JP2008225636A - Template matching device, camera equipped with template matching device, and program for making computer perform template matching - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a template matching device and a template matching method that can accurately specify a matching position even if an input image has geometric change. <P>SOLUTION: The template matching device includes a first calculating means of calculating a first feature quantity by performing operations by a plurality of color components using a predetermined distance table for the intensity of a known template image containing the color components, an extracting means of extracting partial images for operations from a plurality of different positions in an input image containing a plurality of color components, a second calculating means of calculating a second feature quantity by performing operations by the respective color components using the distance table for intensities of the partial images extracted by the extracting means, an arithmetic means of finding similarities between the template image and partial images by comparing identical color components with each other based upon the first feature quantity and second feature quantity, and a specifying means of specifying positions of matching with the template image in the input image by comparing the similarities found at the respective positions in the input image with each other. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a template matching device that performs template matching on an image, a camera including the template matching device, and a program for performing template matching on a computer.

  Template matching is a process of searching for a partial signal (target) that matches a known template signal from input signals and specifying a target position (matching position). In this process, the matching operation between the partial signal for calculation extracted from the input signal and the known template signal is repeatedly performed while the extraction position of the partial signal for calculation is moved little by little. Then, the matching position is specified by comparing the result of the matching calculation at each position in the input signal.

For the matching operation, a well-known cross-correlation method, residual successive test method, or the like is used (see, for example, Patent Document 1). In these methods, signal strength is calculated between corresponding positions of a partial signal for calculation and a template signal, and the results are totaled over the entire signal.
JP-A-5-81433

  However, the above method for calculating the signal intensity between the corresponding positions of the partial signal for calculation and the template signal has the following problems. For example, in the case of a two-dimensional input signal containing image information, if there is a geometric change (rotation, change in scale, etc.) in the input signal, it is not always possible to accurately identify the matching position, and pseudo matching or matching It may be impossible. In the case of a one-dimensional input signal including audio information, if there is a geometric change (such as a change in scale) in the input signal, pseudo matching or matching may not be possible.

  An object of the present invention is to provide a template matching apparatus that can accurately specify a matching position even if there is a geometric change in an input image.

  The template matching apparatus of the present invention performs a first calculation for calculating a first feature amount by performing an operation for each color component using a predetermined distance table with respect to the intensity of a known template image including a plurality of color components. Means for extracting partial images for calculation from a plurality of different positions in the input image including the plurality of color components, and the distance with respect to the intensity of the partial image extracted by the extraction means A second calculation unit that calculates a second feature amount by performing an operation for each color component using a table, and compares the same color components based on the first feature amount and the second feature amount Thus, the calculation means for obtaining the similarity between the template image and the partial image and the similarity obtained at each position of the input image are compared in size to thereby compare the template in the input image. And a specifying means for specifying the matching position between the rate image.

Preferably, the distance table is a table determined according to the size of the template image, and may be composed of a numerical value group depending on the distance from the center of the template image.
Another template matching apparatus of the present invention adds the intensity of a known template image including a plurality of color components for each color component for all pixels in the template image, or adds the intensity in the template image. First calculating means for calculating a first feature amount by adding each color component for a part of pixels, and partial images for calculation from a plurality of different positions in the input image including the plurality of color components An extracting means for extracting, and the intensity of the partial image extracted by the extracting means is added for each color component for all pixels in the partial image, or the intensity is added to some pixels in the partial image The second calculation means for calculating the second feature amount by adding each color component for the same, and comparing the same color components based on the first feature amount and the second feature amount By calculating the similarity between the template image and the partial image, the template image in the input image is compared by comparing the similarity obtained at each position of the input image. And specifying means for specifying a matching position.

Preferably, the calculation means may calculate the similarity only when a difference between the first feature value and the second feature value is equal to or less than a predetermined threshold value.
A camera equipped with the template matching device described above is also effective as a specific aspect of the present invention.
In addition, a configuration obtained by converting the configuration related to the above invention into a template matching program for realizing processing on image data to be processed by a computer is also effective as a specific aspect of the present invention.

