JP4782610B2 - Satellite image processing system - Google Patents

Description

  The present invention distributes satellite images prepared from a server in advance, which are images taken from a high altitude such as satellite images (hereinafter referred to as satellite images) so as to meet the conditions desired by the requester. The present invention relates to an image processing technique for performing processing for the above. In particular, the present invention relates to a technique that can cope with a condition desired by a requester even if the condition is ambiguous or includes a certain range.

  With the launch and operation of high-resolution commercial optical satellites, high-resolution satellite images can be easily acquired. When you want to acquire only the images that the user really needs from among a large number of images, specify the exact location of the target in advance using a map and set detailed information such as the shooting date. Is essential to search for images.

  Recently, a satellite that superimposes a captured area of a satellite image stored in a server and a map on the Web, identifies a target area requested by the user based on the map, and searches for a captured image in the target area. Image search systems are becoming popular. FIG. 11 is a block diagram showing an outline of a general satellite image search system. The satellite image search system distributes a map, an image / search screen, and the like to the web server 1103 and requests a satellite image by specifying a desired target area while displaying the map and the web server 1103 in response to the request. A Web client 1101 having a display unit for displaying images distributed from the side, and an operation unit for performing various operations, and an Internet 1102 for transmitting and receiving data between the Web server 1103 and the Web client 1101. is doing.

  A Web server 1103 that distributes maps, images / search screens, etc. includes a search processing unit 1104 that searches for satellite images, a transmission / reception unit 1105 that performs data transmission / reception processing, a map server 1106 that stores map data, And an image server 1107 for storing area data of captured satellite images.

  In the satellite image search system having the above-described configuration, when information for specifying a search point (for example, latitude, longitude and date / time of four corners of an area) is input from the Web client 1101, a transmission / reception unit is transmitted via the Internet 1102. Information is received in 1105, and the search processing unit 1104 acquires map data and photographed area data of the corresponding point from the map server 1106 and the image server 1107 based on the above information, and passes the data to the transmission / reception unit 1105. . Thereafter, the transmission / reception unit 1105 transmits data to the Web client 1101 via the Internet 1102 and displays the data on the display unit. Next, based on the search result, a process flow of separately creating a product image (a raw image obtained by performing various correction processes) from the raw image (data at the time of shooting) is performed.

  Patent Document 1 describes a system to which a mosaic function and a product image display function are added in addition to the functions of the satellite image search system, and Patent Document 2 describes a satellite image distribution system using a dedicated line network.

Further, Patent Documents 3 and 4 describe a technique for creating a composite image by combining categorized images.
Japanese Patent No. 3394502 JP 2002-112226 A JP 2005-242566 A JP 2001-326806 A

  By the way, satellite images are taken every day in the world and accumulated in a database. Therefore, even if limited to high-resolution satellite images, the number is enormous. However, the data that characterizes these images is only data relating to the situation at the time of image shooting such as latitude / longitude information of the four corners of the image or shooting time.

  Therefore, when trying to acquire an image of a specific area, search for the image from the latitude / longitude information of the specific area, narrow down the target image according to the shooting time, etc., and visually check the thumbnail of the image. The procedure of confirming will be taken. In such a procedure, the situation (geographical position and time) when the image was taken is the so-called “1”, which is whether it meets the requirements such as latitude / longitude information, photographing time, etc. or not. The search result is “0”. Therefore, there is a problem that it is impossible to meet a demand for searching for an image having a condition close to the required condition. However, in reality, the shooting situation regarding the desired image is ambiguous, and there are many cases where it is not possible to completely specify the required conditions, or when it is desired to search using only the state of the actually captured image as a search condition. .

  For example, there are cases where it is desired to search for an image with some of the search conditions including ambiguity, such as “an image with relatively many mountains, the season being around winter, and a lot of snow”. Further, there is a case where the user wants to search without completely specifying the situation at the time of shooting, such as “Search in order from the latitude / longitude is XX, and the shooting date is from XX month in XXXX.”

  If a search as described above can be performed, it may be very convenient for general users, and it is easy to use.

