JP2011008522A - Printer, apparatus and method for processing image, and computer program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently and appropriately perform image retrieval by setting up an appropriate threshold according to a query image.SOLUTION: An image processing apparatus is equipped with: a pivot image selection part which selects a pivot image; a first distance calculation part which calculates a target-pivot distance showing a degree of similarity of a target image and a pivot image; a second distance calculation part which calculates a query-pivot distance showing a degree of similarity of a query image and the pivot image; a threshold setting part (S220) which sets up a threshold based on the query-pivot distance; an approximation distance calculation part (S230) which calculates a query-target approximation distance as a lower bound value of the query-target distance based on the target-pivot distance; a retrieval target setting part (S240) which excludes a target image with a query-target approximation distance larger than the threshold; a third distance calculation part which calculates the query-target distance about the target image which is not excluded; and an image retrieval part which detects the target image similar to the query image based on the query-target distance.

Description

  The present invention relates to image processing for performing an image search for detecting an image similar to a query image from a plurality of target images.

  Image processing for performing image search for detecting an image similar to a query image as a reference for image search for a predetermined feature from a plurality of target images as image search targets is known. For the image search, for example, a feature amount distance representing the degree of similarity between the two images with respect to the predetermined feature (the smaller the feature amount distance, the more similar the two images are) is used.

  As an image search algorithm, an algorithm called Linear Approximating and Eliminating Search Algorithm (hereinafter also referred to as “LAESA” for short) is known (for example, see Non-Patent Document 1). LAESA is a pivot-based search algorithm. That is, in LAESA, feature amounts between each target image and a plurality of images selected from the target images (referred to as pivot images) are pre-processed to reduce the feature amount distance calculation amount during the image search process. A distance (target-pivot distance) is calculated. In the image search, a query-target approximate distance as a lower bound value of a feature amount distance (query-target distance) between the query image and the target image is calculated based on a target-pivot distance calculated in advance. . Since the calculation of the query-target approximate distance is performed using a triangle inequality, the calculation load is very small. In LAESA, target images whose calculated query-target approximate distance is larger than a predetermined threshold (hereinafter also referred to as “query range”) are excluded from image search targets without calculating the actual query-target distance. The For target images that remain without being excluded by the exclusion determination using the query-target approximate distance, the actual query-target distance is calculated, and a target image having a small query-target distance is detected as an image similar to the query image. The

Maria Luisa Mico, Jose Oncina, “A new version of the Nearest-Neighbour Approximating and Eliminating Search Algorithm (AESA) with linear preprocessing time and memory requirements” (Pattern Recognition Letters vol.15, p.9-p.7 1994 January)

  Conventionally, in a search algorithm such as LAESA, a threshold value (query range) for exclusion determination is set in advance and used fixedly through image search. For this reason, image search using a search algorithm such as LAESA may not be executed effectively. For example, if the set threshold value is too large, the number of target images that are actual query-target distance calculation targets may increase, and the time required for image search may increase. On the other hand, if the set threshold value is too small, a target image having a high degree of similarity with the query image may actually be excluded, and the final image search result may be inappropriate.

  Such a problem is not limited to LAESA, but a query-target approximate distance as a lower bound value of the query-target distance is calculated based on the target-pivot distance, and the target image in which the query-target approximate distance is larger than a predetermined threshold value. This is a problem common to a search algorithm that excludes from the calculation target of the actual query-target distance.

  The present invention has been made to solve the above-described problems, and calculates a query-target approximate distance based on a target-pivot distance, and selects a target image whose query-target approximate distance is larger than a predetermined threshold as an actual query. An object of the present invention is to provide a technique that enables an image search to be executed efficiently and appropriately using a search algorithm that excludes the target distance from the calculation target.

  In order to solve at least a part of the above problems, the present invention can be realized as the following forms or application examples.

