JP2005286799A - Index image generator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate an index image with uniform and good quality in a short time.
In an index image generating apparatus, a DSP or the like first captures an image used for index generation in step S1000, and then performs edge detection processing using differentiation in step S1100. At that time, if the edge cannot be clearly detected, edge enhancement processing is performed. Next, in step S1200, a background region other than the closed space surrounded by the edge is filled, and in next step S1300, an optimization process, which is adjustment of the position and area of the closed space, is performed, and in step S1400, the image is displayed. Smoothing process is performed. Then, the index image generated in this way in the last step S1500 is registered in the directory server.
[Selection] Figure 3

Description

  The present invention relates to an index image generation apparatus that generates an index image used for searching for a still image in an image database system that handles still images and the like.

In an image database system that handles still images and the like, there are cases where, for example, high-definition still images of works of art such as butterflies and Edo Kosode are targeted. In such a case, it is required to search and display a high-definition still image using an image as a key. In a conventional image database system, a user generates an index image, which is a sample image, one by one. (For example, refer to Patent Document 1).
JP 2001-14333 A

  By the way, as described above, when an image database system that handles still images, etc., adopts a method for searching and displaying high-definition still images using images as keys, the system performance is affected by the quality of index images. Therefore, it was necessary to generate the index image manually with great care.

  However, in the past, since the user manually generated the index image from the image one by one, it may take several hours to generate one index image, which requires considerable time and effort to generate the index image. There was a problem that it was a necessary work.

  In addition, an index image used as a search key for searching and displaying these high-definition still images can be a high system regardless of the index image if pre-processing for removing noise components in the index image is performed. Although there is a high possibility that performance can be obtained and the significance and importance of the preprocessing are well recognized, if such preprocessing is performed manually by the user one by one, it requires considerable time and labor. There was also a problem that.

  The present invention has been made in view of such a problem, and an object of the present invention is to provide an index image generating apparatus capable of generating an index image having uniform and good quality in a short time.

  In order to solve the above-described problem, an index image generation device according to the present invention includes an edge detection unit that detects an edge of a target image to be an index, and a filling unit that fills a background region other than the target image surrounded by the edge. The structure which has these is taken.

  According to this configuration, since the background area other than the target image surrounded by the edges is filled, the noise component existing in the background area can be removed, and an index image with uniform and good quality can be obtained in a short time. Can be generated.

  Further, the index image generation apparatus of the present invention employs a configuration that further includes optimization means for limiting and optimizing the size and position of the target image surrounded by the edge.

  According to this configuration, since the size and position of the target image surrounded by the edges are two-dimensionally limited, the size and position of the target image are optimized as an index image.

  Moreover, the index image generation device of the present invention employs a configuration that further includes a smoothing unit that smoothes the target image surrounded by the edges.

  According to this configuration, the target image surrounded by the edges is smoothed, so that noise, shadows, and shades included in the target image can be removed, and an index image having a uniform and good quality can be obtained. It can be generated in a short time.

  In the index image generating apparatus according to the present invention, in the configuration described above, the edge detecting unit detects an edge by multiplying a luminance component forming an edge pattern by a predetermined coefficient when detecting an edge of the target image to be an index. Take the configuration.

  According to this configuration, since the luminance component forming the edge pattern is multiplied by the predetermined coefficient, the edge is emphasized, and the edge can be detected even when the edge is not clear.

  Further, the index image generating apparatus of the present invention employs a configuration further comprising registration means for registering the index image in the directory server in association with the address of the target image.

  According to this configuration, since the index image generated as described above is registered in the directory server in association with the address of the target image, an index image having uniform and good quality can be registered in the directory server in a short time. can do.

  The index image generation method of the present invention includes a step of detecting an edge of a target image to be an index, and a step of filling a background area other than the target image surrounded by the edge.

  According to this method, since the background region other than the target image surrounded by the edges is filled, the noise component existing in the background region can be removed, and an index image with uniform and good quality can be obtained in a short time. Can be generated.

  The index image generation program of the present invention causes a computer to execute a function of detecting an edge of a target image as an index and a function of filling a background area other than the target image surrounded by the edge. .

  According to this program, the background area other than the target image surrounded by the edges is filled, so that noise components existing in the background area can be removed, and an index image with uniform and good quality can be created in a short time. Can be generated.