  In addition, the satellite image distribution system with a mosaic function (image stitching function) is easy to use because it can be used on the Web, and it can be selected from multiple images and displayed as a new product image. Easy to use because it can. However, when narrowing down a specific area, the clues are the latitude and longitude information and the shooting time, etc., which is the same as the conventional one, and the search with ambiguous conditions and the search by the feature of the image The problem that it is not possible remains.

  An object of the present invention is to provide a satellite image processing technique that allows an ambiguity in conditions for searching for satellite images.

According to an aspect of the present invention, there is provided an image processing apparatus that processes an image prepared in advance in a raw image server that stores a raw image taken from a high sky so as to meet a search condition desired by a Web client. A category database for storing a category determined as an index for associating the search conditions with each other, and a search condition classified for each category;
A reference value relating to how to extract a category extracted from the category database and search condition candidates included in the category based on a request from the Web client, and to consider each of the extracted search condition candidates An image processing apparatus is provided that includes a search processing unit that performs a process of searching for an image in the image server in consideration of the allowance indicating As a result, it is possible to cope with a condition desired by the requester even if it is an ambiguous condition or a condition including a certain range. Furthermore, a product image generation unit that generates a product image based on the search result may be provided in the present apparatus or another apparatus.

  The category database includes a category combination table that displays a combination of categories. The category combination table is identified by a category landform combination ID that uniquely identifies a combination of categories, and the category and the category are included in the category database. And the tolerance set for each of the included search condition candidates. Thereby, update and registration can be performed for each category.

  For example, the terrain ID 303 and the ratio 304 are included. The terrain ID 303 is an ID registered in the terrain table 201 (see FIG. 2 described later), and is an ID for identifying the terrain name. The ratio 304 is a value indicating the ratio of the area corresponding to the terrain name to the entire image. As shown in FIG. 3, it is preferable that about 1 to 10 combinations of the terrain ID 303 and the ratio 304 can be registered for one terrain combination ID 301. In this way, terrain combinations including a plurality of terrains can also be registered. For example, a combination such as “a mountain is 50%, a sea is 30%, and a flat land is 10%” can be registered. At this time, the ratio does not necessarily have to be 100% as a whole, and it is only necessary to indicate how much each terrain occupies with respect to the entire image. As described above, by providing a category combination table, with respect to search condition items classified in the lower order included in one category, the ratio of the lower search condition is considered at the time of image search. Can be set as needed. However, since it can be expressed only by numerical values, such as brightness, and there is no item (search condition) classified in the lower order for a category that is preferable, particularly in such a case, the database It is not necessary to provide a category and its subordinate items individually.

  In the user interface on the Web client side, a transmission unit that transmits data for displaying a user interface that enables the Web client side to set the category, the search condition candidate, and the tolerance for the category. It is preferable to have a receiving unit that receives the set tolerance and sends it to the search processing unit.

  In the search processing unit, the tolerance is set from the client side with a first tolerance value that is set in advance in the system with respect to a shape-related property among properties such as an image or image information related thereto. When the second tolerance value is compared, and the second tolerance value is smaller, it is determined that it is an acceptable range as a search target, and this is used as a search condition. , It can also be excluded from the search target.

  In the case where the tolerance is a characteristic that can specify a numerical range among the characteristics of the image or the image information related thereto, the search processing unit statistically handles the tolerance range. It is possible to use a search condition if it is within the range, and not to be a search target if it is outside the tolerance range.

  The Web client is a search result display area for displaying a search result, and displays thumbnails of images searched in the image server according to an input condition. When a plurality of images are searched, a plurality of thumbnails are displayed. It is preferable to have an edit processing unit that can edit the entire image displayed in the search result display area based on the plurality of thumbnails.

  As described above, the search conditions are classified into categories, and a plurality of search conditions related to each other are associated with one category and stored as necessary. Make it possible to hold multiple categories. For a certain category, an allowance indicating the degree of weighting can be set from the user side for each search condition. As a result, the condition for searching for satellite images can be made ambiguous.

  For example, among the characteristics such as the image or the image information related thereto, the characteristics related to the shape such as the terrain remain as seen. Therefore, the first allowable value set in the system in advance and the second allowable value set from the client side. When the second allowable value is smaller than the allowable value, it is determined that it is within the allowable range as the search target, and this is adopted as the search condition. When the magnitude relationship is reversed, the search target is not set.