Application Example 1 An image processing apparatus that performs an image search,
A pivot image selection unit for selecting a part of the plurality of target images as a pivot image;
A first distance calculation unit that calculates a target-pivot distance representing a similarity degree for a predetermined feature between each target image and each pivot image;
A second distance calculating unit that calculates a query-pivot distance that represents a degree of similarity for the predetermined feature between a query image and each pivot image;
A threshold setting unit for setting a threshold based on the query-pivot distance;
Approximate distance calculation that calculates a query-target approximate distance as a lower bound value of the query-target distance that represents the similarity of the predetermined feature between the query image and each target image based on the target-pivot distance And
A search target setting unit that excludes the target image whose query-target approximate distance is larger than the threshold from the target of the image search;
A third distance calculating unit that calculates the query-target distance for the target image that is not excluded from the image search target;
An image search unit that detects the target image similar to the query image with respect to the predetermined feature based on the query-target distance.

  In this image processing apparatus, since the threshold value for target image exclusion determination based on the query-target approximate distance is set based on the query-pivot distance, an appropriate threshold value can be set according to the query image, and image search can be performed. It can be executed efficiently and appropriately.

[Application Example 2] The image processing apparatus according to Application Example 1,
The threshold value setting unit calculates a predetermined statistical value of the query-pivot distance, and sets the statistical value as the threshold value.

  In this image processing apparatus, the predetermined statistic value of the query-pivot distance is considered to correspond to the estimated value of the predetermined statistic value for the query-target distance. A threshold value can be set so that an appropriate number of target images are excluded and an appropriate number of target images are included in the query-target distance calculation target, so that an image search can be performed efficiently and appropriately. .

[Application Example 3] The image processing apparatus according to Application Example 2,
The image processing apparatus, wherein the predetermined statistical value is an average value.

Application Example 4 The image processing apparatus according to Application Example 2,
The image processing apparatus, wherein the predetermined statistical value is a q quantile (0 <q <1).

Application Example 5 The image processing apparatus according to any one of Application Examples 2 to 4, further comprising:
A transmission unit for transmitting the query-pivot distance to the other image processing device;
A receiving unit that receives the predetermined statistic about the query-pivot distance transmitted from the plurality of image processing devices calculated in the other image processing device;
The threshold value setting unit is an image processing device that sets the received predetermined statistical value as the threshold value.

  In this image processing apparatus, the same threshold can be set even when image search is executed in parallel for a plurality of independent target image groups by a plurality of image processing apparatuses. The consistency of image search can be ensured.

Application Example 6 The image processing apparatus according to any one of Application Examples 2 to 4, further comprising:
A transmission unit for transmitting the predetermined statistical value to the other image processing device;
A receiving unit that receives a predetermined secondary statistical value for the predetermined statistical value transmitted from the plurality of image processing devices calculated in the other image processing device;
The threshold value setting unit is an image processing device that sets the received predetermined secondary statistical value as the threshold value.

  In this image processing apparatus, the same threshold can be set even when image search is executed in parallel for a plurality of independent target image groups by a plurality of image processing apparatuses. As a result, the consistency of the image search performed can be ensured, and the communication load between the image processing apparatuses can be reduced.

Application Example 7 The image processing apparatus according to any one of Application Examples 1 to 6, further comprising:
A print image setting unit for setting a detection image to be printed from detection images detected by image search;
An image processing apparatus comprising: a printing unit that prints a detection image to be printed.

  It should be noted that the present invention can be realized in various modes. For example, an image processing method and apparatus, an image search method and apparatus, a printing method and apparatus, and a computer for realizing the functions of these methods or apparatuses The present invention can be realized in the form of a program, a recording medium recording the computer program, a data signal including the computer program and embodied in a carrier wave, and the like.

1 is an explanatory diagram schematically illustrating a configuration of a printer 100 as an image processing apparatus according to an embodiment of the present invention. 6 is a flowchart illustrating a flow of image search / print processing according to the embodiment. It is explanatory drawing which shows an example of the initial window W1. It is a flowchart which shows the flow of an image search execution process. It is explanatory drawing which shows an example of search result window W2.