  According to the present invention, it is possible to accurately specify a matching position even if there is a geometric change in the input image.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
(First embodiment)
Here, the template matching apparatus and method according to the first embodiment will be described using the electronic camera 10 shown in FIG. 1 as an example. The electronic camera 10 includes an optical system 11, an optical system control unit 12, an image sensor 13, an image processing unit 14, a recording unit 15, a recording medium 16, an image display unit 17, an operation unit 18, and a CPU 19. The optical system 11 includes a lens, a diaphragm, and the like. The optical system control unit 12 controls the focal position of the diaphragm and lens. The imaging device 13 converts the subject image formed through the optical system 11 into image data and outputs the image data to the image processing unit 14. The image processing unit 14 performs various kinds of digital signal processing such as color signal generation, matrix conversion processing, γ conversion processing, and correction processing, and stores them in a memory (not shown) as an input image. Note that a description of a specific method of each image processing is omitted. The image processing unit 14 performs template matching described later. The recording unit 15 records the image data on the recording medium 16. The recording medium 16 is a removable memory such as a memory card. The image display unit 17 displays an image generated by imaging or an image recorded on the recording medium 16. The operation unit 18 includes a power button and a release button (not shown). The CPU 19 controls each unit in an integrated manner. Note that the CPU 19 performs control such as AF adjustment, AE adjustment, white balance adjustment, main subject detection (face recognition, etc.), automatic shooting, and the like using the result of template matching described later. Further, the CPU 19 detects the presence / absence of a user operation on the operation unit 18 and records a program for executing each process in advance.

Next, template matching in the electronic camera 10 will be described. In template matching, for example, a partial image (target 23) that matches a known template image 22 is searched from the input image 21 shown in FIG. 2, and the position of the target 23 (hereinafter referred to as “matching position (X ₁ , Y ₁ )). ”). The template matching of the first embodiment is performed according to the procedure (steps S1 to S11) of the flowchart shown in FIG. The input image 21 is an image generated by imaging including a through image or an image recorded on the recording medium 16.

In the processing of FIG. 3, the matching operation (S 4 to S 7) between the partial image 24 for calculation extracted from the input image 21 and the template image 22 slightly sets the extraction position (X ₂ , Y ₂ ) of the partial image 24. Repeatedly moving while moving. The matching operations (S4 to S7) are operations for obtaining the similarity between the partial image 24 and the template image 22 based on the moment of inertia described later. The similarity is one of “an index related to similarity” between the partial image 24 and the template image 22 and is an “accurate index related to similarity”.

In the processing of FIG. 3, it is assumed that the input image 21, the partial image 24 for calculation, and the template image 22 each include three color components (that is, RGB components called additive primary colors or light primary colors). Images (21, 24, 22) including RGB components are color images.
The processes (S1 to S11) in FIG. 3 will be described in order.
In step S <b> 1, the CPU 19 calculates an RGB component inertia moment M ₂₂ (hereinafter referred to as “first inertia moment M ₂₂ ”) for each color component from the entire template image 22.

For example, the first inertia moment M ₂₂ of the R component is calculated by the following equation (1) using the density value A _i at the position (x, y) of the R component (FIG. 4) of the template image 22. “I” in the density value A _i is a pixel number in the template image 22.
M ₂₂ = Σ (x _i ² + y _i ² ) A _i (1)
The first moment of inertia M _{22 in} equation (1) is the moment of inertia of the density around the origin (image center), and the moment of inertia of the density around the X axis (M _X = Σx _i ² A _i ) and the Y axis Is equivalent to the sum of the moment of inertia (M _Y = Σy _i ² A _i ).

Further, the moment of inertia of the concentration around the coordinates (x _g , y _g ) is calculated by the following (Expression) 2.
M ₂₂ = Σ ((x _i −x _g ) ² + (y _i −y _g ) ² ) A _i Formula (2)
In addition to the R component, the first inertia moment M ₂₂ of the G component and the B component is obtained from the whole of the G component and the B component of the template image 22 by the same formula as the above formula (1) or formula (2), respectively. Calculated.

