JPH1173510A - Position detecting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To precisely detect the position of an inspection object by performing a mean compressing process between an inspection image and a correct answer image placed in displacing operation for detection sequentially by a rough detection displacement quantity in the inspection image with relaxed compressibility. SOLUTION: A TV camera 1 picks up an image of the inspection object and a memory 3 stores the correct answer image including the inspection object. In a rough detecting process, a rough position of the inspection object is detected according to the result of a comparing process for comparing compressed images of the correct answer image including the detection object and the inspection image with each other. In a precise detecting process, the position of the inspection object is accurately detected nearby the rough position according to the result of a comparing process for comparing the correct answer image placed in displacing operation for detection sequentially by the rough detection displacement quantity in the detection image and the inspection image after a compressing process with compressibility lower than the last compressibility. This compressing process is a mean compressing process wherein the mean value of luminance in an area is regarded as the luminance of one pixel.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a position detecting device using image processing.

[0002]

2. Description of the Related Art Conventionally, as this kind of position detecting device, there is one disclosed in Japanese Patent Application Laid-Open No. 6-160047. This inspection is performed based on a comparison result of a comparison process of comparing the inspection image and a correct image that has undergone a displacement operation for detection that sequentially displaces by a displacement amount for coarse detection in the inspection image, between the compressed images. After performing coarse detection processing to detect the coarse position, which is the approximate position of the inspection target in the range, the previous compression of the inspection image and the correct image that has been sequentially subjected to the detection displacement operation in the inspection image by the coarse detection displacement amount Based on the comparison result of the comparison processing that compares the images compressed by the compression ratio less than the compression ratio, the precise detection of the position of the inspection target in the inspection range in the vicinity of the coarse position is performed more accurately. By performing the processing, the position of the inspection target in the inspection range is detected in multiple stages.

More specifically, the compression processing is performed by a so-called thinning-out processing for extracting a part of a plurality of pixels constituting a correct image.

[0004]

In the above-described conventional position detecting device, the position of the inspection target in the inspection range is detected in multiple stages. The detection time is shorter than that of accurately detecting from the beginning.

[0005] However, since this image is compressed by the thinning process, the image extracted from a plurality of pixels constituting the correct image is not an average value but a specific value. In this case, it becomes impossible to accurately and precisely detect the position of the inspection target in the inspection range.

The present invention has been made in view of the above points, and an object of the present invention is to provide a position detecting device capable of accurately detecting the position of an inspection object in an inspection range. .

[0007]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problem, the invention according to claim 1 is a method for performing a detection displacement operation for sequentially displacing an inspection image and a coarse detection displacement amount within the inspection image. Based on a comparison result of a comparison process of comparing images and compressed images, a coarse detection process of detecting a rough position that is an approximate position of the inspection target in the inspection range, and then the inspection image and the A comparison result of a comparison process in which the correct image that has been subjected to the detection displacement operation that is sequentially displaced by the coarse detection displacement amount in the inspection image is compared with the images that have been subjected to the compression processing at a compression ratio that is less than the previous compression ratio. Based on the, in the position detection device that performs a fine detection process of detecting the position of the inspection target in the inspection range more accurate near the coarse position, and detects the position of the inspection target in the inspection range in multiple stages Serial compression process, are the average is compressed configuration.

According to a second aspect of the present invention, in the first aspect of the present invention, the method further includes a narrowing-down process of narrowing down the plurality of coarse positions after detecting the plurality of coarse positions by the rough detection process. It is configured to perform fine detection processing.

According to a third aspect of the present invention, in the invention according to any one of the first and second aspects, the fine detection displacement is smaller than the coarse detection displacement. The configuration is smaller than the displacement amount for coarse detection.

According to a fourth aspect of the present invention, in the first aspect of the present invention, the coarse detection displacement operation is a parallel movement operation, and the fine detection displacement operation is a rotation and a parallel movement. It is configured to be a composite displacement operation.

According to a fifth aspect of the present invention, in the invention according to any one of the first to fourth aspects, the comparison processing involves a sum-of-products operation. The configuration is made by calculation means that is executed specially.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to FIGS. This position detection device
TV camera 1, monitor 2, memory 3, capture memory 4A, 4B, display memory 5A, 5B, over memory
6, LUT (Look Up Table) 7, CPU8, ASIC9
, And a scan memory (not shown).

