JPH10171831A - Device for retrieving image file - Google Patents

Abstract

PROBLEM TO BE SOLVED: To identify an object at a high speed and with high precision in a simple constitution. SOLUTION: This device is provided with an image scanner 1 which reads the image of an object to be identified, binarizing part 21 which binarizes the read image, center of gravity calculating part 22 which calculates the center of gravity of the binarized image, distance calculating part 23 which calculates a distance from, the calculated center of gravity of the outside periphery of the binarized image by the preliminarily set angle, similarity calculating part 24 which compares the calculated distance data with each distance data in plural picture data stored in an image database 4, and calculates the similarity, and display 3 which displays image retrieved from the image database 4 in the order of the high similarity.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an image file retrieving apparatus capable of identifying an object by image matching.

[0002]

2. Description of the Related Art Conventionally, there has been known an image file search apparatus for storing a large number of images in a file and searching for and displaying a desired image from the files. In this conventional example, a key search is generally performed. In this key search, one or more character strings, such as the name and characteristics of the image, are assigned to each image in advance as a search key and registered together with the image, and the search key is input at the time of search. By doing so, the images are narrowed down, and finally the target image is extracted. This is a search for an image file that has been generally performed conventionally.

[0003]

However, in the above-described conventional image file search apparatus, a plurality of search keys must be set in advance, which not only takes a lot of trouble in the operation at the time of an image file, but also causes a problem. When searching for an image to be searched, unless one of the search keys is always entered, the clues of the search cannot be obtained, and the identification of the target object becomes extremely difficult.

[0004] The present invention has been made in view of the above circumstances, and an object of the present invention is to provide an image file search device capable of quickly identifying an object with a simple configuration.

[0005]

According to a first aspect of the present invention, there is provided an image reading means for reading an image of an identification object, and binarizing the image read by the image reading means. Binarizing means, a center of gravity calculating means for calculating the center of gravity of the image binarized by the binarizing means, and a distance from the center of gravity calculated by the center of gravity calculating means to an outer edge of the binarized image Is calculated by a predetermined angle increment, and the distance data obtained by the distance calculation means is compared with each distance data in a plurality of image data stored in the image database to determine the similarity. It is characterized by comprising: a similarity calculating means to be obtained; and a display means for displaying images retrieved from the image database in descending order of similarity obtained by the similarity calculating means. It is intended to.

In the above configuration, the image data of the object to be identified is registered in the image database in advance. When identifying an object, an image of the object is input from an image scanner, a digital camera, or the like, and the input image is binarized. The center of gravity of the binarized image is obtained, and the distance from this center of gravity to the outer edge of the image is obtained, for example, in units of two degrees. The obtained distance data is compared with the distance data of each image registered in the image database to determine the similarity. And
The similarities are displayed on the display in descending order of similarity. This makes it possible to identify the target object by reliably extracting the corresponding image, while eliminating the need for troublesome search key input. The device of the present invention is particularly effective when automatically identifying tens of thousands of objects such as keys.

[0007]

FIG. 1 is a block diagram showing an embodiment of an image file search device according to the present invention. In this embodiment, the apparatus of the present invention is applied to a key automatic identification system that reads a key image, performs matching with a database image, and automatically identifies a key as an identification target.

As shown in FIG. 1, the image file search apparatus includes an image scanner 1 for reading an image of an object to be identified, and a personal computer main body 2 for calculating a feature amount of image data read from the image scanner 1.
A display 3 for displaying images in the order of similarity calculated by the personal computer main body 2, and an image database 4 storing a large number of images relating to the identification target.
It is composed of

The personal computer main body 2 includes a binarizing section 21, a center-of-gravity calculating section 22, a distance calculating section 23, a similarity calculating section 24, a display control section 25,
It has. Each function will be described later.

The image database 4 has image data and its characteristic amount (distance data from the center of gravity of the key image to the outer edge in this embodiment) and character information on the object for each type. As the character information, for example, various attribute data such as a product name, a manufacturer name, a model number, and specifications can be considered.

<Operation of Image File Searching Apparatus> Next, the operation of this embodiment will be systematically described. FIG. 2 is a flowchart illustrating a processing procedure in the image file search device. As shown in this flowchart, this apparatus includes an image capturing process (Step S1), a binarization process (Step S2), a feature amount calculation process (Step S3),
It is composed of matching similarity calculation processing (step S4), sorting processing (step S6), and candidate display processing (step S7).

First, when an image to be searched is read from the image scanner 1 (image capture processing in step S1), the binarizing section 21 binarizes the read image with a predetermined threshold value ( (Binary processing in step S2). As shown in FIG. 3, the binarized image of the key is represented by f (i,
j), the background portion becomes f (i, j) = 0,
The key part is f (i, j) = 1.