  On the other hand, among the characteristics such as the image or the image information related thereto, the shooting range, the shooting date, the situation, the brightness, and the like can specify a range as a numerical value, for example, assuming a normal distribution, If the tolerance is 0.8, if it is within the range of 0.8, it can be handled as a search target.

The satellite image processing system of the present invention has the following effects.
(1) By specifying imaging information of a requested image and a condition based on a category indicating the feature of the image, it is possible to search and create an image having the feature from a large number of satellite images.
(2) By making it possible to specify ambiguity as a search condition, images can be searched even if the requested image is not decided or the request itself is ambiguous, and it is necessary from the result. It is possible to search for the image that is being displayed.

  Hereinafter, a satellite image processing system according to an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a functional block diagram showing a configuration example of an image processing system according to this embodiment. The satellite image processing system according to the present embodiment includes a Web client 101 and a Web server 103 including a category database 110, and the Web client 101 and the Web server 103 are connected by, for example, an intranet 102. In addition, a product image storage 111 is connected to the intranet 102 and is configured to be accessible from both the Web client 101 and the Web server 103.

  The Web client 101 is generally a personal computer (PC), for example. Usually, an interface (GUI) can be displayed by a browser.

  The Web server 103 is connected to a category database 110, a map server 108 that distributes map data, and an image server 109 that distributes thumbnail images of captured images.

  The category database 110 is a database that stores data that are clues when searching for satellite images in a form associated with the images. As will be described later, categories such as terrain, situation, and brightness, and items included in the categories. The search condition to the image server is stored. Furthermore, shooting information such as latitude / longitude, shooting date / time, and resolution of a shot image may be stored.

  For example, the search conditions are classified into categories, and a plurality of search conditions related to each other are associated with one category and stored as needed. Make it possible to hold multiple categories. Further, in the present embodiment, for a certain category, an allowance indicating the degree of weighting can be set from the user side for each search condition. With these ideas, it is possible to add ambiguity to the conditions for searching for satellite images. This point will be described in detail with reference to FIG.

  The map server 108 registers, for example, a map as a background image at the time of searching for an image and a captured range in a polygon format, for example. It is also possible to distribute this. The map as the background image includes a wide range of objects that are clues when searching from large terrain such as continents and lakes to small artifacts such as highways and railways, corresponding to the map position. . This map is realized, for example, by registering data in a shape file or a GIS compatible database developed by ESRI. The contents of the shape file are a collection of points indicating geographic information and represent the shape. Moreover, a label can be attached to the shape as an accompanying file. The raw image server 109 stores the captured raw image, for example, as a thumbnail. The thumbnail is a compressed image file in a format that can be distributed on the Web, such as the JPEG format.

  In response to an image search request from the Web client 101, the Web server 103 updates the search processing unit 104 that performs image search processing, the image generation unit 105 that generates an image based on the search result, and the category database 110. A category update unit 106 and a transmission / reception unit 107 that performs transmission / reception with the Web client 101 are provided. Furthermore, the Web server 103 can refer to the servers 108, 109, and 110, and can acquire data stored in the servers 108, 109, and 110 as necessary. Generally, a temporary storage unit that can store at least temporarily various data retrieved in response to a request from the Web client 101 is provided.

  The product image storage 111 is a storage for storing an image (product image) generated in response to a request from the Web client 101, as will be described later.

  FIG. 2 is a diagram illustrating a configuration example of the terrain table 201 stored in the category database 110. As shown in FIG. 2, the terrain table 201 includes a terrain ID 202 unique to each terrain and a corresponding terrain name 203. The terrain name is one of the categories, for example, defined as a lower search condition included in “terrain”, and is a search condition related to the image and included in the terrain, “flat”, “mountain”, “ The sea ".

