JP2006139608A - Image search apparatus and image search method - Google Patents

Abstract

An image search apparatus and method capable of performing high-speed search for a large number of data images, particularly compressed images.
A reference image storage device for storing a reference image, a data image storage device for storing a data image, and a maximum luminance value of all pixels included in the reference image stored in the reference image storage device. A reference range determining unit 3 that determines a section (reference range) between the value C _max and the minimum value C _min , and a data image stored in the data image storage device 5, a part of a pixel set included in the data image Sample data pixel set extraction means 7 for extracting (sample data pixel set), and candidate image determination using the data image as a candidate image when the relationship between the sample data pixel set and the reference range satisfies a predetermined requirement Means 4 and image selection means 8 for performing detailed comparison between the candidate image and the reference image and selecting an image that is the same as or similar to the reference image.
[Selection] Figure 1

Description

  The present invention relates to an image search apparatus and an image search method for searching for an image that is the same as or similar to a reference image from a large number of data images.

  An image search for searching for the same image data as the reference image from a large number of image data is required for various applications. The same applies to compressed images (for example, Patent Document 1 and Non-Patent Document 1). Conventional image retrieval of compressed images is performed by pattern matching in which the compressed image is restored (decoded) to the original image data, and the decoded image data and the reference image data are compared in pixel units. Decoding processing and pixel-unit pattern matching require a long time. When searching for compressed images, especially when searching from a large number of images, the search time is enormous. For this reason, it is a problem to speed up the image search of the compressed image.

  In order to solve this problem, various image search techniques have been proposed in which a rough search is performed on all image data to extract candidate image data, a detailed search is performed on the extracted candidate image data, and the result is output.

  For example, Patent Document 2 discloses that variable length decoding processing and inverse quantization processing are performed on image data that has been subjected to discrete cosine transform (DCT) and stored in a buffer, and only a direct current component is obtained from the processed image data. Separating, performing a search based on the DC component, performing a first matching process for extracting image data satisfying the search condition, restoring the extracted image data to image data based on a bitmap, and restoring the restored image data An image search method is disclosed in which a second matching process for pattern matching is performed on image data, and a matched data image is output as a search result.

In Patent Document 3, a candidate area where a search image may exist is first obtained by a rough search for determining similarity based on a color histogram having a low gradation resolution, and then the gradation resolution is determined in the candidate area. An image detection device that performs a detailed search for determining a similarity based on a high color histogram and detects a region where a search image exists is disclosed.
Japanese Patent Application Laid-Open No. 2004-312693 Japanese Patent Laid-Open No. 9-44519 JP 2002-150287 A Seiichiro Kamada, “Consideration on Information Compression of Gray Image Using Hilbert Scanning”, IEICE Transactions, Vol.J80-D-II, No.2, pp.426-433, February 1997

  The inventions described in Patent Document 2 and Patent Document 3 avoid the time loss caused by performing a detailed search on all image data by performing a two-stage search of a coarse search and a detailed search, and image search We are trying to speed up. However, there is a limit to the speeding up of the image search according to these inventions for the following reasons. That is, in the invention described in Patent Document 2, variable length decoding processing, inverse quantization processing, and DC component separation processing are performed on all image data before the first matching processing (coarse search). . Therefore, the time for these processes increases the processing time for the entire image search. In the invention described in Patent Document 3, color histogram creation and similarity determination based on the color histogram are performed in the coarse search stage. Therefore, it takes a long time for the rough search.

  The present invention has been made in view of the above problems, and an object thereof is to provide an image search apparatus and method capable of performing high-speed search for a large number of data images, particularly compressed images.

According to a first configuration of the image search apparatus of the present invention, in an image search apparatus that searches for an image that is the same as or similar to a reference image from a plurality of data images, a reference image storage unit that stores the reference image; Data image storage means for storing a data image, and a section [C _max , between the maximum value C _max and the minimum value C _min of the luminance values of all pixels included in the reference image stored in the reference image storage means C _min ] (hereinafter referred to as “reference range”) and a part of a set of pixels included in the data image (hereinafter referred to as “reference range”) and the data image stored in the data image storage unit. The sample data pixel set extracting means for extracting the sample data pixel set)) and the luminance of the pixels belonging to the sample data pixel set are related to the reference range. When the above requirement is satisfied, a candidate image determination unit that determines the data image as a candidate image and a detailed comparison between the candidate image and the reference image are performed to select an image that is the same as or similar to the reference image. And an image selection means.

