feature, and an image structure storing means for storing the image structure description together with the corresponding image. 11 and an image structure description whose search condition matches the image structure description searched from the image structure description means are searched by collating them with the image structure description stored in the image structure storage means, and the search result is displayed. The image structure matching means 12 displayed on the means is provided. In the image search device according to the second aspect of the present invention, the search condition input means is provided with an input device such as a keyboard for directly inputting the image feature description in characters.

[0009]

In the image retrieval apparatus according to the first aspect of the present invention, an image is input in advance from the image input means, and the image feature extraction means extracts the image feature description from the image. The image structure interpretation means creates an image structure description from the image feature description according to the interpretation rule of the interpretation rule setting means and stores it in the image structure storage means. An image as a search condition is input from the search condition input means, and an image structure description of the image as the search condition is created by the image structure creating means. The image structure matching unit searches the same image structure description as the search condition by collating with many image structure descriptions stored in the image structure storage unit, and displays the search result on the search result display unit. In this way, the image structure matching means, when searching for an image, matches the image structure description with each other, so that the search time is shortened.
In the image search device according to the second aspect of the present invention, the image feature description serving as the search condition can be input as a character string from an input device such as a keyboard of the image structure input means.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of an image search apparatus according to an embodiment of the first invention. In FIG. 1, 1 is image input means, 2 is search condition input means, 3 and 8 are image feature extraction means, 4 and 9 are interpretation rule setting means, 5 and 10 are image structure interpretation means, and 7 is search result display means. , 11 is an image structure storing means, 12 is an image structure collating means, and 13 is an image structure creating means. 1 to 5 and 13 in FIG. 1 have the same functions as those of the conventional ones, and are therefore given the same reference numerals, and the description of the following configuration will be omitted.

The image feature extraction means 8 extracts an image feature description from the image input by the image input means 1. For example, the image input from the image input means 1 is (a) in FIG.
3 to (d), the image feature description as shown in FIG. 3 is performed from the images of those symbols. The interpretation rule setting means 9 has an interpretation rule of the image feature that "when two or more symbols of the same type are included in the connection relationship, these symbols are replaced with a variable X representing an arbitrary symbol." Set. The image structure interpretation means 10 creates an image structure description based on the interpretation rule and the image feature description created by the image feature extraction means 8. This image structure description is as shown in FIG. 4, and corresponds to the image input from the image input means 1. The image structure storage means 11 stores the image structure description together with the corresponding image. The image structure collating means 12 collates the image of the search condition converted into the image structure description with the image structure description of the image structure accumulating means 11, retrieves the matched one, and displays it on the search result display means 7. To do.

FIG. 2 shows a concrete example of an image input to the apparatus of FIG. 1, and the symbols (a) to (d) of FIG. 2 are input from the image input means 1. FIG. 3 shows a specific example of the image feature description extracted by the image feature extraction means 8 of the apparatus of FIG. As shown in FIG. 3, the image feature description consists of three items of image number, number of symbols, and connection relation. The image numbers correspond to (a) to (d) in FIG. Regarding the number of symbols, the code corresponding to each symbol [(S1
~ S3.) Is written on the margin], and the number of these symbols is described in correspondence with the structure.
For example, in FIG. 2A, the symbol code corresponds to S1,
Since there are two S1, "2" is described. As for the connection relation, the connection of each symbol is shown by a symbol. Figure 2
(A) becomes (S1 S1). FIG. 4 shows an image structure description created by the image structure interpretation means of the apparatus of FIG. This image structure corresponds to that in FIG. 3, and is described based on the interpretation rule of the interpretation rule setting means 9 that "the same symbol is two or more and is replaced with the variable X". Therefore, the connection relation is replaced by the variable X.

Next, the operation of the embodiment of the first invention will be described with reference to FIGS. FIG. 5 shows a specific example of an image serving as a search condition input from the search condition input means of FIG. 1, and FIG. 6 shows a specific example of an image structure description created from the image of FIG. 5 by the image structure creating means.