  As described above, according to the present invention, an index image having uniform and good quality can be generated in a short time.

(Embodiment)
The essence of the present invention is that when generating an index image, the edge of the target image is detected, the background area is filled, the arrangement of the target image is optimized, and a smooth image is generated. An index image having a high quality is generated in a short time.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram showing a configuration of an image database system 100 including an index image generation device 140 according to an embodiment of the present invention.

  In FIG. 1, the image database system 100 includes an image server 110, a directory server 120 that provides addresses of a plurality of images stored in the image server 110, and one or more units (here, one unit is used for convenience of explanation). )) And the index image generating apparatus 140 according to the embodiment of the present invention are connected to each other via a broadband network (hereinafter referred to as a network) 150 such as an optical fiber network.

  The image server 110 has an image database 111 that stores a plurality of high-definition images.

  The directory server 120 provides addresses of a plurality of images stored in the image server 110, and has an index table 121 in which a digital code sequence 122, an index image 123, and address information 124 are stored.

  The digital code string 122 is obtained by converting an index image 123 representing an image file in the image server 110 into a digital code string for image search and registering it in advance.

  The address information 124 indicates an address in the image database 111 of the image file corresponding to the index image 123. The address information 124 may be physical or logical. It is not limited to the address information, but may be key information relating the index image 123 and the image in the image database 111.

  The user terminal 130 has a graphic drawing program having a man-machine interface, and is configured to have an LCD-type or PDP-type HDTV display suitable for high-definition still image display. When handling high-definition still images, it is required to faithfully reproduce the color of museum information, such as Edo Kosode, using a graphics drawing program. Consists of items suitable for display.

  FIG. 2 is a block diagram showing a configuration of the index image generating apparatus 140 according to the embodiment of the present invention.

  In FIG. 2, the index image generating apparatus 140 is connected to a network 150 to perform communication processing, a communication interface (I / F) 141, a CPU 142, a memory 143, and a hard disk drive (HDD) 144 storing a program. And a DSP 145 for generating an index image and an external I / F 146, and a display 147, a camera 148, and the like are connected via the external I / F 146. Note that the DSP 145 functions as an edge detection unit, a painting unit, an optimization unit, and a smoothing unit of the present invention by firmware incorporated in the DSP 145. The CPU 142 functions as a registration unit by executing a program stored in a hard disk drive (HDD) 144.

  Next, in the image database system 100, an image search procedure will be specifically described by taking a butterfly image search as an example.

  When the user tries to retrieve the target butterfly image from the image server 110, the user first sketches the target butterfly on the user terminal 130 (see (1) in FIG. 1).

  Next, the sketch drawn by the user is transmitted from the user terminal 130 to the directory server 120 (see (2) in FIG. 1).

  The directory server 120 converts the sketch image received from the user terminal 130 into an image search digital code string 125 (see (3) in FIG. 1), and the digital code string 125 is stored in the index table 121. The search is performed by comparing with the digital code sequence 122 for image search corresponding to the index image 123 (see (4) in FIG. 1).

  As a result of the comparison, several index images 123 corresponding to the similar digital code strings 122 are selected (see (5) in FIG. 1) and displayed on the user terminal 130 (see FIG. 1). 1 (6)).

  The user selects the index image 123 that is considered to represent the target image data from the index images 123 displayed on the user terminal 130 (see (7) in FIG. 1).

  Then, the image data associated with the index image 123 is distributed from the image server 110 to the user terminal 130 and displayed (see (8) in FIG. 1).

  As described above, according to the image database system 100, a sketch image or the like in which an image stored in the image server 110 is freely drawn by a user is used as key information, and the target image is searched for compared with the index image 123. The success or failure of the image search is determined by the quality of the index image 123.

  Therefore, in this embodiment, the user does not generate the index image 123 from the images one by one, but the index image generation device 140 automatically generates an index image with uniform and good quality in a short time. It is what I did.

  Next, the operation of the index image generation device 140, which is a feature of the present embodiment, will be described with reference to the flowchart shown in FIG.

  FIG. 3 is a flowchart showing the operation of the index image generating apparatus according to the embodiment.

  This index image generation apparatus shows an index generation process using arithmetic processing by the DSP 145 or the like. By generating an index image through the above process, an index image having a uniform and good quality can be obtained, and the generated index image can be used by the directory server 120.