According to this configuration, the luminance values of pixels belonging to a part of the pixel set included in the data image (sample data pixel set) are the maximum value C _max and the minimum value C _min of the luminance values of all the pixels included in the reference image. When a predetermined requirement is satisfied in relation to the section [C _max , C _min ] (“reference range”), a data image is determined as a candidate image, and a detailed comparison between the candidate image and the reference image is performed. Thus, the same or similar image as the reference image is selected. As a result, a small number of candidate images are selected from a large number of data images in a short time, and detailed comparison is performed between the small number of candidate images and the reference image, thereby speeding up the image search.

  According to a second configuration of the image search apparatus of the present invention, in the first configuration, partial data image extraction means for extracting a part of the data image stored in the data image storage means as a partial data image is provided. The sample data pixel set extraction unit extracts a part of the pixel set included in the partial data image as a sample data pixel set for the partial data image, and the candidate image determination unit includes the sample data pixel set. The partial data image is determined as a candidate image when the luminance of the pixel belonging to the pixel satisfies a predetermined requirement in relation to the reference range.

  According to this configuration, the partial data image extraction unit extracts a part of the data image as a partial data image. As a result, a data image including the same or similar portion as the reference image can be searched.

  According to a third configuration of the image search apparatus of the present invention, in the first or second configuration, the candidate image determining means has a ratio of the luminance of the pixels belonging to the sample data pixel set belonging to the reference range being predetermined. When it is equal to or greater than the threshold, the data image or the partial data image is determined as a candidate image.

  According to a fourth configuration of the image search apparatus of the present invention, in the first or second configuration, the candidate image determination unit is configured such that when the luminances of all the pixels belonging to the sample data pixel set belong to the reference range, The data image or the partial data image is determined as a candidate image.

  According to a fifth configuration of the image search apparatus of the present invention, in the first or second configuration, the candidate image determination unit has a ratio of luminance of pixels belonging to the sample data pixel set belonging to the reference range of 80% or more. In this case, the data image or the partial data image is determined as a candidate image.

  According to the third to fifth configurations, the candidate image determination unit is configured to determine whether the candidate image determination unit has the data image or the partial data when the luminance of the pixels belonging to the sample data pixel set is greater than or equal to the predetermined threshold. An image is determined as a candidate image. Thereby, since the candidate image is determined based on the ratio of the pixels belonging to the reference range, the calculation of the candidate image determination is further simplified, and the time required for determining the candidate image can be further shortened. Further, by adjusting the threshold value, it is possible to adjust the roughness of the mesh for determining candidate images according to the object and purpose of image search.

According to a sixth configuration of the image search device of the present invention, in the first or second configuration, the candidate image determination unit is configured such that an average value C _mean of luminances of pixels belonging to the sample data pixel set is within the reference range. In the case of belonging, the data image or the partial data image is determined as a candidate image.

According to this configuration, the candidate image determining unit determines the data image or the partial data image as a candidate image when the average value C _mean of the luminance of the pixels belonging to the sample data pixel set belongs to the reference range. Thereby, the conditions for determining candidate images can be relaxed. Therefore, the possibility that a data image that is the same as or similar to the reference image is erroneously excluded from the candidate image is reduced, and the accuracy of the image search is improved.

According to a seventh configuration of the image search apparatus of the present invention, in the first or second configuration, the candidate image determination unit uses an average value C _mean and a standard deviation σ of luminances of pixels belonging to the sample data pixel set. On the other hand, when C _mean + σ and C _mean −σ belong to the reference range, the data image or the partial data image is determined as a candidate image.