  If the size of the template image 22 is fixed, the first half of the equation (1) or (2) is fixed and represented by a distance table. FIG. 5 shows a distance table for a template image having a size of 15 × 15. As shown in FIG. 5, the distance table of the moment of inertia has a large value at the periphery and zero at the center. On the other hand, as shown in FIG. 6, the normal template image has many main subjects such as a person in the center. That is, when the distance table shown in FIG. 5 is used, important information is handled lightly and unimportant information is handled heavy. Therefore, in the first embodiment, the distance table shown in FIG. 7 is used. The distance table in FIG. 7 is a distance table having a large central value and a small peripheral value. The distance table shown in FIG. 7 can be created based on the distance table shown in FIG.

In step S <b> 2, the CPU 19 extracts a partial image 24 for calculation from the input image 21. The calculation partial image 24 has a rectangular shape having the same size (number of pixels) as the template image 22 in the vertical and horizontal directions. The partial image 24 for calculation and the template image 22 are both smaller than the input image 21 (the number of pixels is small).
In step S3, the CPU 19 calculates an RGB component inertia moment M ₂₄ (hereinafter referred to as “second inertia moment M ₂₄ ”) for each color component from the entire calculation partial image 24 extracted in step S2. For example, the calculation of the second moment of inertia M ₂₄ of the R component is calculated in the same manner as in step S 1 described above from the entire R component of the partial image 24. The same applies to the G component and the B component.

In step S4, CPU 19 is in the partial image 24 and the second moment of inertia M ₂₄ of the R component, the template image 22 is compared with the first moment of inertia M ₂₂ of the R component, the two R components of the moment of inertia of the differential Output the absolute value. The output value O _{R in} this case is an index relating to the similarity between the second inertia moment M ₂₄ and the first inertia moment M ₂₂ of the R component.
In step S <b> 5, the CPU 19 compares the second inertia moment M ₂₄ of the G component of the partial image 24 with the first inertia moment M ₂₂ of the G component of the template image 22, and calculates the difference between the inertia moments of the two G components. An absolute value is output (output value O _G ).

In step S <b> 6, the CPU 19 compares the second inertia moment M ₂₄ of the B component of the partial image 24 with the first inertia moment M ₂₂ of the B component of the template image 22, and calculates the difference between the inertia moments of the two B components. and it outputs the absolute value (the output value O _B).
In step S7, CPU 19 includes an output value O _R of the R component in step S4 described above, the output value O _G of the G component in step S5, and the output value O _B of the B component in step S6, a first inertia respectively Normalization is performed by dividing by the color components I _R , I _G and I _B of the moment M ₂₂ . Then, the average of the similarities S _R , S _G and S _B for each color component calculated by normalization is obtained, and the value is set as “similarity between the partial image 24 and the template image 22”.

The degree of similarity (= (S _R + S _G + S _B ) / 3) obtained in step S7 has a tendency that the smaller the value, the higher the similarity between the partial image 24 and the template image 22, and the extraction of the partial image 24. It represents that the position (X ₂ , Y ₂ ) is close to the matching position (X ₁ , Y ₁ ).
When the matching calculation (S4 to S7) is completed, in step S8, the CPU 19 uses the similarity (= (S _R + S _G + S _B ) / 3) as a result of the matching calculation to extract the partial image 24 ( X ₂ , Y ₂ ) are stored in the memory in association with each other. After the processing of these steps S2 to S8, processing is completed for one extraction position in the input image _{21 (X 2, Y 2)} .

In step S9, the CPU 19 determines whether or not to move the extraction position (X ₂ , Y ₂ ) of the partial image 24 for calculation to the next position. If it determines with moving to the next position, CPU19 will progress to step S10, and if it determines with not moving to the next position, it will progress to step S11.
In step S < _b > 10, the CPU 19 performs a process for moving the extraction position (X ₂ , Y ₂ ) of the calculation partial image 24. The CPU 19 moves the extraction position (X ₂ , Y ₂ ) of the partial image 24 from the current position to the next position. Thereafter, the processing returns to step S2, and at the new extraction position (X ₂ , Y ₂ ), extraction of the partial image 24 for calculation (S2) → calculation of the second moment of inertia M ₂₄ of RGB components (S3) → matching The calculation (S4 to S7) → result storage (S8) is repeated.