  The automatic index generation method in the image database system according to the present invention performs image capture processing, edge detection processing, background filling processing, layout optimization processing, and image smoothing processing. The layout optimization process and the image smoothing process may be omitted, but by performing these processes together, an index image with more uniform and good quality can be generated. Thus, the generated index image can be used by the directory server 120.

  Hereinafter, these processes will be described in detail with reference to the flowchart of FIG.

(Image import processing)
First, an image capturing process used for index generation in step S1000 is performed.

  Specifically, the index generation device 140 captures an image to be indexed by the camera 147 and captures an image used for index generation.

  The image used for index generation is not limited to the image captured by the camera 147, but may be an image registered in the image server 110 or another server (not shown). That is, for example, when an index image is not generated for an image registered in the image server 110 and the index image is not registered in the directory server 120, the image stored in the image server 110 is stored. Of course, the index image may be generated by downloading.

(Edge detection processing)
Next, an edge detection process using the differentiation in step S1100 is performed. In this edge detection process, an edge enhancement process may be added as necessary for the following reason.

  That is, the edge pattern is composed of luminance components in each pixel (pixel) in the image frame. Therefore, the DSP 145 functions as an edge detection unit, generates an edge pattern by differentiating and synthesizing the luminance component in each pixel (pixel) in the x-axis direction and the y-axis direction. By overlapping and drawing, it can be clearly identified as an edge. Also, by making the curve forming the edge thick, a closed space is formed by the edge, so that the target image and the background can be distinguished.

  In the present embodiment, the edge of the target image is detected by the following several edge detection methods that extend the basic concept. The outline is as follows.

  For example, with respect to the x component and the y component of the function f [i, j] at the point (i, j) in the xy coordinate space, the differentiation (difference) is defined by the following equation.

Δ _x f = f [i + 1, j] −f [i−1, j] (Formula 1)
Δ _y f = f [i, j + 1] −f [i, j−1] (Expression 2)

  Next, the gradient g [i, j] of the function f [i, j] is taken into consideration in order to eliminate the edge directionality (x axis, y axis) in (Expression 1) and (Expression 2). Define the following function.

g [i, j] = [ (Δ x f) 2 + (Δ y f) 2] 1/2 ··· ( Equation 3)
g [i, j] = | Δ x f | + | Δ y f | ··· ( Equation 4)
g [i, j] = max (| Δ x f |, | Δ y f |) ··· ( Equation 5)

Therefore, a spatial filter function g [i, j] representing an edge is generated by Δ _x f and Δ _y f according to the formula (formula 3), (formula 4), or (formula 5).

  An edge detection operation is executed by applying an operator (differential operator by computer) generated by expanding these spatial filters to a region including 3 × 3 neighboring pixels in a two-dimensional space.

  For edge detection, operators such as Prewitt filter and Sobel filter (differential operator by computer) are used. Prewitt filters, Sobel filters, and the like are excellent in making the edges stand out, and are suitable for generating index images used in the present invention.

On the other hand, Laplacian taking into account the second derivative has excellent characteristics for edge detection and image definition improvement. That is, Laplacian ▽ ² f is defined by the following (formula 6).

▽ ² f = ∂ ² f / ∂x ² + ∂ ² f / ∂y ² (Formula 6)

Here, if the first differentiation is performed with i + 1 and i and the second differentiation is performed with i and i-1, ∂ ² f / ∂x ² is given by the following (formula 7).

^{∂ 2 f / ∂x 2 = f} [i + 1, j] + f [i-1, j] - 2f [i, j] ··· ( Equation 7)

Accordingly, Laplacian ▽ ² f is expressed by the following (formula 8).

▽ ² f = f [i-1, j] + f [i + 1, j] + f [i, j-1] + f [i, j + 1] -4f [i, j]
... (Formula 8)

  Next, a process of generating an image without blur from an image with blur is executed using Laplacian as follows.

  Let B (x, y, t) be a function representing an image without blur, and f (x, y, t) be an image containing blur. Assuming that B (x, y, t) can be expanded in series near t = τ and expressed by Taylor expansion of f (x, y, t), the following (Equation 9) is obtained. It is done.

B (x, y, t) = f (x, y, τ) -τ∂f / ∂t + τ 2/2 · ∂ 2 f / ∂t 2 - ···
... (Formula 9)

  Here, if the blur of the image is expressed by a model of the diffusion process, ∂f / ∂t is given by Laplacian. That is, the following (Formula 10).