  A part of each data is registered as an initial value, and others are additionally registered as necessary when an image search result is received. As the terrain names registered as initial values, for example, terrain names that are generally expected to be frequently used, such as unclassified, flat land, mountains, and sea, are included as search condition candidates. It is desirable that the terrain name is easy to classify automatically, but even if it is difficult to classify automatically, it can be corrected manually, so depending on whether automatic classification is possible or not It is not limited. In addition, it is possible to prepare a value of “Unclassified” for images that are uncharacteristic in terms of topography or images that cannot be identified (cannot be narrowed down) using a single term due to a mixture of multiple terrains. preferable.

  As a standard for new names that are registered based on the results of image search, they are not registered as initial values and can be judged to some extent scientifically. It is desirable that the name be highly usable. However, it is also possible to register terrain that is not expected to be used in only one image. Here, for example, if a newly created terrain called “dune” is assigned to an image extracted as a result of specifying an uncategorized terrain category and performing an image search, the terrain called “dune” After being registered in 201 and selected whether to associate with other terrain, the terrain combination ID registered in the terrain combination table 301 described later and associated with the image in the image classification table 401 is also changed. Details of the terrain combination table 301 and the image classification table 401 will be described later.

  Also, data can be managed by a similar management method for categories other than terrain, such as the situation. However, since it can be expressed only by numerical values, such as brightness, and there is no item (search condition) classified in the lower order for a category that is preferable, particularly in such a case, the database It is not necessary to provide a category and its subordinate items individually.

  FIG. 3 is a diagram showing a configuration example of a category combination table (in this case, a terrain combination table) 301 that displays combinations of categories stored in the category database 110 (FIG. 1). As shown in FIG. 3, the terrain combination table 301 according to the present embodiment includes a terrain combination ID 302 that uniquely identifies a combination of terrain, the terrain ID 303, and a ratio 304. The landform ID 303 is an ID registered in the landform table 201 (FIG. 2), and is an ID for identifying the landform name. The ratio 304 is a value indicating the ratio of the area corresponding to the terrain name to the entire image.

  As shown in FIG. 3, it is preferable that about 1 to 10 combinations of the terrain ID 303 and the ratio 304 can be registered for one terrain combination ID 301. In this way, terrain combinations including a plurality of terrains can also be registered. For example, a combination such as “a mountain is 50%, a sea is 30%, and a flat land is 10%” can be registered. At this time, the ratio does not necessarily have to be 100% as a whole, and it is only necessary to indicate how much each terrain occupies with respect to the entire image. As described above, by providing a category combination table, it is necessary to determine how much of the lower classification included in one category should be considered when searching for images. It can be set according to. Similar to the category table, categories other than the terrain can be similarly configured.

  FIG. 4 is a diagram showing a configuration example of the image classification table 401 stored in the category database 110 shown in FIG. As illustrated in FIG. 4, the image classification table 401 includes an image ID 402, category information 403, and shooting information 404. The situation combination ID is an ID for specifying a combination of situations relating to an image at the time of shooting described later. The situation indicates, for example, the state of the subject to be photographed that is not directly related to the terrain itself such as snow, fresh green, and increased water. In the case shown in FIG. 4, the category information 403 is specified by the terrain combination ID and the situation combination ID. The shooting information 404 is information such as a shooting date that is not related to the image itself.

  Note that the image ID 402, the category information 403, and the shooting information 404 are preferably registered immediately after the image shooting because they are easily forgotten later. However, regarding the category information 403, it is preferable that manual correction is possible because there is a possibility that manual correction is performed after automatic registration of information once immediately after shooting.

  As described above, the image classification table 401 is a table for associating the category information 403 and the shooting information 404 with an image. By creating this table, for example, terrain: “mountain”, situation: “snow” Information such as brightness: “3”, shooting date: “December 31”, resolution: “0.6 m”, and the like can be associated with an image. It should be noted that categories having options such as terrain and situation can be selected and associated from combinations registered in each category combination table.

  Since brightness is a very important characteristic in an image, the brightness level is determined as necessary, and the brightness level is registered as a numerical value. For example, the brighter side is set to 1, the darker side is set to 10 or the like, the brightness is evaluated in 10 levels, and registered as a numerical value. It should be noted that items other than brightness are handled in the same manner with respect to a category in which information is associated with only numerical values.