In this way, by repeating the processing of steps S2 →... → S10 → S2..., The partial image at each position is moved while the extraction position (X ₂ , Y ₂ ) of the partial image 24 for calculation is moved little by little. extracting 24 sequentially, each time extracting the partial image 24, it is possible to perform a matching operation (S4 to S7) based on the first moment of inertia M ₂₂ and the second moment of inertia M ₂₄ of the RGB components.

Then, when the extraction position (X ₂ , Y ₂ ) of the partial image 24 for calculation reaches the end point, and the processing of steps S2 to S8 ends there, the CPU 19 differs in the input image 21 in step S11. Extraction of the partial image 24 having the smallest value and the highest similarity to the template image 22 by comparing the similarity (= (S _R + S _G + S _B ) / 3) obtained at a plurality of positions. The position (X ₂ , Y ₂ ) is identified as the matching position (X ₁ , Y ₁ ). The template matching process for the input image 21 is thus completed.

As described above, the template matching in the first embodiment, based on the first moment of inertia M ₂₂ and the second moment of inertia M ₂₄ of the RGB components, the matching operation by comparing the same color components each other (S4 to S7) And the degree of similarity (= (S _R + S _G + S _B ) / 3) is compared. The RGB component first inertia moment M ₂₂ is a feature amount of each of the RGB components of the template image 22, and the RGB component second inertia moment M ₂₄ is a feature amount of each of the RGB components of the partial image 24 for calculation. . As described above, since the feature amount is obtained as an amount that does not depend on the position of the pixel, even if there is a geometric change (rotation, change in scale, etc.) in the input image 21, it is possible to accurately match the position (X ₁ , Y ₁ ) Can be specified.

Further, by using the distance table having a large center value and a small peripheral value shown in FIG. 7, important information can be sufficiently reflected in template matching.
In the template matching of the first embodiment, the matching calculation (S4 to S7) is performed using all the color components (that is, RGB components) included in each image (21, 24, 22). For this reason, a large amount of information can be secured, and the matching position (X ₁ , Y ₁ ) can be specified more accurately.

Further, in the first embodiment described above, after normalizing the output values of steps S4 to S6 in step S7 of FIG. 3, the average value (= (S _R + S _G + S _B ) / 3) is obtained, Although the value is “similarity between the partial image 24 and the template image 22”, the present invention is not limited to this. After normalizing the output values in steps S4 to S6, the sum (= (S _R + S _G + S _B )) is obtained, and the value may be set as “similarity between partial image 24 and template image 22”. Even in the case of the sum (= (S _R + S _G + S _B )), the smaller the value is, the higher the similarity between the partial image 24 and the template image 22 is, and the extraction position (X ₂ , Y ₂ ) of the partial image 24 is higher. This means that it is close to the matching position (X ₁ , Y ₁ ).

Further, although the distance table of FIG. 7 shown in the first embodiment has been described as an example, the present invention is not limited to this. Any distance table may be used as long as the center value is large and the peripheral values are small.
In the first embodiment described above, an example of an image (21, 24, 22) including three color components (that is, RGB components) has been described, but the present invention is not limited to this. The present invention can be applied regardless of whether the number of color components is two or four or more. That is, the present invention can be applied to a case where a plurality of color components are included.
(Second Embodiment)
The second embodiment is a modification of the first embodiment. In the second embodiment, only different parts from the first embodiment will be described. The second embodiment will be described using the same reference numerals as those in the first embodiment.

In the second embodiment, a case will be described in which a matching calculation between the calculation partial image 24 and the template image 22 is performed based on a density addition value described later. The template matching of the second embodiment uses a “density addition value” instead of the “moment of inertia” of the first embodiment. In the second embodiment, an image (21, 24, 22) including RGB components will be described as an example.
When all the values in the distance table of FIG. 7 shown in the first embodiment are set to “1”, the expressions (1) and (2) described in the first embodiment become the following expressions (3).

N ₂₂ = ΣA _i Equation (3)
N ₂₂ calculated by Expression (3) is an addition value of the density values A _i of all the pixels in the template image 22. In the second embodiment, the first density addition value N ₂₂ is calculated for the template image 22, and the second density addition value N ₂₄ is similarly obtained for the partial image 24.
Template matching of the second embodiment is performed according to the procedure of the flowchart shown in FIG. 8 (steps S21 to S31). The input image 21 is an image generated by imaging including a through image or an image recorded on the recording medium 16.