The TV camera 1 captures an image of the inspection object together with the monitor 2. The memory 3 stores the correct image including the inspection target. The access to the memory 3 is switched between the access to the CPU 8 and the access to the ASIC 9 by the changeover switch SW1, and is not simultaneously accessed by the ASIC 9 and the CPU 8.

The capture memories 4A and 4B capture the inspection image of the inspection range where the inspection object exists from the TV camera 1 and record the inspection image. This capture memory
4A and 4B are connected to the CPU 8 by the changeover switches SW2a and SW2b.
Access and ASIC9 access and TV camera 1
And the connection to the camera is switched, so that when one of the capture memory 4A and the capture memory 4B is connected to either the CPU 8 or the ASIC 9, the other is connected to the TV camera 1. . Therefore, when one of the capture memory 4A and the capture memory 4B is being processed by either the CPU 8 or the ASIC 9, the other is connected to the TV camera 1 so that the next inspection image is captured. It has become.

The display memories 5A and 5B store images to be displayed on the monitor 2, and are provided with a changeover switch SW3.
By a and SW3b, access to the CPU 8, connection to the TV camera 1 for capturing an inspection image, and connection to the monitor 2 for display are respectively switched.

The over memory 6 draws characters and figures and displays them on the monitor 2 so as to overlap with the inspection image. LUT (L
The ook up table 7 has a table for determining output values to the monitor 2.

The CPU 8 executes general-purpose calculations, and executes calculations of all processes other than the calculations performed by the ASIC 9.

The ASIC (calculation means) 9 is suitable for repeatedly performing simple multiplication and addition, and specially executes a product-sum operation (Σxy) as a target calculation. This sum-of-products operation is a summation calculation when x = 1, a narrowly-defined sum-of-products calculation when x ≠ y, and a square-sum calculation when x = y. The operation of the ASIC 9 and the product-sum operation expression will be described later in detail.

The scan memory stores scan data for actually operating the ASIC 9, and constitutes a memory section together with the memory 3 and the capture memories 4A and 4B. The scan data is created by the CPU 8.

Next, the operation of the ASIC 9 will be described in detail with reference to FIG. First, when the CPU 8 selects the ASIC 9 and transfers the scan data, the ASIC 9
Memory selection and write mode selection are performed. Thus, the memory unit writes the scan data to the scan memory. Then, when a checker device (not shown) is started, the CPU 8
The ASIC 9 generates a memory selection and a read address of the scan memory. Thus, the memory section reads data from the scan memory. In response to the reading of the data, the ASIC 9 checks the number of data, sets a mode, sets various parameters, generates scan data, and reads data from the capture memories 4A and 4B, which are video memories, and the memory 3. . ASIC
9 performs, based on the set mode select, the read data, performs a gray-scale data count, that is, a sum-of-products calculation in a narrow sense, and a sum-of-squares calculation, and reads the result of the sum-of-products calculation. It makes a request to CPU8. CPU
8 performs timing adjustment and reads the result of the product-sum operation.

Next, the overall operation of this device will be described. As shown in FIG. 1, a correct image including an inspection target that has undergone a displacement operation for detection that is sequentially displaced by a coarse detection displacement amount in an inspection image is compared with an inspection image between images that have been subjected to compression processing as shown in FIG. Based on the comparison result of the comparison process, the detection displacement is performed by performing a coarse detection process of detecting a rough position, which is an approximate position of the inspection target in the inspection range, and then sequentially displacing the coarse detection displacement amount in the inspection image. The position of the inspection target in the inspection range is set to the coarse position based on the comparison result of the comparison processing in which the operated correct image and the inspection image are compared with each other at the compression ratio relaxed from the previous compression ratio. The multi-stage detection for detecting the position of the inspection target in the inspection range in multiple stages by performing the fine detection process for more accurately detecting in the vicinity of, and the compression ratio in both the X and Y directions from 1/16 1/8, 1/4 Is carried out in five steps was relaxed to 1 / 2,1 / 1 and sequentially. Here, the compression processing refers to an average compression processing in which an average value of luminance in a square area is calculated and the calculated value is set as the luminance of one pixel of the compressed image. For example, if the compression ratio is 1/16, the average value of the luminance in a region of 16 × 16 pixels is the luminance of one pixel of the compressed image.