According to this configuration, the candidate image determination unit is configured such that both C _mean + σ and C _mean −σ belong to the reference range with respect to the average value C _mean and standard deviation σ of the luminance values of the pixels belonging to the sample data pixel set. Then, the data image or the partial data image is determined as a candidate image. As a result, the conditions for determining the candidate image can be relaxed, so that the possibility that a data image that is the same as or similar to the reference image is erroneously excluded from the candidate image is reduced, and the accuracy of the image search is improved.

  According to an eighth configuration of the image search apparatus of the present invention, in any one of the first to seventh configurations, the image selection unit includes a difference between a pixel value histogram of the candidate image and a pixel value histogram of the reference image. When the difference is smaller than a predetermined threshold, the candidate image is selected as an image that is the same as or similar to the reference image.

  According to this configuration, the image selection unit calculates a difference between the pixel value histogram of the candidate image and the pixel value histogram of the reference image, and when the difference is smaller than the predetermined threshold, the candidate image is determined as the reference image. Select the same or similar images. Thereby, since image selection can be performed with relatively simple calculation, image search can be performed at higher speed and with higher accuracy.

  According to a ninth configuration of the image search apparatus of the present invention, in any one of the first to eighth configurations, the data image is a run-length encoded compressed image, and the sample data pixel set extraction unit includes: A set of representative values of luminance of all or part of the data image or the partial data image is extracted as the sample data pixel set.

  According to this configuration, the sample data pixel set extraction unit extracts a set of representative values of the luminance values of all or part of the run-length encoded data image as the sample data pixel set. This facilitates the extraction of the sample data pixel set, so that the image search can be performed more quickly and accurately.

A first configuration of the image search method of the present invention is an image search method for searching for an image that is the same as or similar to a reference image from a plurality of data images, the reference image stored in a reference image storage means A reference range determination step for determining a section [C _max , C _min ] (hereinafter referred to as “reference range”) between the maximum value C _max and the minimum value C _min of the luminance values of all pixels included in the data image, A sample data pixel set extraction step for extracting a part of a pixel set included in the data image (hereinafter referred to as “sample data pixel set”) for the data image stored in the storage means, and belongs to the sample data pixel set A candidate image determining step of determining the data image as a candidate image when the luminance of the pixel satisfies a predetermined requirement in relation to the reference range; and By a detailed comparison of the serial candidate image and the reference image, characterized by having an image selection step, of selecting the reference image and the same or similar image.

  According to a second configuration of the image search method of the present invention, in the first configuration, a partial data image extraction step of extracting a part as a partial data image from the data image stored in the data image storage means. The sample data pixel set extraction step extracts a part of the pixel set included in the partial data image as a sample data pixel set for the partial data image, and the candidate image determination step includes the sample data The partial data image is determined as a candidate image when the luminance of a pixel belonging to the pixel set satisfies a predetermined requirement in relation to the reference range.

  According to a third configuration of the image search method of the present invention, in the first or second configuration, in the candidate image determination step, a ratio belonging to the reference range of the luminance of the pixels belonging to the sample data pixel set is predetermined. The data image or the partial data image is determined as a candidate image when it is equal to or greater than the threshold value.

  According to a fourth configuration of the image search method of the present invention, in the first or second configuration, in the candidate image determination step, the luminance of all pixels belonging to the sample data pixel set belongs to the reference range. The data image or the partial data image is determined as a candidate image.

  According to a fifth configuration of the image search method of the present invention, in the first or second configuration, in the candidate image determination step, the ratio of the luminance of the pixels belonging to the sample data pixel set to the reference range is 80%. In this case, the data image or the partial data image is determined as a candidate image.

According to a sixth configuration of the image search method of the present invention, in the first or second configuration, in the candidate image determination step, an average value C _mean of luminances of pixels belonging to the sample data pixel set is the reference range. The data image or the partial data image is determined as a candidate image.

According to a seventh configuration of the image search method of the present invention, in the first or second configuration, in the candidate image determination step, a luminance average value C _mean and a standard deviation σ of pixels belonging to the sample data pixel set On the other hand, when C _mean + σ and C _mean −σ belong to the reference range, the data image or the partial data image is determined as a candidate image.