The processes (S21 to S31) in FIG. 8 will be described in order.
In step S21, the CPU 19 calculates a first density addition value N ₂₂ of RGB components for each color component from the entire template image 22.
In step S22, the CPU 19 extracts a partial image 24 for calculation from the input image 21 as in step S2 of FIG.

In step S23, CPU 19 calculates the total of the partial image 24 for computation extracted in step S22 the second density addition value N ₂₄ of the RGB components for each color component.
In step S24, CPU 19 has a second density addition value N ₂₄ of the R component of the partial image 24 is compared with the first density addition value N ₂₂ of the R component of the template image 22, density sum of the two R components Output the absolute value of the difference. The output value T _{R in} this case is an index related to the similarity between the second density addition value N ₂₄ of the R component and the first density addition value N ₂₂ .

In step S25, CPU 19 has a second density addition value N ₂₄ of the G component of the partial image 24 is compared with the first density addition value N ₂₂ of the G component of the template image 22, density sum of the two G components Is output (output value T _G ).
In step S26, CPU 19 has a second density addition value N ₂₄ of the B component of the partial image 24 is compared with the first density addition value N ₂₂ of the B component of the template image 22, density sum of the two B components The absolute value of the difference is output (output value T _B ).

In step S27, CPU 19 includes an output value T _R of the R component in step S24 described above, the output value T _G of the G component in step S25, and the output value T _B of the B component in step S26, the first concentration, respectively Normalization is performed by dividing by the color components J _R , J _G , and J _B of the added value N ₂₂ . Then, the average of the similarities U _R , U _G , and U _B for each color component calculated by normalization is obtained, and the value is set as “similarity between the partial image 24 and the template image 22”.

In step S28 to step S31, the CPU 19 performs the same processing as that in step S8 to step S11 in FIG. 3 to perform template matching processing.
As described above, in the template matching according to the second embodiment, the matching operation is performed by comparing the same color components based on the intensity of the RGB components (first density addition value N ₂₂ and second density addition value N ₂₄ ). (S24 to S27) are performed, and the degree of similarity (= (U _R + U _G + U _B ) / 3) as a result is compared in magnitude. The RGB component first density addition value N ₂₂ is a feature amount of each RGB component of the template image 22, and the RGB component second density addition value N ₂₄ is a feature amount of each RGB component of the calculation partial image 24. It is. As described above, since the feature amount is obtained as an amount that does not depend on the position of the pixel, even if there is a geometric change (rotation, change in scale, etc.) in the input image 21, it is possible to accurately match the position (X ₁ , Y ₁ ) Can be specified. It can also be expected to reduce the processing load in template matching.

In the template matching of the second embodiment, the matching calculation (S24 to S27) is performed using all the color components (that is, RGB components) included in each image (21, 24, 22). For this reason, a large amount of information can be secured, and the matching position (X ₁ , Y ₁ ) can be specified more accurately.
Note that the processing of the second embodiment may be modified as follows. After step S26 in FIG. 8, the CPU 19 sets the R component output value T _R , the G component output value T _G in step S25, and the B component output value T _B in step S26 to respective predetermined threshold values. Compare. When the output value T _R of the R component, the output value T _G of the G component in step S25, and the output value T _B of the B component in step S26, is less than the predetermined threshold value, CPU 19, the step It is determined that the extraction position (X ₂ , Y ₂ ) of the partial image 24 extracted in S22 is far from the matching position (X ₁ , Y ₁ ). Then, the process proceeds to step S28 without performing the process of step S27. In step S28, the similarity as a result of the matching calculation is set to “out of range” and is associated with the extraction position (X ₂ , Y ₂ ) of the partial image 24. Save to memory.