Next, the operation of this device will be described in detail with reference to FIG. First, the inspection image and the correct image are compressed with the above-mentioned compression ratio. then,
As shown in FIG. 4, the compressed correct answer image is rotated by one of the displacement operations by a displacement angle for detection (displacement amount for detection) θ, and then subjected to a parallel movement, one of the displacement operations. So-called raster scanning is performed in order within a predetermined range by a detection displacement length (detection displacement amount), which is one pixel of the compressed image. The displacement operation including the rotation and the parallel movement is performed until the rotation angle reaches the predetermined angle φ.

The predetermined angle φ in the first stage of the detection process
Is 360 °, for example. The detection displacement angle θ is obtained by Expression (1) that calculates the angle at which the pixel farthest from the rotation center is displaced by one pixel. In the equation (1), Dist indicates the distance between the rotation center and the pixel farthest from the rotation center. This Dist is the distance between the center of rotation and the pixel at the position farthest from the center of rotation, and thus gradually increases as the compression ratio is reduced.
Accordingly, the detection displacement angle θ gradually decreases as the compression ratio is reduced. That is, the displacement angle for fine detection used in the fine detection process is smaller than the displacement angle for coarse detection used in the coarse detection process.

The detection displacement angle θ = Sin ⁻¹ (1 / Dist) (1) On the other hand, the predetermined range in the first-stage detection processing is the entire region of the compressed image. As described above, since the detection displacement length is one pixel of the compressed image, the displacement length decreases as the compression ratio is reduced. That is, the displacement length for fine detection used in the fine detection process is smaller than the displacement length for coarse detection used in the coarse detection process.

After the first-stage detection process is performed, the next-stage detection process is performed. As shown in FIG. 5, the predetermined angle φ in the detection processing of the next and subsequent steps is, as shown in FIG. 5, a value obtained by subtracting an angle obtained by adding the detection displacement angle θ of the previous step about the angle at which the approximate position of the inspection target is detected. Between angles. FIG. 5 is a diagram for explaining the predetermined angle φ, and the compression process is not performed.

As shown in FIG. 6, the predetermined size LL defining the predetermined range in the next detection processing is, in any of the X and Y directions, centered on the coarse position where the approximate position of the inspection object is detected. It is assumed that the position is between the position where the dimension value L composed of (the next stage of the current compression ratio) / (the previous compression ratio) is added and the position where the dimensional value L is subtracted. For example, if the previous compression ratio is 1/16 and the next compression ratio is 1/8, the dimension value L is 2, and the predetermined dimension LL in the next stage of detection processing is the position detected in the previous stage. The center is within a range of ± 2 pixels. FIG.
Is a diagram for explaining a predetermined size LL, in which a compression process is not performed.

After the above displacement operation, a correlation value R indicating the degree of coincidence between the correct image and the inspection image in the overlapping area where the correct image and the inspection image overlap is calculated by Expression (2) involving the product-sum operation, The correlation value R is compared, and the position of the inspection target in the detection inspection range is detected by obtaining the combination of the coordinate and the rotation angle at which the correlation value R becomes the maximum.

In equation (2), f (x, y) indicates the luminance value of the correct image at coordinates (x, y), and g (x, y)
y) indicates the luminance value of the inspection image at the coordinates (x, y), and S
Indicates a superimposed area where the correct image and the inspection image are superimposed, and m
n indicates the number of pixels in the overlapping area.

[0029]

(Equation 1)

When the position of the inspection object in the detection inspection range is detected by the above-described procedure, for example, the size of the inspection image is 512 × 480 pixels, and the size of the correct image is 128.
When × 128 pixels and the displacement angle θ for detection at the final stage are 1 °, the total number of calculation processes involved in the detection process is 19,300, and the total processing time is about 1 second. This is because the total number of calculation processes when detecting the position of the inspection target in the detection inspection range by one-stage detection without performing multi-stage detection is four.
The number of detection processes and the detection time are significantly reduced as compared with 8 million times and the total processing time is several seconds. In addition, about 98 of the above-mentioned 19300 processing times
% Is the number of times of processing in the first-stage detection processing.

In such a position detecting device, the image obtained by the average compression process is different from the image obtained by the so-called decimation process in which a part of the plurality of pixels constituting the correct image is extracted. Has an averaged value, so even if a plurality of pixels have a specific value, since the specific value is averaged, the pixel does not have a specific value itself. In addition, the position of the inspection target in the inspection range can be accurately detected.