Next, embodiments of the present invention will be described in the following order based on examples.
A. Example:
A-1. Configuration of image processing device:
A-2. Image search / print processing:
B. Variation:

A. Example:
A-1. Configuration of image processing device:
FIG. 1 is an explanatory diagram schematically showing the configuration of a printer 100 as an image processing apparatus according to an embodiment of the present invention. The printer 100 of this embodiment is an ink jet color printer that supports so-called direct printing, in which an image is printed based on image data acquired from a memory card MC or the like. The printer 100 includes a CPU 110 that controls each unit of the printer 100, an internal memory 120 configured by a ROM and a RAM, an operation unit 140 configured by buttons and a touch panel, a display unit 150 configured by a liquid crystal display, and a printer. An engine 160 and a card interface (card I / F) 170 are provided. The printer 100 may further include an interface for communicating with another device (for example, a digital still camera or a personal computer). Each component of the printer 100 is connected to each other via a bus.

  The printer engine 160 is a printing mechanism that performs printing based on print data. The card interface 170 is an interface for exchanging data with the memory card MC inserted into the card slot 172. In this embodiment, an image file including image data is stored in the memory card MC.

  The internal memory 120 stores an image processing unit 200, a display processing unit 310, and a print processing unit 320. The image processing unit 200 is a computer program for executing image search / print processing under a predetermined operating system. The image search / print processing of this embodiment is processing for detecting an image similar to the query image for a predetermined feature from a plurality of target images and printing the detected image. The image search / print process will be described in detail later.

  The image processing unit 200 includes an image search unit 210 and a print image setting unit 220 as modules. The image search unit 210 includes a pivot image selection unit 211, a distance calculation unit 212, a threshold setting unit 213, an approximate distance calculation unit 214, and a search target setting unit 215. The functions of these units will be described in detail in the description of image search / print processing described later. The distance calculation unit 212 corresponds to the first, second, and third distance calculation units in the present invention.

  The display processing unit 310 is a display driver that controls the display unit 150 to display a window screen including a processing menu and a message, an image, and the like on the display unit 150. The print processing unit 320 is a computer program for generating print data from image data, controlling the printer engine 160, and printing an image based on the print data. The CPU 110 reads out and executes these programs (the image processing unit 200, the display processing unit 310, and the print processing unit 320) from the internal memory 120, thereby realizing the functions of these units.

A-2. Image search / print processing:
FIG. 2 is a flowchart illustrating a flow of image search / print processing in the embodiment. The image search / print processing of this embodiment detects and detects an image similar to a query image as a reference in image search for a predetermined feature of the image from a plurality of target images to be subjected to image search. This is a process for printing a printed image. More specifically, a feature amount distance representing the similarity between the query image for the predetermined feature and the target image is calculated, and a target image whose feature amount distance is equal to or less than the maximum feature amount distance is detected. Note that the smaller the feature distance, the more similar the two images. In the present embodiment, as the predetermined feature, a wavelet coefficient calculated by wavelet decomposition of an image using a Haar wavelet as a base is used. Although the target image can be set arbitrarily, in this embodiment, an image (image data included in the image file) stored in a predetermined area of the memory card MC is set as the target image.

  In the image search of this embodiment, as described below, a LAESA algorithm that reduces the number of times of calculation of the feature amount distance (query-target distance) between the query image and the target image to speed up the processing is used. Adopted. For image retrieval using LAESA, “A new version of the Nearest-Neighbour Approximating and Eliminating Search Algorithm (AESA) with linear preprocessing time and memory requirements” by Maria Luisa Mico, Jose Oncina (Pattern Recognition Letters vol.15, p 9-p.7 January 1994) and Edgar Chavez, JL Marroquin, Ricardo Baeza-Yates, “Spaghettis: An Array Based Algorithm for Similarity Queries in Metric Spaces” (spire, pp. 38, String Processing and Information Retrieval Symposium & International Workshop on Groupware, 1999), detailed description of LAESA will be omitted in this specification.