  FIG. 5 is a diagram illustrating a configuration example of an image search screen 501 displayed on the display unit of the Web client 101 when performing an image search using a category. As shown in FIG. 5, the image search screen 501 includes a search result display area 502, a search result thumbnail 503, a search range condition display component 504, a shooting date condition display component 505, a terrain category condition display component 506, A situation category condition display component 507, a brightness condition display component 508, a condition addition button 509, a redo button 510, a return button 511, and a decision button 512 are configured.

  As shown in FIG. 5, in the search result display area 502, images searched according to the conditions input by the user on the GUI shown on the right side of the screen are displayed as search result thumbnails 503. If a plurality of images are found as a result of the search, a plurality of search result thumbnails 503 are displayed. As a result of the search, each hit image is given a score. For example, the images are displayed in order from the upper left to the lower right in descending order. The score is set in a descending order from the one that hits under a condition with a high “search priority” and the one that has a high “requirement degree” within the same condition. The “search priority” is a degree indicating which condition is preferentially handled among a plurality of set search conditions, and the “requirement suitability” is that requested for each condition. It shows how much they match. For example, if the search score criteria are set to 300, 200, and 100 in descending order of “search priority”, the score is 300 when the “relevant requirement” is 90%, 50%, and 70%, respectively. × 0.9 + 200 × 0.5 + 100 × 0.7 = 440. On the other hand, when the “required fitness” is 70%, 90%, and 90%, 300 × 0.5 + 200 × 0.9 + 100 × 0.9 = 420, and 440 is larger than 420. Will be displayed first (before). In this way, by devising an arithmetic expression so that the required adaptability multiplied by the higher search priority greatly affects the calculation result, the “required adaptability” is higher under two of the three conditions. The device is devised so that, in one condition having a higher priority than the case, a higher “required fitness” can be easily found in the search result. If the device can be set in this way, the setting method of the score reference is arbitrary.

  In addition, an allowable value (reliable: 1 to ambiguous: 0) is set as a setting item so that the search is performed with a certain accuracy. If the “required fitness” is a value less than that, it can be removed from the search results. It is preferable to set a calculation method for the allowable value for each target condition. For example, when a mountain is first specified in the terrain and the allowable value is specified as 0.8, the ratio value is used as it is as the “required fitness”, the mountain is associated with the image, and the ratio is 80 Only images set at% or higher are searched. The method according to the present embodiment can be applied to any system that is categorized as a type that allows a condition to be selected regarding the appearance of an image.

  Next, when the brightness is 3 and the allowable value is specified as 0.7, a normal distribution is assumed in which the center position of x in f (x) is 3, and the “required fitness” is When the brightness value is viewed in a normal distribution, the brightness value is given depending on where the brightness value comes. This method can be applied to a category of a type in which conditions are given only by numerical values.

  If the shooting date is January 1st and the allowable value is 0.5, January 1 will be the starting point “Required adaptability = 1” and the end of July will be the end point “Required adaptability = 0”. For example, a cosine curve is defined, where each date is located on the curve, and a “required fitness” is calculated. This method is preferably applied to categories that handle values that circulate by year or month, such as dates. The method described above is a method for calculating a score based on “search priority” and “required fitness”.

  In this example, the “search priority” is higher as the position of the search condition display component is higher, and they can be replaced by a mouse operation. It is also possible to delete by pressing the delete button. By doing so, the priority of the search condition can be arbitrarily changed even after the condition is once set. Further, after the search condition is set and the priority of the condition is changed, the display result of the search result display area 502 is also immediately changed. This makes it easy to perform searches one after another while checking how search results change when conditions are added / deleted / replaced.

  When the add button 509 is pressed, a process for displaying a condition designation screen 601 for adding a new search condition to the current search condition list can be activated. The condition can be added on this screen. The redo button 510 is a button used when redoing an image search from the beginning. By pressing this button, all the search results so far can be erased, and it is possible to return to a state in which no conditions are specified.

  The return button 511 is a button used when it is desired to return the search by one step. By pressing this button, the condition designation state is moved back by one level (invalidating operations such as adding / deleting / changing conditions performed immediately before).