Further, by sequentially displacing the fine detection displacement amount smaller than the coarse detection displacement amount, it is possible to displace the fine detection displacement amount in the same order as the coarse detection displacement amount.
This makes it possible to perform a fine and precise detection process, and to achieve the effect of being able to accurately detect the position of the inspection target in the inspection range.

Further, since the product-sum operation is performed by the ASIC 9 that executes a target calculation exclusively, and is not performed by the CPU 8 used for general-purpose calculation, the time required for the product-sum operation is longer than that of the CPU 8. Can be shortened. For example, if the size of the inspection image is 512 × 4
When the product-sum operation is executed only by the CPU 8 when the size of the correct image is 80 × 80 pixels and the size of the correct image is 128 × 128 pixels, the processing time takes several seconds. In the present embodiment, the processing time is reduced to about 1 second. .

Next, a second embodiment of the present invention will be described below with reference to FIG. The same members as those in the first embodiment are denoted by the same reference numerals, and only the parts different from those in the first embodiment will be described. In the first embodiment, at the time of the detection processing at each stage,
While the position of the inspection target is detected by obtaining the coordinates and the rotation angle at which the correlation value R becomes the maximum, in the present embodiment, at the time of the detection processing of each stage other than the final stage,
After obtaining a plurality of combinations of coordinates and rotation angles that have a relatively high correlation value R and exceed a threshold, when detecting the next stage, narrow down the plurality of combinations obtained in the previous stage. Detection accompanied by processing is performed.

More specifically, for example, in the first stage of the detection process, as shown in FIG. 8, when there are combinations of the coordinates and rotation angles of the top seven positions where the correlation value R is not so low, the correlation value R The top five combinations having a relatively high R are selected, and when the combinations having the correlation value R exceeding the selection threshold are the top four combinations, the combinations up to the top four, that is, the first to fourth combinations 4th combination C ₁ , C ₂ ,
By obtaining only C ₃ and C ₄ , a plurality of positions to be inspected are detected, and in the next stage of the second stage of the detection process,
Based on each of the top four detection positions, a predetermined range and a detection displacement length L for parallel movement as one of the displacement operations, a predetermined angle φ for rotation as one of the displacement operations, and detection. Is calculated, and a second-stage detection process is performed.

In such a position detecting device, when only one position is detected in the detection processing in the previous stage, the detected position is not accurately detected due to a factor such as noise and an error is detected. Although it may be possible to detect the noise, if a plurality of positions are detected in the detection process in the previous stage, and then a plurality of detection positions are narrowed down in the detection process in the next stage, noise or the like may occur. Due to the above factor, even if the plurality of detected positions include an error that is not accurately detected and has an error, an error occurs during the narrowing-down process of narrowing down the detected position. Since the object can be eliminated, the effect that the position of the inspection target in the inspection range can be accurately detected can be further exhibited.

Further, similarly to the first embodiment, the position of the inspection object in the inspection range can be accurately detected by sequentially displacing the fine detection displacement amount smaller than the coarse detection displacement amount. Can be played further.

Further, as in the first embodiment, since the product-sum operation is performed by the ASIC 9 which executes the target calculation exclusively, the time required for the product-sum operation can be shorter than that by the CPU 8.

Next, a third embodiment of the present invention will be described below with reference to FIG. The same members as those in the first embodiment are denoted by the same reference numerals, and only the parts different from those in the first embodiment will be described. In the first embodiment, in the first stage of the detection processing, a parallel movement and a rotational displacement operation are performed.
In the present embodiment, the displacement operation is performed only by the parallel movement.

More specifically, in the first stage of the detection processing, after the inspection image and the correct image have been compressed at a compression ratio of, for example, 1/4, the compressed correct image is displaced in parallel. The movement is performed in a predetermined range, that is, in the entire region of the compressed image, by the detection displacement length, which is one pixel of the compressed image.

Thus, after the first-stage detection processing is performed, the second-stage detection processing involving translational and rotational displacement operations is performed. The compression ratio after the second stage is 1 /
It is set to 1. The predetermined range after the second step is a dimension value L of (current compression rate for next detection processing) / (previous compression rate) centering on the position where the approximate position of the inspection target is detected. Between the position at which is added and the position at which it is subtracted. For example, the compression ratio in the detection processing in the previous stage is 1/4,
If the current compression ratio for the next-stage detection processing is 1/1, the dimension value L becomes 2, and the predetermined range of the next-stage detection processing is ± 4% around the position detected in the previous stage. This is the range of pixels. Further, the displacement length for detection after the second stage is 1
/ (2 × previous compression ratio). For example, if the previous compression rate is 1/4 and the current compression rate is 1/1, the detection displacement length is 2 pixels. On the other hand, the displacement length for detection in the first-stage detection process is one pixel of the compressed image having a compression ratio of ４ and four pixels in the uncompressed image. The displacement length for detection is smaller than the displacement length for detection. In addition, when the displacement length for detection is less than one pixel, with the progress of the detection step,
The predetermined range and the detection displacement length are not changed.