  In response to an instruction from the user via the operation unit 140 (FIG. 1), the image processing unit 200 starts an image search / print process (FIG. 2). When the image search / printing process is started, the image processing unit 200 controls the display processing unit 310 to display the initial window W1 on the display unit 150 (step S110). FIG. 3 is an explanatory diagram showing an example of the initial window W1. In this embodiment, a query image as a reference for image search is set by drawing. Accordingly, the initial window W1 includes various user interfaces for drawing. That is, the initial window W1 includes a color palette button Bu11 including a plurality of color buttons having different colors, a brush diameter specifying button Bu12 including a plurality of brush buttons having different diameters, and a color palette button Bu11 and a brush diameter specifying button Bu12. The drawing image display area Ar10 that displays the drawing image DI according to the user's drawing operation used, the clear button Bu13 that clears the display of the drawing image display area Ar10, and the drawing image DI that is displayed in the drawing image display area Ar10. A return button Bu14 for changing to a drawing image DI before one drawing operation, a button Bu15 for changing the drawing image DI displayed in the drawing image display area Ar10 to a drawing image DI after one drawing operation, and a drawing image display area Ar10 And an OK button 16 for setting the image displayed on the screen as a query image.

  In step S120 (FIG. 2), pre-processing is executed. In the pre-processing, first, the pivot image selection unit 211 (FIG. 1) randomly selects some images from a plurality of target images, and sets the selected images as pivot images. Next, the distance calculation unit 212 calculates a feature amount distance (target-pivot distance) that represents the degree of similarity between each target image and each pivot image for a predetermined feature. The calculated target-pivot distance is stored in a predetermined area of the internal memory 120. The pre-processing can be executed using only the target image (including the pivot image) (can be executed off-line), and does not require a query image, so before the query image setting processing (step S130 described later). Can be executed. Accordingly, the pre-processing may be executed before the start of the image search / print processing.

  In step S130 (FIG. 2), the image search unit 210 (FIG. 1) sets a query image. The user can draw a desired image in the drawing image display area Ar10 by selecting and operating each interface included in the initial window W1 (FIG. 3) via the operation unit 140. After the image is drawn in the drawing image display area Ar10, when the user selects the OK button Bu16, the image search unit 210 sets the drawing image displayed in the drawing image display area Ar10 as a query image.

  When the query image setting (step S130 in FIG. 2) is completed, an image search is executed (step S140). FIG. 4 is a flowchart showing the flow of the image search execution process. In step S210 of the image search execution process, the distance calculation unit 212 calculates a feature amount distance (query-pivot distance) that represents the degree of similarity between the query image for each predetermined feature and each pivot image.

  In step S220 (FIG. 4), the threshold setting unit 213 (FIG. 1) sets a threshold (hereinafter also referred to as “query range R”) based on the calculated query-pivot distance. The query range R is a threshold value used for target image exclusion determination based on a query-target approximate distance described later. The query range R is set to a predetermined statistical value for the calculated query-pivot distance. In this embodiment, an average value is adopted as the predetermined statistical value. Therefore, the query range R is an average value of the query-pivot distance.

  When the query range R is set in this way, the query range R directly depends on the query image. That is, the query range R has a value corresponding to the query image. Further, since the pivot image is an image selected at random from the target images, the calculated statistical value is an estimated value of the statistical value for the entire target image. That is, it can be said that the calculated statistical value for the query-pivot distance is an estimated value of the statistical value for the query-target distance.

  In step S230 (FIG. 4), the approximate distance calculation unit 214 (FIG. 1) calculates the feature amount between the query image and the target image based on the target-pivot distance calculated in the preprocessing (step S120 in FIG. 2). A query-target approximate distance as a lower bound value of the distance (query-target distance) is calculated. Since the calculation of the query-target approximate distance is performed using a triangle inequality, the calculation load is very small.

  In step S240 (FIG. 4), the search target setting unit 215 (FIG. 1) performs target image exclusion determination based on the calculated query-target approximate distance. Specifically, the search target setting unit 215 excludes a target image whose query-target approximate distance is larger than the query range R from a target of a query-target distance calculation process (step S250) described later. Since the query-target approximate distance is a lower bound value of the query-target distance, when the query-target approximate distance is larger than the query range R, the actual query-target distance is also larger than the query range R. Therefore, a target image having a query-target approximate distance larger than the query range R is excluded from the object of image search (step S260) without calculating the actual query-target distance.

  In step S250 (FIG. 4), the distance calculation unit 212 (FIG. 1) calculates an actual query-target distance for the target images that remain without being excluded in the exclusion determination (step S240). In step S260, the image search unit 210 performs an image search for detecting a target image whose calculated query-target distance is equal to or less than a preset maximum feature amount distance. Note that when the maximum number of detected images is set in advance and the number of detected images exceeds the maximum number of detected images, the exceeding number of images is excluded in order from the larger query-target distance. . If the maximum feature amount distance is not set and only the maximum number of detected images is set, images with the maximum number of detected images are detected in order from the smallest query-target distance.