First, an image as shown in FIGS. 2A to 2D is input from the image input means 1, and the image feature extraction means 8 extracts the image feature description of FIGS. 2A to 2D. Then, the image feature description as shown in FIG. 3 is created. The image structure interpreting means 10 converts the image feature description of FIG. 3 into the image structure description shown in FIG. 4 according to the interpretation rule of the interpretation rule setting means 9, and stores it in the image structure storing means 11. On the other hand, an example of an image of the search condition as shown in FIG. 5 input from the search condition input means 2 is an image structure description (connection relationship = (X X), where X = S1 shown in FIG. , S2, S3 Here, the image feature extraction means 3 converts the image of Fig. 5 into the image feature structure shown in Fig. 8. The image structure matching means 12 matches the image structure description with the image structure description. As a result, since the image numbers (a) and (b) in Fig. 4 are the same as the images corresponding to Fig. 5, the image number (a),
(B) will be detected [connection relation (X
X)]. Finally, the retrieved image is displayed on the CRT image or the like by the retrieval result display means 7.

Next, an image retrieval apparatus which is an embodiment of the second invention will be described. In the image search device of the second invention, the search condition input means 2 of FIG. 1 is provided with an input device such as a keyboard instead of the device for reading an image to create image data. The image feature description as shown in FIG. 8 is directly input as a character string from this input device.

[0016]

As described above, according to the first aspect of the present invention, the input image is converted into the image structure description in advance and stored in the image structure storage means together with the input image to retrieve the image. In this case, the image structure description of the image to be searched is searched from the image structure description stored in the image structure storage means. Therefore, since the search can be performed by matching the image structure description comrades,
There is no need to convert the image structure description for each search, which has the effect of enabling high-speed search. According to the second aspect of the invention, since the image feature description of the search condition can be input in characters, there is an effect that the search time can be further shortened as compared with the case of inputting an image.

[Brief description of drawings]

FIG. 1 is a block diagram of an image search apparatus according to an embodiment of the first invention.

FIG. 2 is a specific example of an image input to the device of FIG.

FIG. 3 is a diagram showing an example of image feature description of an image input to the apparatus of FIG.

4 is a diagram showing an example of an image structure description of an image input to the apparatus of FIG.

5 is a diagram showing an example of an image input as a search condition to the apparatus of FIG.

6 is a diagram showing an image structure description of the image of FIG. 5;

FIG. 7 is a block diagram showing an example of a conventional image search device.

8 is a diagram showing an image structure description when the image of FIG. 5 is input to the apparatus of FIG.

[Explanation of symbols]

 8 image feature extraction means 9 interpretation rule setting means 10 image structure interpretation means 11 image structure storage means 12 image structure matching means

[Procedure amendment]

[Submission date] June 24, 1992

[Procedure Amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0014

[Correction method] Change

[Correction content]

First, an image as shown in FIGS. 2A to 2D is input from the image input means 1, and the image feature extraction means 8 extracts the image feature description of FIGS. 2A to 2D. Then, the image feature description as shown in FIG. 3 is created. The image structure interpreting means 10 converts the image feature description of FIG. 3 into the image structure description shown in FIG. 4 according to the interpretation rule of the interpretation rule setting means 9, and stores it in the image structure storing means 11. On the other hand, an example of an image of the search condition as shown in FIG. 5 input from the search condition input means 2 is an image structure description (connection relationship = (X X), where X = S1 shown in FIG. , S2, S3 Here, the image feature extraction means 3 converts the image of Fig. 5 into the image feature structure shown in Fig. 8. The image structure matching means 12 matches the image structure description with the image structure description. As a result, since the image numbers (a) and (b) in Fig. 4 are the same as the images corresponding to Fig. 5, the image number (a),
(B) will be detected [connection relation (X
X)]. Finally, the retrieved image is displayed on the CRT screen or the like by the search result display unit 7.

Claims (2)

[Claims] 1. A search condition input means for inputting an image of a search condition, and an image structure creating means for extracting an image feature from the image and creating an image structure description of the image based on an interpretation rule of the image structure. In the image search device, which is provided with a search result display unit for displaying the searched image and an image input unit for inputting the image, and which searches for and displays an image matching the search condition from a plurality of images accumulated in advance, An image feature extraction means for extracting an image feature description from an image of the image input means, an interpretation rule setting means for setting an interpretation rule of a predetermined image feature, and an image for creating an image structure description from the interpretation rule and the image feature description. The structure interpreting means, the image structure storing means for storing the image structure description together with the corresponding image, and the image to be searched from the image structure creating means match the search condition. An image retrieving device comprising an image structure collating means for collating the image structure description stored in the image structure storing means with the image structure description to retrieve and displaying the image structure description on the retrieval result display means. 2. The image search device according to claim 1, wherein the search condition input means is provided with an input device such as a keyboard for directly inputting the image feature description by a character string.