  The decision button 512 is a button used when the search is finished with the current search result and the next work is started. When the search is finished, the user can move to the next image generation / category correction request screen 701 by pressing the enter button 512. The determination button 512 is a device that functions effectively when pressed at any stage of image search.

  FIG. 6 is a diagram illustrating a configuration example of a condition designation screen 601 that can be displayed on the display unit of the Web client 101 so that a condition can be added when a search is performed. As shown in FIG. 6, the condition designation screen 601 includes a category selection component 602, a condition selection component 603, an allowable value selection component 604, and an OK button 605.

  First, the category selection component 602 can select a category from a selection menu display. From this category selection component 602, a category for specifying a condition can be selected. The options include options related to shooting information such as a search range, shooting date, and resolution, and options that specify category information such as terrain, status, and brightness. When a category is selected, an input form corresponding to it appears. FIG. 6 shows a screen when the terrain is selected, and the input form is an option format that can be selected from unclassified, sea, mountain, flat land,. Based on this form, a search condition can be added by selecting a corresponding terrain, inputting an allowable value, and pressing an OK button 605.

  Here, if the selected category is the shooting range (search range) of the image, a form for entering the latitude and longitude of the upper left and lower right of the rectangle for specifying the shooting range (assumed to be a rectangle), A form or the like for designating the center point and width of the initial range is displayed, and a search condition can be added by inputting a numerical value from the GUI and pressing an OK button 605 after confirmation. Returning to FIG. 5, the image itself can be confirmed.

  As described above, the input form suitable for the selected category is provided by changing to the input form suitable for the selected category. After adding as a search condition, each condition is set as one unit, and the search condition can be added or deleted by the same operation regardless of the contents as shown in the search condition list of FIG.

  FIG. 7 is a diagram illustrating a configuration example of an image generation / category correction request screen 701 displayed on the display unit of the Web client 101 when an image generation request and a category correction request are made based on an image search result. FIG. 6 is a diagram corresponding to FIG. 5. As shown in FIG. 7, the image generation / category correction request screen 701 includes a product image summary display area 702, a search result thumbnail 703, an image generation request selection component 704, a use image selection component 705, and a category correction request selection. A component 706, a search redo button 707, and a determination button 708 are configured.

  The product image summary display area 702 is an area for displaying an image in which the search result thumbnails 703 are arranged based on geographic information as a prediction of a product image that is generated when an image is generated in the current selection state. While looking at this area, the user (operator) can determine which image to use and whether to perform mosaic processing. The display at this stage is also displayed in an overlapping manner in the order of the scores when the search is performed (the one with the higher score is displayed in front).

  The image generation request selection component 704 can select whether or not to request image generation. As a result of looking at the search result, if it is determined that there is no need to generate an image, “Never Generate” is selected here. However, even when this is selected, a category correction request described later can be issued. That is, the process of editing an image using a thumbnail can be performed as described below. In general, such editing can be executed by an operation unit such as a mouse and a keyboard provided in the WEB client 101, editing processing software, and a CPU.

  The use image selection component 705 selects the use image and the presence / absence of a mosaic. Image generation requests can be issued for all search results, but only necessary ones can be selected and issued. In this case, on the product image summary display area 702, a search result thumbnail 703 of a necessary image is clicked to set the thumbnail 703 in a selected state. A plurality of thumbnails 703 can be selected. In addition, when all images or a plurality of images are selected, it is possible to select whether or not to perform mosaic processing. When selecting to perform mosaic processing, the product image is displayed with all selected images being mosaicked. Is output as When mosaic processing is not selected, each image is output as an individual product image.

  The category correction request selection component 706 can select whether to make a correction request for the category associated with the searched image. When the category correction request is selected, the category is corrected or newly registered on the table of each category and the combination table is corrected or newly registered as necessary, and then the association with the image is performed on the image classification table 401. It can be corrected. The search redo button 707 can be returned to the initial state of the image search screen 501 by pressing this button when the user wants to redo the search from the beginning. The decision button 708 can actually start the processing by pressing this button 708 when it is decided to execute the processing based on the result selected by each selection component.