On the other hand, the predetermined angle φ in the second stage is, for example,
± 16 °. In addition, the predetermined angle φ after the third stage
Is ± (1 / previous compression ratio). For example, if the previous compression ratio is 1/4 and the current compression ratio is 1/1, the predetermined angle φ is ± 4 °. Since the predetermined angle φ after the third stage is ± (1 / compression ratio of the previous time), when the compression ratio is smaller than 1/1, it decreases as the detection stage proceeds. In addition, the detection displacement angle θ is a predetermined angle φ of 1
/ 4. For example, if the predetermined angle φ is ± 16 °, the displacement angle for detection is 8 °. This detection displacement angle θ
Is 1/4 of the predetermined angle φ, so that the predetermined angle is
As mentioned above, as the detection phase progresses,
The detection displacement angle θ itself becomes smaller as the detection stage proceeds. When the detection displacement angle θ exceeds the limit accuracy, the predetermined range and the detection displacement length are not changed with the progress of the detection step.

In such a position detecting device, the first-stage detection processing, which is a coarse detection processing for detecting an approximate coarse position, is a fine detection processing for detecting a more accurate fine position near the coarse position. Compared with the detection process in the second stage, the detection range is wide and the amount of calculation for the detection process is large, so that it takes a long time for the process, but only involves a displacement operation of only translation. The time required for the processing is shorter than in a case involving a displacement operation combining rotation and translation. More specifically, for example, the size of the inspection image is 512 × 480 pixels, and the size of the correct image is 128 × 12.
8 pixels, the predetermined angle φ in the second stage detection processing is ± 16
°, when the compression ratio in the first-stage detection processing is 1/4 and the limit accuracy of the angle is 1 °, the total number of calculation processing involved in the detection processing is 880, and the time required for the detection processing is also 400 ms.
ec.

Further, similarly to the first embodiment, the position of the inspection target in the inspection range can be accurately detected by sequentially displacing the fine detection displacement smaller than the coarse detection displacement. Can be played further.

Further, as in the first embodiment, the product-sum operation is performed by the ASIC 9 which executes the target calculation exclusively, so that the time required for the product-sum operation can be shorter than that by the CPU 8.

Next, a fourth embodiment of the present invention will be described below with reference to FIG. Note that the same members as those of the third embodiment are denoted by the same reference numerals, and only different points from the third embodiment will be described. In the third embodiment, the position of the inspection target is detected by calculating the coordinate and the rotation angle at which the correlation value R is maximum at the time of the second-stage detection processing.
In the present embodiment, as in the first to fourth stages of the detection processing of the second embodiment, a plurality of combinations of coordinates and rotation angles having a relatively high correlation value R and exceeding a threshold are obtained. At the time of detection in the third stage processing, detection accompanied by a narrowing-down process for narrowing down a plurality of combinations obtained in the second stage detection processing is performed.

In such a position detecting device, in addition to the effects of the third embodiment, similarly to the second embodiment, among the plurality of detected positions, it is not necessary to accurately detect the positions. Even if an object having an error is included, an object having an error can be excluded during the narrowing-down process for narrowing the detection position, so that the position of the inspection target in the inspection range can be accurately detected. The effect can be further enhanced.

In the first to fourth embodiments, the detection displacement angle θ may decrease as the detection step proceeds. For example, when the detection displacement angle θ is extremely small, the detection step proceeds. It does not have to be made smaller as it goes.

The procedure for reducing the displacement length for detection and the displacement angle θ for detection as the detection step proceeds is not limited to the procedure shown in the first to fourth embodiments.

In the first to fourth embodiments, the multiply-accumulate operation is performed by the ASIC 9 which executes the target calculation exclusively. For example, when the inspection image is small, the calculation for the detection process is not performed. If not, the processing may be performed by the CPU 8.