As described above, only when the difference between the density addition values calculated in step S24 to step S26 is equal to or less than the predetermined threshold value, the processing for calculating the similarity (step S27) can be performed, so that useless processing can be omitted. . Therefore, the template matching process can be expected to be speeded up.
In the second embodiment described above, in step S27 of FIG. 8, after normalizing the output values of steps S24 to S26, the average value (= (U _R + U _G + U _B ) / 3) is obtained, Although the value is “similarity between the partial image 24 and the template image 22”, the present invention is not limited to this. After normalizing the output values in steps S24 to S26, the sum (= (U _R + U _G + U _B )) is obtained, and the value may be set as “similarity between partial image 24 and template image 22”. Even in the case of the sum (= (U _R + U _G + U _B )), the smaller the value is, the higher the similarity between the partial image 24 and the template image 22 is, and the extraction position (X ₂ , Y ₂ ) of the partial image 24 is higher. This means that it is close to the matching position (X ₁ , Y ₁ ).

In the second embodiment described above, an example is shown in which an addition value obtained by adding the density values A _i of all the pixels in the template image 22 and the partial image 24 is used, but an average value may be used. That is, the values obtained by adding the density values A _i of all the pixels in the template image 22 and the partial image 24 and dividing each by the area may be used.
In the second embodiment described above, an example is shown in which the density values A _i of all the pixels in the template image 22 and the partial image 24 are added. However, even if the density values A _i of some pixels are added. good. In this case, in the template image 22 and the partial image 24, the density values A _i of the corresponding portions of the pixels may be added.

In the second embodiment described above, an example of an image (21, 24, 22) including three color components (that is, RGB components) has been described. However, the present invention is not limited to this. The present invention can be applied regardless of whether the number of color components is two or four or more. That is, the present invention can be applied to a case where a plurality of color components are included.
(Third embodiment)
In the third embodiment, a computer will be described as an example.

  FIG. 9 is a diagram illustrating a configuration of a computer 100 according to the third embodiment. As shown in FIG. 9, the computer 100 includes an acquisition unit 101, an image processing unit 114, a recording unit 115, an image display unit 117, an operation unit 118, and a CPU 119. The acquisition unit 101 acquires image data from an external device such as an electronic camera or a recording medium via a wired, wireless, or recording medium drive. The image processing unit 114 is the same as the image processing unit 14 of the first embodiment. The recording unit 115 records image data and the like. The image display unit 117 includes an image display element such as a liquid crystal display element. The operation unit 118 includes a power button, a mouse, a keyboard, and the like. The CPU 119 controls each unit in an integrated manner. In addition, the CPU 119 records a program for executing each process in advance.

The CPU 119 performs the processing described in the flowchart of FIG. 3 of the first embodiment and the flowchart of FIG. 8 of the second embodiment on the image acquired by the acquisition unit 101 and the image recorded in the recording unit 115. The program is recorded.
Then, based on a user instruction via the operation unit 118, processing similar to that described in the flowchart of FIG. 3 of the first embodiment and the flowchart of FIG. 8 of the second embodiment is performed.

As described above, according to the computer 100 of the third embodiment, the same effects as those of the first embodiment and the second embodiment can be obtained.
In each of the above-described embodiments, the example in which the shapes of the calculation partial image 24 and the template image 22 are rectangular has been described, but the present invention is not limited to this. The partial image 24 and the template image 22 do not have to be rectangular. The essential process is the same for other shapes.

  Further, in each of the above-described embodiments, the template matching has been described by taking the electronic camera 10 of FIG. 1 as an example, but the present invention is not limited to this. Other observation devices for observing, inspecting and aligning samples (semiconductor wafers, liquid crystal substrates, printed boards, biological specimens (for example, cells)), electron microscopes that scan the local area of the sample with an electron beam and capture images The present invention can also be applied to devices, surveillance cameras, and the like. Further, the present invention can be applied not only to an image of a local region of a sample but also to an apparatus that captures an image of the entire surface of the sample at once. The same effect can be obtained even when an external computer is connected to the electronic camera of the first and second embodiments and a part of the processing described in the first and second embodiments is realized by the computer. Obtainable.

  In the above-described embodiment, an example in which template matching is performed on an input image (a two-dimensional input signal including image information) has been described. However, the present invention is not limited to this. In addition, the present invention can be similarly applied to the case of performing template matching (signal processing) on a one-dimensional input signal (input speech) including speech information. When a one-dimensional input signal is targeted, the above-mentioned “image density” may be replaced with “signal intensity”. In the case of a one-dimensional input signal, even if there is a geometric change (change in scale, etc.) in the input signal, the matching position can be specified accurately.