  The operation of the satellite image processing system configured as described above will be described below. FIG. 8 is a flowchart showing an outline of the flow of image registration processing in the satellite image processing system. In the image registration process, first, when an image is captured and a raw image arrives at the Web server 103, the Web server 103 automatically registers shooting information (step 801). As described above, the shooting information includes general shooting conditions and information in the satellite image such as the shooting range, shooting date, shooting angle, and resolution. Next, automatic categorization is performed by the Web server 103, and data for displaying the result on the Web client 101 is sent (step 802). Automatic categorization classifies each category based on a threshold value based on a result using a calculation formula. Some categories, such as brightness, and the ratio of each category, should be registered with emphasis on the human sense rather than the quantitative calculation results. However, the accuracy is ignored for the time being. Automatic classification and registration.

  In the steps so far, the captured image is directly transferred to the storage unit (hard disk) under the management of the Web server 103, and at the same time, the following two-stage registration processing is performed. Since the main registration process is performed independently of the search process, it can be performed simultaneously. That is, based on the result of the automatic categorization, it is determined whether to manually correct the categorization (step 803). The automatic categorization only performs a classification process based on a mechanical algorithm, and may be deviated from human senses. Therefore, when the classification is clearly wrong at the time of registration, it is possible to manually correct the classification. If manual categorization is selected (Y), manual categorization is executed (step 804). In manual categorization, the web server 103 receives a result corrected to an appropriate category while observing the result of automatic categorization sent to the web client 101. The Web server 103 receives the value and registers it in the image classification table 401 (FIG. 4) in association with the image. If manual categorization is not selected (N), the first classification is accepted and registered in the image classification table 401 in association with the image.

  FIG. 9 is a flowchart showing an overview of the flow of image search processing in the Web server 103 of the satellite image processing system. In the image search process, it is first selected whether or not to add a search condition (step 901). When adding a search condition, the user presses an add button 509 on the image search screen 501 to call the condition designation screen 601 and perform an operation for adding the search condition.

  In the search condition adding operation, first, a category (including shooting information) to be added to the search condition is selected (step 902). In this step, a category to be added to the category currently specified as the search condition is selected. At this time, it is possible to select the same category as the category already specified. In this case, “and” or “or” can be selected. Such a condition designation makes it easy to search for an image including both “mountain” and “sea” with respect to the category landform.

  After the category to be added is selected, conditions regarding the category are designated (step 903). Since the method for specifying the condition varies depending on the selected category, the specification is made in accordance with the selected category. Since the method is the same as the explanation part of the condition designation screen 601, it can be understood by referring to that part.

  When the condition designation is completed, the addition process is completed by pressing an OK button 605 on the condition designation screen 601. After completion, the process returns to step 901. Looking at the current search results, no more search conditions are added, and when the search is to be terminated (N), the determination button 512 on the image search screen 501 is pressed (step 904). If not finished (Y), the processing from step 902 is repeated. Here, in the search condition addition step (902), the image search process can be redone from the beginning by using the redo button 510. Further, a return button 511 can be used to return to the previous state.

  FIG. 10 is a flowchart showing an outline of the flow of image generation / category correction processing of the satellite image processing system. In the image generation process, it is first checked whether image generation is requested (step 1001). If image generation is not requested (N), the image generation process is terminated and the process proceeds to category correction process.

  If image generation is requested (Y), it is checked whether a mosaic is requested (step 1002). If requested (Y), a product image is generated with a mosaic (step 1003). If not requested (N), a product image is generated without mosaic (step 1004). When there is no mosaic and a plurality of images are designated, a product image for each is generated. When the product image is completed, it is transferred to the product image storage (step 1005). When the transfer is completed, the Web client 101 that is the image request source is notified of this (step 1006).

  In the category correction process on the right side of FIG. 10, when image generation is not requested, it is performed immediately after the image generation request check. When image generation is requested, it is performed in parallel with the product image generation processing after the mosaic request check. First, a category correction request is checked (step 1007). If not requested (N), the process is terminated. If requested (Y), a new category is added or an existing category is corrected for each category table on the category database 110 (step 1008). Next, those combinations are designated, each category combination table is corrected, and then the image classification table 401 is updated to correct the association with the images. In this way, by receiving the image search result and updating the category associated with the image each time, it is possible to perform categorization as if viewed from the human side.