∂f / ∂t = k∇ ² f (Expression 10)

  Here, if a second-order or higher term is omitted for τ in (Expression 9), B (x, y, t) from (Expression 9) and (Expression 10) is given by (Expression 11) below. .

B (x, y, t) = f (x, y, τ) −kτ∇ ² f (Equation 11)

  As described above, since the image can be made high-definition with the Laplacian, it is difficult to form an edge. For example, it can be used for distinguishing the same color pattern and the background.

Operators such as the Prewitt filter and Sobel filter described above (differential operators by computer) are applied to Δ _x f and Δ _y f. An operator of a Laplacian filter (a computer differential operator) is applied to two-dimensional second-order differentiation. In the above description, each filter is 3 × 3, but 5 × 5 or more may be used.

  As an example, in the simulation result by MATLAB, a favorable result is obtained in the case of the Sobel filter and the Laplacian filter. As a result of conducting a preliminary experiment using both filters based on this simulation result, a method of applying an Sobel filter to the x-axis and y-axis and generating an edge by the sum of their absolute values was invented. As a result of the experiment, the superiority of this absolute value sum method was confirmed. Using this result, index generation performed using a Texas Instruments DSP (TMS32C6416) was also performed by applying the Sobel filter, and good results were obtained.

(Edge enhancement)
For a special image in which the edge cannot be clearly detected, a process for enhancing the edge is added during the edge detection process in step S1100.

  Edge enhancement is generated by multiplying the value of the luminance component (Y signal component) forming the edge pattern by the proportional coefficient K. Normally, values such as K = 1.3 and K = 1.5 are desirable values, but are not limited to these values. In a demonstration experiment conducted as an example, an edge is emphasized without affecting the background in an image obtained by multiplying an edge pattern by a proportional coefficient (K = 1.3, 1.5).

  Further, if necessary, there is a method for enhancing the γ characteristic of the image. There are two ways to do this.

The first method is as follows.
(1) The γ value is set larger than 1, and an index image is generated by the method of the present embodiment.
(2) An edge pattern is formed using the generated index image, and the background is filled with a single color to verify the integrity as a sample image.

The second method is as follows.
(1) An edge pattern is formed, and the background is painted white to generate a sample image with γ = 1.
(2) Based on the generated sample image of γ = 1, the γ value is set to be larger than 1, and an index image is generated by the method of the present invention.

  In the demonstration experiment conducted as an example, the γ value was set larger than 1 by the first and second methods, respectively, and the index image was generated by combining the γ correction for the still image and the generation of the normal edge pattern. In a normal image, an image having a value in the range of γ of 2.5 or less is observed without a sense of incongruity.

(Background painting)
Next, a background painting process that is outside the closed space in step S1200 is performed.

  That is, the DSP 145 functions as a filling means, and fills the background area outside (or inside) the closed space surrounded by the edges with a single color having a uniform level and hue. As a result, the background noise component existing outside the target still image can be removed. Here, the noise includes unnecessary image components and pattern noise in addition to random noise on the image.

  For example, it is desirable that the background of an image with an edge of a small sleeve mainly composed of red is a signal of a uniform level and hue such as white or gray. In this way, since the hue is greatly different between the small sleeve and the background, the small sleeve and the background can be easily separated by the edge.

  Therefore, it is possible to form a closed space by the edge pattern, and fill the background which is outside the closed space with any single color such as white, blue or gray. In the image search, for example, a background filled with a reference color such as white is used.

  As described above, in the example of the small sleeve in which the background is painted white by the processing in step S1200 after the edge is detected by the processing in step S1100, the background portion is obtained by painting the background in white after the edge is detected. The noise at is automatically removed.

  From these experimental findings, it is clear that in actual processing, only the closed space surrounded by the edges needs to be recognized as a small sleeve.

(Optimization processing for index image placement)
Next, an optimization process of index image arrangement in step S1300 is performed.

  In this optimization process, the DSP 145 functions as an optimization unit, and controls the size and position of the closed space recognized by the edge in a two-dimensional manner by limiting the x-axis direction and the y-axis direction. Then, the closed space area is maximized and the closed space position is optimized.