<< Characteristics Calculation Processing in Step S3 >>
For the binarized image, distance data from the center of gravity is calculated as described below, and this is used as a feature amount.

First, the center-of-gravity calculating section 22 obtains the center of gravity G of the binarized image based on the following equation.

[0015]

(Equation 1) Here, S is an area, and is represented by the following equation (2).

[0016]

(Equation 2) Next, as shown in FIGS. 4A and 4B, the distance calculation unit 23 calculates the distance from the calculated center of gravity G (i _G , j _G ) to the peripheral edge at each angle. h (θ) angle θ
Shall be in increments of two. That is, θ = {0}, 2 ゜, 4 ゜, 6 ゜,..., 358 ゜｝ Next, the distance data obtained as described above is displayed in a graph with the horizontal axis representing the angle. FIG. 5 is a graph of key distance data, in which the input image t (k) is indicated by a solid line, and the search image (also referred to as a database image or a search result image) fn (k) is indicated by a dashed line. .

<< Pattern Matching and Similarity Calculation in Step S4 >> Pattern matching is performed by the following normalized correlation.

Cauchy-Schwarz inequality

(Equation 3) Is expanded as follows.

[0019]

(Equation 4) Since the above equation (5) is a quadratic equation for a,
Discriminant of quadratic equation shown in the following equation (6)

(Equation 5) From

(Equation 6) However, in equation (7), “=” is t (k) = αfn
In the case of (k), that is, when the distance data of the input image and the distance data of the database image are similar (when both images have the same shape). K indicates a 2-degree angle, and indicates a value of 0 to 179.

The left side of the above equation (7) indicates the normalized correlation, and the normalized correlation indicates the similarity between the images. However, since the normalized correlation does not become larger than 1, 10
The value multiplied by 0 is displayed as the similarity.

In this way, the similarity between the input image t (k) and the searched database image fn (k) is sequentially obtained, and displayed on the display in descending order of similarity (steps S6 to S8). FIG. 6 shows a specific display screen configuration example. As shown in FIG. 6, the input image and the search result image are displayed on the screen, and the similarity is displayed. Also, attribute information attached to the search result image in advance,
For example, information such as the key type, name, model name, and manufacturer is displayed. The operator can see various information such as the type of the key to be identified and the manufacturer by looking at the screen of the search result.

In this embodiment, the image scanner 1 is used as the image reading means. Alternatively, image data captured using a digital camera or the like may be taken in.

In this embodiment, the similarity between images is determined by comparing distance data from the center of gravity in the binarized image. However, a color image is used as an input image, and the distance data and the distance data are used. The use of the color information together with the color information is effective for identifying a colored object.

Furthermore, in this embodiment, the key is used as the identification object, but the present invention is not limited to this, and the invention can be applied to various objects such as shape recognition of automobiles and motorcycles, and identification of firearms.

[0025]

As described above, according to the first aspect of the present invention, the image of the object to be identified is read, the read image is binarized, and the center of gravity of the binarized image is obtained. The distance from the center of gravity to the outer edge of the binarized image is obtained at preset angle intervals, and the obtained distance data is
Since the similarity is obtained by comparing each distance data in a plurality of image data stored in the image database, and the images searched from the image database are displayed in descending order of the obtained similarity. Even with a simple device configuration, the object can be identified at high speed and with high accuracy. Moreover, since a plurality of images can be searched in the order of similarity, the identification work is dramatically improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of an image file search device according to the present invention.

FIG. 2 is a flowchart showing the operation of the embodiment of the image file search device according to the present invention.

FIG. 3 is an explanatory diagram illustrating a binarization process performed by a binarization unit;

FIG. 4 is an explanatory diagram showing a method for obtaining distance data from the center of gravity.

FIG. 5 is a graph showing the obtained distance data developed by angles.

FIG. 6 is an explanatory diagram showing a screen configuration example of a search result.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Image scanner 2 Personal computer main body 3 Display 4 Image database 21 Binarization part 22 Center of gravity calculation part 23 Distance calculation part 24 Similarity calculation part 25 Display control part

Claims (1)

[Claims] 1. An image reading means for reading an image of an object to be identified, a binarizing means for binarizing the image read by the image reading means, and a binarizing means for binarizing the image read by the binarizing means. A center-of-gravity calculating means for obtaining a center of gravity; a distance calculating means for obtaining a distance from the center of gravity obtained by the center-of-gravity calculating means to an outer edge of the binarized image in increments of a predetermined angle; A similarity calculating means for comparing the obtained distance data with each distance data in a plurality of image data stored in the image database to obtain a similarity; and a similarity calculating means having a high similarity obtained by the similarity calculating means. Display means for displaying images retrieved from the image database in order.