As described above, the satellite image processing system according to the present embodiment has the following advantages.
(1) By specifying imaging information of a requested image and a condition based on a category indicating the feature of the image, it is possible to search and create an image having the feature from a large number of satellite images.
(2) By making it possible to specify ambiguity as a search condition, images can be searched even if the requested image is not decided or the request itself is ambiguous, and it is necessary from the result. It is possible to search for the image that is being displayed.

  The present invention can be used as an image processing system for providing a satellite image.

It is a figure which shows the example of 1 structure of the image processing system by one embodiment of this invention. It is a figure which shows the example of 1 structure of the topography table stored in the category database. It is a figure which shows the example of 1 structure of the topography combination table stored in the category database. It is a figure which shows the example of 1 structure of the image classification table stored in the category database. It is a figure which shows the specific example of the setting screen which performs the image search by a category. It is a figure which shows the specific example of the screen for designation | designated which performs the conditions specification of a category. It is a figure which shows the specific example of the selection screen which selects the request | requirement of image generation and category correction. It is a flowchart figure which shows the outline | summary of the flow of an image registration process. It is a flowchart figure which shows the outline | summary of the flow of an image search process. It is a flowchart figure which shows the outline | summary of the flow of an image generation process and category correction process. It is a schematic block diagram of the conventional satellite image search tool.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 ... Web client, 102 ... Intranet, 103 ... Web server, 104 ... Search processing part, 105 ... Image generation part, 106 ... Category update part, 107 ... Communication part, 108 ... Map server, 109 ... Image server, 110 ... Category DB: 111 ... Product image storage storage, 201 ... Terrain table, 301 ... Terrain combination table, 401 ... Image classification table, 403 ... Category information, 404 ... Shooting information, 501 ... Image search screen, 601 ... Condition designation screen, 701 ... Image generation / category correction request screen

Claims (7)

An image processing apparatus for processing an image prepared in advance in a raw image server that stores a raw image taken from a high sky so as to meet a search condition desired by a Web client,
A database for storing data that is a clue when searching for an image in association with an image, the category determined as an index for associating the search conditions with each other, and the search conditions classified for each category A category database for storing
Criteria regarding how to extract categories extracted from the category database and search condition candidates included in the categories based on a request from the Web client, and to consider each of the extracted search condition candidates A search processing unit that performs a process of searching for an image in the image server that is equal to or less than the tolerance based on each tolerance for each of the search condition candidates set with tolerance indicating a value;
An image processing apparatus comprising: Further, for a plurality of items of the category or a plurality of items classified in the lower level included in the category, a setting interface capable of setting a plurality of percentages to consider the plurality of items,
The search processing unit
The image processing apparatus according to claim 1, wherein the search is performed using the plurality of setting items set by the interface as a search condition. The image processing apparatus according to claim 1, further comprising a setting interface that allows the tolerance to be set across a plurality of categories. The category database has a category combination table that displays combinations of categories,
The category combination table includes a category combination ID for uniquely identifying a combination of certain categories, a category, and a tolerance that can be set for each of the search condition candidates included in the category. The image processing apparatus according to any one of 1 to 3 . The image processing apparatus according to claim 4, wherein the category includes a terrain, and the category combination table includes a flat land, a mountain, and a sea as a terrain name that is a subordinate item of the terrain. . Creating a data for displaying a user interface that allows the web client to set the category, the search condition candidate, and the tolerance for the category, and transmitting the data;
The image according to any one of claims 1 to 5 , further comprising: a receiving unit that receives the tolerance set in the user interface on the Web client side and sends the received tolerance to the search processing unit. Processing equipment. The search processing unit
Among the characteristics of the image or related image information, the tolerance can be designated as a geographical shape. The first tolerance value set in advance in the system and the second tolerance value set from the client side. When the second tolerance value is smaller than the allowable value, it is determined that the search target is within the allowable range and this is used as the search condition. When the magnitude relationship is opposite, the search target is The image processing apparatus according to any one of claims 1 to 6 , wherein the image processing apparatus is configured not to become.

Images