  According to an eighth configuration of the image search method of the present invention, in any one of the first to seventh configurations, the image selection step includes a pixel value histogram of the candidate image and a pixel value histogram of the reference image. A difference is calculated, and when the difference is smaller than a predetermined threshold, the candidate image is selected as an image that is the same as or similar to the reference image.

  According to a ninth configuration of the image search method of the present invention, in any one of the first to eighth configurations, the data image is a run-length encoded compressed image, and in the sample data pixel set extraction step, Is characterized in that a set of representative values of brightness of all or part of the run of the data image or the partial data image is extracted as the sample data pixel set.

  A program according to the present invention causes a computer to execute the image search methods having the first to ninth configurations.

  As described above, according to the present invention, since a data image satisfying a predetermined requirement in the relationship between the luminance of the pixel belonging to the sample data pixel set and the reference range is determined as a candidate image, the algorithm for determining the candidate image is simple. Become. For this reason, the present invention requires only a short time for determining candidate images, and has the effect of speeding up image retrieval. Therefore, the present invention is useful as a high-speed image retrieval apparatus and method for retrieving a data image that is the same as or similar to a reference image from among a large number of data images.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram illustrating a configuration of a particle discrimination device including an image search device according to Embodiment 1 of the present invention.

  The image search device 1 includes a reference image storage unit 2, a reference range determination unit 3, a candidate image determination unit 4, a data image storage unit 5, a partial data image extraction unit 6, a sample data pixel set extraction unit 7, and an image selection unit 8. It has.

  The fine particle discriminating apparatus of this embodiment inputs a fine particle image obtained by imaging the fine particles to be discriminated by the CCD camera 11 to the image search device 1 and stores an image similar to the fine particle image in the data image storage means 5. This is a device that discriminates the kind of fine particles by collating with the stored fine particle data image and outputs the result to the output device 12. The output device 12 is not limited to a display device or a printing device, and an external storage device such as a magnetic disk, a communication interface that outputs to a network, or the like may be used.

  The fine particle image picked up by the CCD camera 11 is digitized by the AD converter 13 and temporarily stored in the frame memory 14. A portion in which the particles to be discriminated are captured from the image stored in the frame memory 14 is extracted by the region extraction unit 15 and stored in the reference image storage unit 2 as a reference image.

The reference range determination unit 3 extracts the maximum value C _max and the minimum value C _min of the luminance values of all the pixels included in the reference image read from the reference image storage unit 2, and the reference range [C _max , C _min ]. Is output to the candidate image determination means 4.

  The data image storage means 5 stores compressed image data of repetitive fine particles in the form of run-length encoded compressed image data, and functions as an image database of fine particle specimens. Details of this run-length encoding will be described later.

  The partial data image extraction unit 6 divides the data image read from the data image storage unit 5 into partial data images and outputs the partial data image to the sample data pixel set extraction unit 7.

  The sample data pixel set extraction unit 7 extracts a part of the pixel set (sample data pixel set) included in the partial data image input from the partial data image extraction unit 6 and outputs it to the candidate image determination unit 4. A method for extracting the sample data pixel set will be described later.

  The candidate image determining unit 4 satisfies the predetermined requirements in relation to the reference range input from the reference range determining unit 3 with respect to the luminance of the pixels belonging to the sample data pixel set input from the sample data pixel set extracting unit 7. In this case, the partial data image is determined as a candidate image. The predetermined requirements will be described later.

  The image selection unit 8 selects a partial data image that is the same as or similar to the reference image by performing a detailed comparison between the partial data image that is selected as the candidate image by the candidate image determination unit 4 and the reference image, and the result is obtained. Output to the output device 12. A detailed comparison method will be described later.

  Note that the image search apparatus 1 of this embodiment may be configured as hardware as an LSI chip, but each component may be configured as a functional module by loading a program into a general-purpose computer.

