JPH05159063A - Image retrieving device - Google Patents

Abstract

PURPOSE:To alleviate a restriction condition in a various conventional cases so as to extract an objective image from a screen by turning an object image to be retrieved into mosaic and using it as the feature of the objective image. CONSTITUTION:A retrieving objective image dictionary 104 stores the retrieving- objective image such as an apple, etc., by turning it into mosaic. An unknown image mosaicking part 203 turns an unknown image into mosaic and input it to a result output part 304 obtaining a retrieving result by way of a distance calculation part 300. By turning the image into mosaic like this, it easily comes to possible to find the object from within an optional screen and to cut it off from a background. Then, the objective image is easily found without giving a special condition such as uniformity to the background and generating an error owing to a noise in the case of using a line segment by turning the retrieving objective image and the unknown image into mosaic and retrieving them.

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 retrieving a target image from input grayscale images.

[0002]

2. Description of the Related Art A typical natural image is a grayscale or color image, but in the past, attention was paid to the shape of the target image in order to find the target image from the natural images,
With the shape information as a feature, a candidate having a matching feature is searched from the input image.

[0003]

Naturally, the first feature of the image is the shape, but it is very difficult to extract the correct shape from the natural image. Therefore, in the past, in order to separate the target object and the background, the background was pre-learned or the background was limited to a uniform one, and bright illumination was used to obtain the maximum contrast.
In order to make it easy to extract line segments, various restrictions were imposed, such as limiting the object to an artificial rigid object.

As described above, it is extremely difficult to accurately extract a shape from an arbitrary natural image, and therefore, there are many practical restrictions and difficulty in usefulness. It is an object of the present invention to relax various constraint conditions in the conventional case so that a target image can be extracted from the screen.

[0005]

FIG. 1 is a block diagram showing the principle of the present invention. Reference numeral 104 in the figure denotes a search target image dictionary, which stores, for example, mosaic images of search targets such as apples. An unknown image mosaicing unit 203 mosaics an unknown image. Reference numeral 300 is a distance calculation unit, and 304 is a result output unit that obtains a search result.

In the present invention, when extracting the feature of the object, not the shape but the grayscale or color information of coarse resolution is used. Specifically, the object image to be searched is mosaicized and this is used as a feature of the target image. In other words, it is based on the fact that even if the image has a significantly reduced resolution so that Lincoln can be identified from the Lincoln mosaic image, the characteristics of the object can be expressed.

[0007]

[Operation] By mosaicing the image in this way,
It is possible to easily find the target from any screen and separate it from the background. Of course, it does not prohibit the combined use of conventional means.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 2 is a diagram showing an embodiment of the present invention.

FIGS. 3 and 4 are diagrams for explaining the process of mosaicking an image to be searched or an unknown image and extracting candidates for searching from the unknown image. In FIG. 2, reference numeral 101 denotes a search target image input unit that uses a TV camera or a scanner to input a search target object, for example, a typical apple image in the case of fruit apples.
Input to the search target image buffer 102. Therefore,
The content of this buffer 102 is a set of pixels represented by grayscale or color.

This search target image is represented by a matrix F = [f], and as shown in FIG. 3, it is divided into M × N by a block having a size of Wb pixels × Wb pixels, and each block FMmn of this mosaic image FM. Let be represented by a representative value in the block, for example, an average value in the block if it is a shade, and a color having the maximum frequency in the block if it is a color.

[0011]

[Equation 1]

[0012] Search target image mosaic unit 103
Stores the contents of the buffer 102 as search target mosaic data in this procedure and stores it in the search target image dictionary 104. It should be noted that, if necessary, the average value of a plurality of search objects, in this example, an image of apples, may be taken as representative apple mosaic data, or a plurality of search object images may be prepared.

On the other hand, the unknown image U is stored in the unknown image input unit 201.
It is taken into the unknown image buffer 202 via. Therefore, as shown in FIG. 4, the unknown image U is divided into P × Q with a block size of Wb pixels × Wb pixels by the unknown image mosaic unit 203, and a representative value is calculated for each block to generate a mosaic image UM. obtain. The unknown image input unit 201
May be the same as the search target image input unit 101.

Therefore, the mosaic of the unknown image is scanned to search for a location that matches the mosaic to be searched. That is, it is based on the following principle. When Ui is arbitrary M × N mosaic data among P × Q mosaic data of the input image, U
The position of Ui when the distance Di between i and the mosaic data FM to be searched is the minimum value is taken as the search result.