  For example, in the image of the small sleeve, since the skirt is arranged in parallel with the x-axis, the skirt line is recognized as the reference line. Therefore, the axis orthogonal to the x axis is taken as the y axis. A quadrilateral surrounded by two orthogonal straight lines parallel to the x-axis and the y-axis, respectively, is an area occupied by the small sleeve and the background that are the search target images.

  The vertical size of the small sleeve is enlarged and adjusted so that the small sleeve that is the search target image is exactly inserted in the vertical direction of this quadrilateral, and the small sleeve on the background is adjusted so that the left and right margins are equal. Adjust the position and place the small sleeves in place. At this time, the background other than the small sleeves is painted with the reference color (in the present invention, white) by the process of step S1200 described above. This operation is herein referred to as optimization of index image arrangement. In an example of an optimization experiment for the arrangement of sample images of Kosode, the practicality of this processing method was confirmed.

(Generate smooth image)
Next, the image smoothing process in step S1400 is performed.

  In this smoothed image generation processing, in order to automatically register the smoothed image as an index image in the directory server 120, the DSP 145 functions as a smoothing means to maximize the size and position of the closed space generated by the edge. Following the optimization, a smoothed image is generated for the image generated by the closed space. This makes it possible to remove noise, shadows, and shades contained in the target image.

  That is, when a closed space including the search target image is formed by the edge pattern, the edge pattern surrounds the search target image. If the search target image includes shades of unnecessary shadows, too fine patterns, or high spatial frequency noise, the search success rate may be reduced. Therefore, in order to generate an index image from which these unnecessary components in the search target image are removed, a spatial averaging process by the DSP 145 is desirable. In the smooth image of the small sleeves generated by this spatial averaging process, unnecessary components are removed.

(Automatic index image registration)
Next, the index image automatic registration processing in step S1500 will be described.

  The index image generated as described above is registered in the index table 121 of the directory server 120 when the CPU 142 functions as a registration unit in association with the address of the high-definition still image registered in the image server 110. .

  When the index image is transmitted from the index image generating apparatus 140 to the directory server 120, the index image may be transmitted as it is. However, in order to reduce the transmission data amount, the index image may be transmitted after being reduced. .

  Therefore, in automatic registration, an index code representing a feature amount generated from a high-definition still image is registered in the index table 121 of the directory server 120 in association with the address of the high-definition still image stored in the image server 110.

  FIG. 4 is an explanatory diagram for explaining the process until the index image is registered in the directory server 120 from the index image generating apparatus 140 when the target image to be indexed is a small sleeve image.

  That is, as shown in FIG. 4, the index image generating device 140 reads the original content 401 of the small sleeve photographed by the camera 147 or the like by the application executed by the DSP 145 firmware, extracts the edges, and is surrounded by the edges. The background area other than the target image area is filled, the layout optimization process and the target image smoothing process are performed, and the image is reduced as necessary and transmitted to the directory server 120.

  Also in the directory server 120, the index image from the index image generation device 140 is reduced by the server application as necessary, and is associated with the address of the high-definition still image stored in the image server 110, and the directory server as the index image 402 It is registered in the index table 121 of 120.

  In the system according to the present embodiment, the automatic registration program is created by a Java (R) program or the like, and all image data of Kosode is registered.

  About 300 small sleeve sample images are stored in the image server 110 and the directory server 120 together with the index images, respectively, so that the database image can be searched. As a result of using the automatically registered index image for search display and verifying the system performance, a good performance of 99.6% accuracy was confirmed.

  As described above, according to the present embodiment, by automatically generating an index image used in the image database system 100, an index image having uniform and good quality can be generated in a short time.

  As a result, according to the present embodiment, it is possible to obtain the effects of improving the success rate of the search using the index image as the search key and reducing the cost of generating the index.

  In other words, by automating edge detection and background area filling, it is possible to eliminate background noise components that exist outside the still image to be indexed. Image components and pattern noise can be removed. As a result, it is possible to improve the success rate of search using such an index image as a search key.

  In addition, by automatically registering the smoothed image as an index image in the directory server 120, it is possible to remove noise, shadows, and shades. It becomes possible to improve.

  Using the closed space formed by the edges generated by image differentiation, maximizing the closed space area formed by this target image and optimizing the closed space position, the index generation work time can be reduced. Since it can be shortened, it is possible to reduce the cost of index generation in the generation of still image content.