(Run-length encoding compression algorithm)
The fine particle data image stored in the data image storage means 5 quantizes the pixel value of each pixel of the original image data, and the variance of the pixel value of the original image data corresponding to the run in each run is a predetermined threshold value. The run length is determined by determining the run length to be equal to or less than Γ. An example of the compression algorithm will be briefly described below. The compression algorithm is described in detail in Patent Document 1 or Non-Patent Document 1.

First, image data of a two-dimensional original image (hereinafter referred to as “original image data”) is scanned into one-dimensional original image data. This one-dimensional original image data is denoted as {x _i ; i = 1, 2,..., N}. Here, N is the number of pixels of the original image data.

  Next, the following recursive bisection processing is executed on the one-dimensional original image data.

First, the one-dimensional pixel column in the section l is set to {x _i ; i = 1, 2,..., N}. The section l is divided into a section l ₁ and a section l ₂ . The average values of the sections l, l ₁ , and l ₂ are expressed by the expressions (1), (2), and (3), respectively. However, the number of pixels in the interval l ₁ is N ₁ .

In performing the recursive division into _two , assuming that the accumulated square errors of the sections l, l ₁ , and l ₂ are e, e ₁ , and e ₂ , respectively, these are the expressions (4), (5), and (6 ).

The bisection point is N ₁ (1 <N ₁ <N) that minimizes the next evaluation formula.

This two-divided operation is recursively repeated until Eval is equal to or less than a predetermined threshold Γ and N is equal to or less than _Nmax .

  By the algorithm as described above, the one-dimensional original image data is divided into a plurality of sections. Run-length encoding is performed using each divided section as a run and using the average value of pixels in that section (run) as the pixel value in that run. The original image data that has been one-dimensionalized by this run-length encoding has its pixel values quantized in each run, and the variance of the pixel values of the original image data corresponding to that run in each run is below a predetermined threshold Γ. Become.

  Next, the operation of the fine particle discriminating apparatus according to the present embodiment will be described. FIG. 2 is a flowchart illustrating the image search method according to the first embodiment of the invention. The following description will be made with reference to the step numbers attached to this flowchart and the reference numerals attached to FIG. Here, as an example of image search, a case where a partial data image that is the same as a reference image composed of 256 (= 16 × 16) pixels is retrieved from a data image composed of 131072 (= 512 × 256) pixels. Think.

  (Step 1) The reference range determination means 3 reads the reference image from the reference image storage means 2.

  (Step 2) The reference range determining means 3 determines the reference range of the reference image and outputs it to the candidate image determining means 4.

  (Step 3) The partial data image extraction means 6 reads out the data image from the data image storage means 5.

  (Step 4) The partial data image extraction unit 6 divides the data image into 512 partial data images each having 16 × 16 pixels, and sequentially outputs them to the sample data pixel set extraction unit 7.

  (Step 5) If all the partial data images have been searched, the process proceeds to Step 10; otherwise, the process proceeds to Step 6 to continue the partial data image search.

(Step 6) The sample data pixel set extraction means 7 extracts a part of the pixel set (sample data pixel set) included in the partial data image and outputs it to the candidate image determination means 4. Here, the sample data pixel set extraction means 7, the partial data image to select the 25 pixels, each of the luminance values C _1, C _2, obtains the ‥‥ C _25. That is, about 10% of 256 pixels included in the partial image is sampled. The 25 pixels are extracted according to a specific rule (for example, every 10 pixels are picked up from an array of 256 pixels) or a pattern.

(Step 7) The candidate image determining means 4 determines the partial data when the luminance {C ₁ , C ₂ ,... C ₂₅ } of the pixels belonging to the sample data pixel set is greater than or equal to a predetermined threshold. The image (N) is determined as a candidate image. Then, the process proceeds to step 8 to perform a detailed search. Otherwise, return to Step 5.

  (Step 8) The image selection means 8 performs a detailed comparison between the candidate image and the reference image. If both are the same or similar, the process proceeds to step 9; otherwise, the process returns to step 5.

  (Step 9) The image selection means 8 displays the partial data image that is the same as or similar to the reference image on the output device 12, and ends.