It is the schematic of the electronic camera 10 of 1st Embodiment. It is a figure explaining the input image 21, the template image 22, the target 23, and the partial image 24 for a calculation. It is a flowchart which shows the process sequence of the template matching of 1st Embodiment. It is a diagram illustrating a moment of inertia M ₂₂ of the R component. It is a figure explaining a distance table. It is a figure explaining a template image. It is another figure explaining a distance table. It is a flowchart which shows the process sequence of the template matching of 2nd Embodiment. It is the schematic of the computer 100 of 3rd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Electronic camera, 11 ... Optical system, 12 ... Optical system control part, 13 ... Imaging device, 14 * 114 ... Image processing part, 15 * 115 ... Recording part, 17 * 117 ... Image display part, 18 * 118 ... Operation Part, 19.119 ... CPU, 100 ... computer, 101 ... acquisition part

Claims (7)

First calculating means for calculating a first feature amount by performing an operation for each color component using a predetermined distance table with respect to the intensity of a known template image including a plurality of color components;
Extracting means for extracting partial images for calculation from a plurality of different positions in the input image including the plurality of color components;
Second calculating means for calculating a second feature amount by performing an operation for each color component using the distance table with respect to the intensity of the partial image extracted by the extracting means;
An arithmetic means for obtaining a similarity between the template image and the partial image by comparing the same color components based on the first feature amount and the second feature amount;
A template matching device comprising: specifying means for specifying a matching position with the template image in the input image by comparing the similarity obtained at each position of the input image. . The template matching apparatus according to claim 1,
The said distance table is a table defined according to the magnitude | size of the said template image, and consists of a numerical value group depending on the distance from the center of the said template image. The template matching apparatus characterized by the above-mentioned. Add the intensity of a known template image containing multiple color components for each color component for all pixels in the template image, or add the intensity for each color component for some pixels in the template image First calculating means for calculating the first feature amount;
Extracting means for extracting partial images for calculation from a plurality of different positions in the input image including the plurality of color components;
The intensity of the partial image extracted by the extraction means is added for each color component for all pixels in the partial image, or the intensity is added for each color component for some pixels in the partial image. Second calculating means for calculating the second feature amount;
An arithmetic means for obtaining a similarity between the template image and the partial image by comparing the same color components based on the first feature amount and the second feature amount;
A template matching device comprising: specifying means for specifying a matching position with the template image in the input image by comparing the similarity obtained at each position of the input image. . The template matching apparatus according to claim 3,
The template matching apparatus characterized in that the calculation means calculates the similarity only when a difference between the first feature value and the second feature value is equal to or less than a predetermined threshold.   A camera comprising the template matching device according to any one of claims 1 to 4. A first calculation procedure for calculating a first feature amount by performing an operation for each color component using a predetermined distance table with respect to the intensity of a known template image including a plurality of color components;
An extraction procedure for extracting partial images for calculation from a plurality of different positions in the input image including the plurality of color components;
A second calculation procedure for calculating a second feature amount by performing an operation for each color component using the distance table with respect to the intensity of the partial image extracted in the extraction procedure;
A calculation procedure for obtaining a similarity between the template image and the partial image by comparing the same color components based on the first feature amount and the second feature amount;
A specific procedure for specifying a matching position with the template image in the input image by comparing the degree of similarity obtained at each position of the input image with a computer. Matching program. Add the intensity of a known template image containing multiple color components for each color component for all pixels in the template image, or add the intensity for each color component for some pixels in the template image A first calculation procedure for calculating the first feature amount;
An extraction procedure for extracting partial images for calculation from a plurality of different positions in the input image including the plurality of color components;
The intensity of the partial image extracted in the extraction procedure is added for each color component for all pixels in the partial image, or the intensity is added for each color component for some pixels in the partial image. A second calculation procedure for calculating the second feature amount;
A calculation procedure for obtaining a similarity between the template image and the partial image by comparing the same color components based on the first feature amount and the second feature amount;
A specific procedure for specifying a matching position with the template image in the input image by comparing the degree of similarity obtained at each position of the input image with a computer. Matching program.

Images