[0051]

According to the first aspect of the present invention, an image obtained by the average compression processing is different from an image obtained by so-called thinning-out processing in which a part of a plurality of pixels constituting a correct image is extracted.
Since the values of a plurality of pixels have an averaged value, even if there are those having specific values for a plurality of pixels,
Since the specific value is averaged, the specific value does not have the specific value itself, so that the position of the inspection target in the inspection range can be accurately and precisely detected.

According to the second aspect of the present invention, when a single coarse position is detected by the coarse detection process, the detected coarse position is not accurately detected and has an error due to factors such as noise. Although it is possible in some cases, if a plurality of coarse positions are detected by the coarse detection process, and then a narrowing process of narrowing down the plurality of coarse positions by the fine detection process is performed, the extracted plurality of coarse positions may be lost due to factors such as noise. Even if the position includes an error that is not accurately detected and has an error, it is possible to exclude an error in the narrowing process for narrowing the coarse position. The effect of the invention described in claim 1 that the position of the inspection target in the range can be accurately detected can be further exhibited.

According to the third aspect of the present invention, the displacement is sequentially displaced by a fine detection displacement smaller than the coarse detection displacement, thereby sequentially displacing by the same fine detection displacement as the coarse detection displacement. More precisely, it is possible to perform a finer detection process than to displace, and it is possible to more accurately detect the position of the inspection target in the inspection range than to the invention according to claim 1 or 2. The effects of the described invention can be further achieved.

According to a fourth aspect of the present invention, in addition to the effect of any one of the first to third aspects, the rough detection processing for detecting the rough position is more accurate in the vicinity of the rough position. Compared to the fine detection processing for detecting the position, the detection range is wider and the processing time is longer, but the displacement operation involves only the parallel movement operation. The time required for the processing is shorter than that in the case where.

According to a fifth aspect of the present invention, in addition to the effects of the first to fourth aspects, the product-sum operation is performed by an adder for executing a target calculation exclusively. In addition, since the calculation is not performed by calculation means such as a CPU used for general-purpose calculation, the time required for the product-sum operation can be shorter than that of the calculation means such as the CPU.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing a compression process according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of the above.

FIG. 3 is an explanatory diagram showing operations of a CPU, an ASIC, and the like in the embodiment.

FIG. 4 is an explanatory diagram showing a rotational displacement operation of the compressed reference image.

FIG. 5 is an explanatory diagram showing a predetermined angle of rotation in a next stage of the above.

FIG. 6 is an explanatory diagram showing a predetermined range of parallel movement in a next stage of the above.

FIG. 7 is a flowchart showing a detection operation of the above.

FIG. 8 is an explanatory diagram showing a plurality of positions detected by the first-stage detection processing according to the second embodiment of the present invention.

FIG. 9 is a flowchart illustrating a detection operation according to the third embodiment of the present invention.

FIG. 10 is a flowchart illustrating a detection operation according to a fourth embodiment of the present invention.

[Explanation of symbols]

 9 ASIC (Calculation means)

Claims (5)

[Claims] 1. A comparison process of comparing an inspection image and a correct image that has been subjected to a detection displacement operation in which the inspection image is sequentially displaced by a coarse detection displacement amount within the inspection image, based on a comparison result of a comparison process. After performing a coarse detection process of detecting a coarse position that is an approximate position of an inspection target in an inspection range, and then performing a detection displacement operation of sequentially displacing the inspection image and a coarse detection displacement amount in the inspection image, Based on the comparison result of the comparison processing that compares the image and the compressed image with the compression ratio relaxed from the previous compression ratio, the position of the inspection target in the inspection range is more precisely adjusted in the vicinity of the coarse position. A position detecting device that performs a fine detection process for detecting a position and detects a position of an inspection target in an inspection range in multiple stages, wherein the compression process is an average compression process. 2. The method according to claim 1, wherein a plurality of the coarse positions are detected by the coarse detection process, and then the fine detection process is performed together with a narrowing process for narrowing down the plurality of coarse positions. Item 4. The position detecting device according to Item 1. 3. The position detecting device according to claim 1, wherein the fine detection displacement amount is smaller than the coarse detection displacement amount. 4. The displacement operation for coarse detection is a translation operation, and the displacement operation for fine detection is a composite displacement operation in which rotation and translation are combined. 3. The position detecting device according to claim 1. 5. The method according to claim 1, wherein the comparison process involves a sum-of-products operation, and the sum-of-products operation is performed by a calculation unit that specially executes a target calculation. The position detecting device according to any one of the above.