  (Step 10) The image selection means 8 displays on the output device 12 that there is no partial image that is the same as or similar to the reference image, and ends.

Here, the “reference range” is a pixel value section for selecting candidate images in step 7, and is between the maximum value C _max and the minimum value C _min of the luminance values of all the pixels included in the reference image. [C _max , C _min ].

Further, in step 7, if the “threshold value” for determining a candidate image is 100%, the condition is too strict, and if a detailed search is performed, the same or similar data image is erroneously excluded by the rough search. May end up. Therefore, the value of less than 100% "threshold", and set to, for example, 80%, sample data pixel intensity of the pixel belonging to the set _{{C 1, C 2, ‥‥} C 25} of, the luminance value of more than 20 points It may be a requirement for selecting candidate images to be in the “reference range” section.

The requirements for determining candidate images are selected according to the purpose and conditions of the search. For example, the average value C _mean of the luminance {C ₁ , C ₂ ,... C ₂₅ } of the pixels belonging to the sample data pixel set exists in the “reference range” section, or the pixels belonging to the sample data pixel set The mean value C _mean and standard deviation σ of the luminance {C ₁ , C ₂ ,... C ₂₅ } are calculated, and it is determined that both C _mean + σ and C _mean −σ belong to the “reference range” section. It may be a requirement for selection.

  The sample data pixel set may be a set of representative values (quantized luminance values) of the luminance of all or part of the run-length encoded data image.

Further, in the detailed comparison between the candidate image and the reference image in step 8, the difference S _i between the brightness histogram data of the candidate image and the histogram data of the reference image is calculated, and if the difference S _i is smaller than a predetermined threshold, Identical or similar. If the histogram data of the candidate image is H _i (x) and the histogram data of the reference image is H ₀ (x), the difference S _i is expressed by the equation (8). However, x, _{x i} represents a pixel value.

It is a block diagram showing the structure of the fine particle discrimination apparatus provided with the image search device which concerns on Example 1 of this invention. It is a flowchart explaining the image search method which concerns on Example 1 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image search device 2 Reference image storage means 3 Reference range determination means 4 Candidate image determination means 5 Data image storage means 6 Partial data image extraction means 7 Sample data pixel set extraction means 8 Image selection means 11 CCD camera 12 Output device 13 AD conversion 14 Frame memory 15 Area extraction means

Claims (19)