  In the above-described embodiment, the user terminal 130 that draws a sketch image and performs an image search and the index image generation device 140 that generates an index image are described as separate devices. However, the present invention is not limited to this. Instead, the user terminal 130 that performs image search may have the function of the index image generation device 140 that generates an index image. That is, the user terminal 130 does not draw a sketch image and performs an image search using this as a search key, but automatically performs an index image from an image captured from another terminal or server via a camera, a scanner, or the Internet. It is of course possible to search for an image using this as a search key. In this way, the user does not need to draw a sketch image, and an index image can be generated from an image taken from a camera or scanner, and the image can be searched using this as a search key. An image search of the image can be performed.

  In this embodiment, the image server 110 and the directory server 120 are configured independently and connected via the network 150. However, the present invention is not limited to this, and the image server 110 and the directory server are connected. Of course, the server 120 may be configured as the same server.

  Further, in the process of generating an image suitable for search by applying the edge detection process in step S1100 and the image smoothing process by the spatial averaging process in step S1400, for example, the following two methods are used. Can be considered

(1) A first method in which an edge pattern is first generated by a Sobel filter, and then an image is smoothed by a spatial averaging process.
(2) A second method in which an image is first smoothed by a spatial averaging process and then an edge pattern is generated by a Sobel type filter.

  That is, the edge detection process in step S1100 and the image smoothing process by the spatial averaging process in step S1400 may be interchanged.

  As a result of an experimental comparison of the first and second methods described above, it was found that the first method is superior when attention is paid to noise, and the second method is superior when attention is paid to contours. Therefore, both of these methods are effective.

  In the above embodiment, the entire image of Edo Kosode has been described as an example of the target image. However, since many designs drawn on Edo Kosode are similar, they are drawn on a part of Edo Kosode. It is also possible to recognize and extract a crane design, which is a partial image, and use it as an index image. In this case as well, by applying the method described above, as with the pattern of the entire image, a design recognized as a partial image is selected, and the position and size are automatically controlled to generate an index image. Is possible. That is, the design of a crane, which is a partial image drawn on Edo Kosode, is selected as a pattern, and an image that can be used as an index is generated by the above-described method using the selected design of the crane.

  As an example, in a demonstration experiment that picked up and extracted a pattern of cranes drawn on Edo Kosode, an edge was extracted by applying a Sobel filter to the original image. Was superimposed on the color signal, and the background was painted or the image of the crane was smoothed. The experimental result was good and it was confirmed that the pattern of the partial image can be used as an index image.

  The index image generation apparatus according to the present invention has the effect of being able to generate an index image with uniform and good quality in a short time, index generation as part of information management in a museum, It is useful as a technique for generating index images for teaching materials in the educational industry, schools, etc., and for general purposes of index generation in image database systems using image search.

1 is a block diagram showing a configuration of an image database system including an index image generation device according to an embodiment of the present invention. The block diagram which shows the structure of the index image generation apparatus which concerns on embodiment The flowchart which shows operation | movement of the index image generation apparatus which concerns on embodiment Explanatory drawing which shows until a small sleeve image is registered as an index image

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Image database system 110 Image server 120 Directory server 130 User terminal 140 Index image generation apparatus 142 CPU (registration means)
145 DSP (edge detection means, painting means, optimization means, smoothing means)

Claims (7)

Edge detection means for detecting the edge of the target image as an index;
Painting means for painting a background area other than the target image surrounded by the edge;
An index image generating apparatus comprising: Further, optimization means for limiting and optimizing the size and position of the target image surrounded by the edge,
The index image generating apparatus according to claim 1, wherein: Further, a smoothing means for smoothing the target image surrounded by the edge,
The index image generating apparatus according to claim 1, wherein: The edge detection unit detects an edge by multiplying a luminance component forming an edge pattern by a predetermined coefficient when detecting an edge of a target image as an index.
The index image generating apparatus according to claim 1, wherein: Furthermore, a registration means for registering the index image in the directory server in association with the address of the target image,
The index image generating apparatus according to claim 1, wherein: Detecting an edge of the target image as an index;
Filling a background region other than the target image surrounded by the edge;
An index image generation method characterized by comprising: A function for detecting the edge of the target image as an index;
A function of filling a background area other than the target image surrounded by the edge;
A program for generating an index image, which causes a computer to execute.

Images