There are various ways to obtain the distance Di, but a typical one is the Euclidean distance. This Euclidean distance is calculated as follows.

[0016]

[Equation 2]

Specifically, it is as follows. Reference numeral 204 denotes a candidate selection unit that selects candidate data within the frame of M × N blocks from the unknown image mosaic unit 203. Generally, the upper left M × N unknown images are generally selected. This mosaic data and the search target image dictionary 104
The mosaic data to be searched from is input to the distance calculation unit 300, and the distance is calculated. This result is stored in the position / distance storage unit 301 together with the M × N position data in the unknown image mosaic UM at this time.

Further, the candidate selecting section 204 selects the next candidate from the unknown image mosaicizing section 203, and the distance calculating section 3
00 calculates the distance at this time and stores it in the position / distance storage unit 301 together with the candidate position. Similarly, M × N candidates are sequentially selected for the entire region of the unknown image mosaic, and the selected position and the calculated distance are all stored in the position / distance storage unit 301.

By the way, in the above, in the mosaicing of the unknown image, the same block size as the image to be searched, Wb pixels × Wb pixels, was used. If the search target image and the search target image in the unknown image have the same size, this size may be left unchanged, but generally, the size of the search target image in the unknown image is also unknown. Therefore, when mosaicing an unknown image, the number of pixels of the block size is, for example, 1 in Wb.
Several types such as / 5 to 10/5 are taken, and the above procedure is repeated for each block size, and the obtained distance and the position at that time are stored in the position / distance storage unit 301.

Also, when a plurality of search target images are stored in the dictionary, the above procedure is similarly repeated for each search target image, and the obtained distance and the position at that time are stored in the position / distance accumulation. Store in part 301.

In this way, when the distances in all cases are stored in the position / distance storage unit 301, the minimum distance detection unit 302
Of the distance data in the position / distance storage unit 301,
Find the one with the minimum value. This value is sent to the search target presence / absence determining unit 303, and when this value is less than or equal to a threshold value, it is determined that the target image exists in the unknown image, its position is extracted from the position / distance accumulating unit 301, and the result is output as a result. Part 304
Output to. On the other hand, when this value exceeds a certain threshold value, it is determined that the target image does not exist in the unknown image, and “no” is output to the result output unit 304.

[0022]

As described above, by mosaicing and searching a search target image and an unknown image, it is possible to apply a special condition such as uniformity to the background, and
It is possible to easily find the target image without causing an error due to noise when the line segment is used. In particular, it is effective for objects having similar shapes and colors, such as fruits, human faces and butterflies.

[Brief description of drawings]

FIG. 1 is a principle configuration diagram of the present invention.

FIG. 2 is a block diagram illustrating the entire invention.

FIG. 3 is a diagram showing a mosaic of a search target image.

FIG. 4 is a diagram in which an unknown image is mosaiced.

[Explanation of symbols]

 101 Search Target Image Input Unit 102 Search Target Image Buffer 103 Search Target Image Mosaicing Unit 104 Search Target Image Dictionary 201 Unknown Image Input Unit 202 Unknown Image Buffer 203 Unknown Image Mosaicing Unit 204 Candidate Selection Unit 300 Distance Calculation Unit 301 Position / Distance Accumulation unit 302 Minimum distance detection unit 303 Search target presence / absence determination unit 304 Result output unit

Claims (2)

[Claims] 1. A first means for dividing a grayscale or color natural image to be searched into M × N in advance, calculating a grayscale or color representative value of each block, and storing it as dictionary data. Second means for dividing the input image into P × Q and calculating the representative value of each block when the natural image is given as an input, and third means for extracting an arbitrary M × N area from the second means. And fourth means for calculating the distance formed by the two representative value groups obtained by the first and third means and accumulating the distance, and repeating the third and fourth means over the entire area of P × Q. Fifth means of applying the obtained minimum value from the obtained distances, and when this minimum value is smaller than a threshold value given in advance, if the search target image is at the position derived from the minimum value in the input image If the minimum value is larger than the threshold value, it is detected in the input image. Image retrieval apparatus characterized by having a sixth means for determining a target image does not exist. 2. In the first means, a target image is preliminarily set as a plurality of similar but not unique images, and each image is divided into M × N to calculate a representative value of each block. 2. The image retrieval apparatus according to claim 1, wherein an average value is calculated for each corresponding block among a plurality of images and the average value is stored as a dictionary.