  When the image search execution process (FIG. 4) is completed, the image processing unit 200 (FIG. 1) controls the display processing unit 310 to display the search result window W2 on the display unit 150 (step S150). FIG. 5 is an explanatory diagram showing an example of the search result window W2. The search result window W2 is a window for displaying a list of target images (detected images DeI) detected in the image search. That is, as shown in FIG. 5, the search result window W2 includes a detection image display area Ar62 in addition to the query image display area Ar61. In the detection image display area Ar62, the detection images DeI are displayed side by side in a rank order (in ascending order of feature amount distance) indicating the similarity to the query image.

  The search result window W2 (FIG. 5) includes a print designation area Ar63. In the search result window W2, for example, the user drags and drops the icon of the detection image DeI to the print designation area Ar63 to display the icon IC of the detection image DeI in the print designation area Ar63, thereby detecting the detection image DeI to be printed. Can be designated (step S160 in FIG. 2). In the example of FIG. 5, two detection images DeI are designated as images to be printed. The print designation area Ar63 includes a print execution button Bu64 and a print designation release button Bu65. The user selects the icon IC of the image displayed in the print designation area Ar63 and selects the print designation release button Bu65 (or drags and drops the image icon IC onto the print designation release button Bu65). The print designation of the image can be canceled. In addition, the user can instruct the print execution start of the detected image DeI corresponding to the icon IC displayed in the print designation area Ar63 by selecting the print execution button Bu64. When an instruction to start printing is given, the print processing unit 320 (FIG. 1) generates print data based on the image data of the designated detection image DeI, and controls the printer engine 160 to print the detection image DeI. This is executed (step S170 in FIG. 2).

  As described above, in the image search / print processing by the printer 100 according to the present embodiment, selection of a pivot image from a plurality of target images and calculation of a target-pivot distance are performed as pre-processing. Further, in the image search execution process, the query range R is set based on the query-pivot distance, and target image exclusion determination based on the query-target approximate distance is performed using the set query range R, and the target that is not excluded An actual query-target distance is calculated for only the image, and an image similar to the query image is detected based on the calculated actual query-target distance. In this embodiment, since the query range R is set to a predetermined statistical value (average value) for the query-pivot distance, the query range R directly depends on the query image, and the statistical value ( The query range R as an average value is an estimated value of the statistical value for the entire target image. Therefore, in this embodiment, the query range R is excluded from the appropriate number of target images in the entire target image regardless of the contents of the query image, and the appropriate number of target images is set to the actual query-target distance. It can be set to be a calculation target. Therefore, in this embodiment, the query-target approximate distance is calculated based on the target-pivot distance, and a search algorithm for excluding target images whose query-target approximate distance is larger than a predetermined threshold from the actual query-target distance calculation target. Can be used to efficiently and appropriately perform image retrieval.

B. Variation:
The present invention is not limited to the above-described examples and embodiments, and can be implemented in various modes without departing from the gist thereof. For example, the following modifications are possible.

B1. Modification 1:
In the above embodiment, the image search is executed using the image stored in the memory card MC as the target image. However, the target image in the image search can be arbitrarily set. For example, an image stored in a predetermined area of the internal memory 120 (FIG. 1) may be set as a target image, or in a storage area of another device connected to the printer 100 via a cable or a network. The stored image may be set as the target image. Regardless of the storage location of the target image, the pre-processing (step S120 in FIG. 2) can be performed using only the target image (can be performed offline).

  Further, the image search may be executed in parallel on a plurality of mutually independent target image groups by one or a plurality of image processing apparatuses. In this case, it is preferable that the query range R is set to the same value in order to ensure the consistency of the image search executed in parallel. Therefore, when the query-pivot distance is calculated in each processing unit that performs image search, the calculated query-pivot distance is aggregated into one processing unit (master) via, for example, the network, and all queries are performed in the master. -A statistical value (for example, average value) about the pivot distance is calculated as the query range R, and the calculated query range R is transmitted to each processing unit. Each processing unit executes an image search using the transmitted query range R. In this way, the query range R as a statistical value (average value) is an estimated value of the statistical value for all the target image groups as a whole.