In an image search device for searching for an image that is the same as or similar to a reference image from a plurality of data images,
Reference image storage means for storing the reference image;
Data image storage means for storing the data image;
[C _max , C _min ] (hereinafter referred to as “reference range”) between the maximum value C _max and the minimum value C _min of the luminance values of all the pixels included in the reference image stored in the reference image storage means A reference range determining means for determining.
Sample data pixel set extraction means for extracting a part of the pixel set included in the data image (hereinafter referred to as “sample data pixel set”) for the data image stored in the data image storage means;
Candidate image determining means for determining the data image as a candidate image when the luminance of the pixels belonging to the sample data pixel set satisfies a predetermined requirement in relation to the reference range;
Image selection means for selecting an image that is the same as or similar to the reference image by performing a detailed comparison between the candidate image and the reference image;
An image search apparatus comprising: A partial data image extracting means for extracting a part of the data image stored in the data image storage means as a partial data image;
The sample data pixel set extraction means extracts a part of the pixel set included in the partial data image as the sample data pixel set for the partial data image,
The candidate image determining means determines the partial data image as a candidate image when the luminance of a pixel belonging to the sample data pixel set satisfies a predetermined requirement in relation to the reference range. Item 2. The image search device according to Item 1. The candidate image determining means includes
The data image or the partial data image is determined as a candidate image when the ratio of the luminance of pixels belonging to the sample data pixel set belonging to the reference range is equal to or greater than a predetermined threshold value. 2. The image search device according to 2. The candidate image determining means includes
The image search according to claim 1 or 2, wherein when the luminance of all pixels belonging to the sample data pixel set belongs to the reference range, the data image or the partial data image is determined as a candidate image. apparatus. The candidate image determining means includes
3. The data image or the partial data image is determined as a candidate image when the ratio of the luminance of pixels belonging to the sample data pixel set belonging to the reference range is 80% or more. The image search device described in 1. The candidate image determining means includes
3. The data image or the partial data image is determined as a candidate image when an average value C _mean of luminances of pixels belonging to the sample data pixel set belongs to the reference range. Image search device. The candidate image determining means includes
When C _mean + σ and C _mean −σ belong to the reference range with respect to the average value C _mean and standard deviation σ of the brightness of pixels belonging to the sample data pixel set, the data image or the partial data image is selected as a candidate image. The image search device according to claim 1, wherein the image search device is determined as follows. The image selection means includes
A difference between a pixel value histogram of the candidate image and a pixel value histogram of the reference image is calculated, and when the difference is smaller than a predetermined threshold, the candidate image is selected as an image that is the same as or similar to the reference image. The image search device according to claim 1, wherein: The data image is a run-length encoded compressed image,
The sample data pixel set extraction means includes:
9. The image search apparatus according to claim 1, wherein a set of representative values of brightness of all or part of the data image or the partial data image is extracted as the sample data pixel set. . An image search method for searching for an image identical or similar to a reference image from a plurality of data images,
A section [C _max , C _min ] (hereinafter referred to as “reference range”) between the maximum value C _max and the minimum value C _min of the luminance values of all the pixels included in the reference image stored in the reference image storage means. ) To determine the reference range,
A sample data pixel set extraction step for extracting a part of the pixel set included in the data image (hereinafter referred to as “sample data pixel set”) for the data image stored in the data image storage means;
A candidate image determination step of determining the data image as a candidate image when the luminance of the pixels belonging to the sample data pixel set satisfies a predetermined requirement in relation to the reference range;
And an image selection step of selecting an image that is the same as or similar to the reference image by performing a detailed comparison between the candidate image and the reference image.
An image search method characterized by comprising: A partial data image extraction step of extracting a part of the data image stored in the data image storage means as a partial data image;
In the sample data pixel set extraction step, for the partial data image, a part of the pixel set included in the partial data image is extracted as a sample data pixel set,
In the candidate image determining step, the partial data image is determined as a candidate image when the luminance of a pixel belonging to the sample data pixel set satisfies a predetermined requirement in relation to the reference range. The image search method according to claim 10. In the candidate image determination step,
11. The data image or the partial data image is determined as a candidate image when the ratio of the luminance of the pixels belonging to the sample data pixel set belonging to the reference range is a predetermined threshold value or more. 11. The image search method according to 11. In the candidate image determination step,
The image search according to claim 10 or 11, wherein when the luminance of all pixels belonging to the sample data pixel set belongs to the reference range, the data image or the partial data image is determined as a candidate image. Method. In the candidate image determination step,
12. The data image or the partial data image is determined as a candidate image when the ratio of the luminance of pixels belonging to the sample data pixel set belonging to the reference range is 80% or more. The image search method described in 1. In the candidate image determination step,
12. The data image or the partial data image is determined as a candidate image when an average luminance value C _mean of pixels belonging to the sample data pixel set belongs to the reference range. Image search step. In the candidate image determination step,
When C _mean + σ and C _mean −σ belong to the reference range with respect to the average value C _mean and standard deviation σ of the brightness of pixels belonging to the sample data pixel set, the data image or the partial data image is selected as a candidate image. The image search method according to claim 10 or 11, wherein the image search method is determined as: In the image selection step,
The difference between the pixel value histogram of the candidate image and the pixel value histogram of the reference image is calculated, and when the difference is smaller than a predetermined threshold, the candidate image is selected as an image that is the same as or similar to the reference image. The image search method according to claim 10, wherein: The data image is a run-length encoded compressed image,
In the sample data pixel set extraction step,
18. The image search method according to claim 10, wherein a set of representative values of brightness of all or part of the data image or the partial data image is extracted as the sample data pixel set. . A program for causing a computer to execute the image search method according to any one of claims 10 to 18.

Images