  Further, in each processing unit that performs image search, calculation of the query-pivot distance and calculation of a statistical value (for example, average value) for the calculated query-pivot distance are performed, and the statistical value calculated in each processing unit. The values are aggregated to one processing unit (master) via a network, for example, and secondary statistical values (for example, average values) for all statistical values are calculated as a query range R in the master, and the calculated query range R is It may be transmitted to each processing unit, and each processing unit may execute an image search using the transmitted query range R. In this way, the data transmitted from each processing unit to the master during aggregation is only one statistical value, and the communication load is reduced.

B2. Modification 2:
In the above embodiment, the predetermined statistical value for the query-pivot distance used for setting the query range R is an average value, but a statistical value other than the average value may be used. For example, a q quantile (a point that divides the distribution into q: 1-q (0 <q <1)) may be used as a statistical value. Further, when the q quantile is used as the statistical value, the value of q may be set according to the allowable processing time. For example, when the allowable processing time is short, the value of q may be set to a smaller value than when it is long.

B3. Modification 3:
In the above embodiment, the predetermined feature of the image that is the target of the similarity between the images is the wavelet coefficient calculated by the wavelet decomposition of the image using the Haar wavelet as a base, but the predetermined feature of the image is Other features such as a color histogram representing the color distribution of an image expressed in a predetermined color space may be used.

B4. Modification 4:
In the above embodiment, the query image as a reference for image search is set by drawing. However, the query image may be set by other methods such as file designation and color selection.

B5. Modification 5:

  The contents and layout of the initial window W1 and the search result window W2 in the above embodiment are merely examples, and various modifications can be made. Further, the configuration of the printer 100 as the image processing apparatus in the above embodiment is merely an example, and the configuration of the printer 100 can be variously changed. In the above-described embodiments, the image search / print processing by the printer 100 as the image processing apparatus has been described. However, part or all of the processing may be another type of server, personal computer, digital still camera, digital video camera, or the like. It may be executed by the image processing apparatus. The printer 100 is not limited to an ink jet printer, and may be another type of printer such as a laser printer or a sublimation printer.

  In the above embodiment, a part of the configuration realized by hardware may be replaced with software, and conversely, a part of the configuration realized by software may be replaced by hardware.

  In addition, when part or all of the functions of the present invention are realized by software, the software (computer program) can be provided in a form stored in a computer-readable recording medium. In the present invention, the “computer-readable recording medium” is not limited to a portable recording medium such as a flexible disk or a CD-ROM, but an internal storage device in a computer such as various RAMs and ROMs, a hard disk, etc. It also includes an external storage device fixed to the computer.

100 ... Printer 110 ... CPU
DESCRIPTION OF SYMBOLS 120 ... Internal memory 140 ... Operation part 150 ... Display part 160 ... Printer engine 170 ... Card interface 172 ... Card slot 200 ... Image processing part 210 ... Image search part 211 ... Pivot image selection part 212 ... Distance calculation part 213 ... Threshold setting part 214 ... Approximate distance calculation unit 215 ... Search target setting unit 220 ... Print image setting unit 310 ... Display processing unit 320 ... Print processing unit

Claims (10)

An image processing apparatus for performing image search,
A pivot image selection unit for selecting a part of the plurality of target images as a pivot image;
A first distance calculation unit that calculates a target-pivot distance representing a similarity degree for a predetermined feature between each target image and each pivot image;
A second distance calculating unit that calculates a query-pivot distance that represents a degree of similarity for the predetermined feature between a query image and each pivot image;
A threshold setting unit for setting a threshold based on the query-pivot distance;
Approximate distance calculation that calculates a query-target approximate distance as a lower bound value of the query-target distance that represents the similarity of the predetermined feature between the query image and each target image based on the target-pivot distance And
A search target setting unit that excludes the target image whose query-target approximate distance is larger than the threshold from the target of the image search;
A third distance calculating unit that calculates the query-target distance for the target image that is not excluded from the image search target;
An image search unit that detects the target image similar to the query image with respect to the predetermined feature based on the query-target distance. The image processing apparatus according to claim 1,
The threshold value setting unit calculates a predetermined statistical value of the query-pivot distance, and sets the statistical value as the threshold value. The image processing apparatus according to claim 2,
The image processing apparatus, wherein the predetermined statistical value is an average value. The image processing apparatus according to claim 2,
The image processing apparatus, wherein the predetermined statistical value is a q quantile (0 <q <1). The image processing apparatus according to any one of claims 2 to 4, further comprising:
A transmission unit for transmitting the query-pivot distance to the other image processing device;
A receiving unit that receives the predetermined statistic about the query-pivot distance transmitted from the plurality of image processing devices calculated in the other image processing device;
The threshold value setting unit is an image processing device that sets the received predetermined statistical value as the threshold value. The image processing apparatus according to any one of claims 2 to 4, further comprising:
A transmission unit for transmitting the predetermined statistical value to the other image processing device;
A receiving unit that receives a predetermined secondary statistical value for the predetermined statistical value transmitted from the plurality of image processing devices calculated in the other image processing device;
The threshold value setting unit is an image processing device that sets the received predetermined secondary statistical value as the threshold value. The image processing apparatus according to any one of claims 1 to 6, further comprising:
A print image setting unit for setting a detection image to be printed from detection images detected by image search;
An image processing apparatus comprising: a printing unit that prints a detection image to be printed. An image processing method for performing image search,
(A) selecting a part of the plurality of target images as a pivot image;
(B) calculating a target-pivot distance representing a similarity degree for a predetermined feature between each of the target images and each of the pivot images;
(C) calculating a query-pivot distance representing a similarity degree for the predetermined feature between a query image and each of the pivot images;
(D) setting a threshold based on the query-pivot distance;
(E) Based on the target-pivot distance, a query-target approximate distance is calculated as a lower bound value of the query-target distance representing the degree of similarity of the predetermined feature between the query image and each target image. Process,
(F) excluding the target image whose query-target approximation distance is greater than the threshold from the image search target;
(G) calculating the query-target distance for the target images not excluded from the image search target;
(H) detecting the target image similar to the query image with respect to the predetermined feature based on the query-target distance. A computer program for image processing that performs image search,
Pivot image selection function to select a part of multiple target images as a pivot image,
A first distance calculation function for calculating a target-pivot distance representing a degree of similarity for a predetermined feature between each target image and each pivot image;
A second distance calculation function for calculating a query-pivot distance that represents a similarity degree of the predetermined feature between a query image and each of the pivot images;
A threshold setting function for setting a threshold based on the query-pivot distance;
Approximate distance calculation that calculates a query-target approximate distance as a lower bound value of the query-target distance that represents the similarity of the predetermined feature between the query image and each target image based on the target-pivot distance Function and
A search target setting function for excluding the target image whose query-target approximation distance is larger than the threshold from the target of the image search;
A third distance calculating function for calculating the query-target distance for the target image that is not excluded from the image search target;
A computer program for causing a computer to realize an image search function for detecting the target image similar to the query image with respect to the predetermined feature based on the query-target distance. A printing device,
A pivot image selection unit for selecting a part of the plurality of target images as a pivot image;
A first distance calculation unit that calculates a target-pivot distance representing a similarity degree for a predetermined feature between each target image and each pivot image;
A second distance calculating unit that calculates a query-pivot distance that represents a degree of similarity for the predetermined feature between a query image and each pivot image;
A threshold setting unit for setting a threshold based on the query-pivot distance;
Approximate distance calculation that calculates a query-target approximate distance as a lower bound value of the query-target distance that represents the similarity of the predetermined feature between the query image and each target image based on the target-pivot distance And
A search target setting unit that excludes the target image whose query-target approximate distance is larger than the threshold from the target of the image search;
A third distance calculating unit that calculates the query-target distance for the target image that is not excluded from the image search target;
An image search unit for detecting the target image similar to the query image with respect to the predetermined feature based on the query-target distance;
A print image setting unit for setting a detection image to be printed from detection images detected by image search;
A printing unit that prints a detection image to be printed.

Images