JPH1166270A - Method and processor for image processing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically correct a shift of an identical image part and to lighten the burden on an operator by finding shifts of the positions of conduits at the upper and lower end parts of an image to be corrected from the position of a corresponding conduit in a reference position. SOLUTION: Images of two films are read by a scanner and binarized (S2). Vectorization is performed so as to obtain information on positions and information on shapes, conduit by a conduit (S4). The reference image and image to be corrected are divided into blocks of the same size, and images of the conduits included in those blocks are matched to search for corresponding conduits (S5 and S6). The quantities of shifts between conduits which are made to correspond to each other in corresponding blocks of the reference image A and image B to be corrected are calculated and on the basis of them, and the shift quantity of the blocks is found (S7). For canceling the shift quantities of the conduits included in the respective blocks of the image B to be corrected, the positions are shifted in parallel, and vector data of the conduits of the image B to be corrected are converted into raster data (S8 and S9).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method and an apparatus suitable for processing an image of a wood grain conduit.

[0002]

2. Description of the Related Art It is known that, when embossing is applied to decorative paper on which a wood grain pattern is printed, it is desirable to process the wood grain so that the conduit portion is depressed. For this reason, it is necessary to obtain an image of only the conduit portion of the wood grain.The technique is to fill the conduit portion of the wood with a filler such as abrasive powder or toner, photograph the film, and read the film with a scanner. Normally, an image of only the conduit portion is obtained by digitizing and digitizing the digitized image by a predetermined threshold value. Filling a pipe of wood with a filler such as abrasive powder prior to film photography is performed by binarizing the pipe by binarizing it so that the pipe becomes particularly bright or dark in the captured image. This is so that only one can be extracted.

Here, a film obtained by photographing a grain is not used as it is, but is first subjected to an endless process.
This is as follows. First, two films of the same image of a grain are prepared. Assuming that these films are films A and B, first, for film A, FIG.
8 (a), the lower end is cut out by a predetermined width 、, and then, as shown in FIG. 8 (b), the upper end is cut out by the same width 、, and the film B is cut off. The portion is connected to the portion where the portion A of the film A is cut.

Then, a retouching operation is performed on the image of the conduit near the upper side of the connecting portion so that the image of the conduit at the connecting portion is continuous. As a result, a film as shown in FIG. 8C is obtained. In this film, it is clear that the portion having the width の at the upper end and the portion having the width の at the lower end have the same image. . Therefore, FIG.
By overlapping the images of the film shown in (c) by the width △ and connecting them vertically, it is possible to obtain a long image in which the images of the conduit are continuous. That is, the image on the film in FIG. 8C is endless. Although FIG. 8 has described the endless processing of the image of the film conduit in the vertical direction, the processing of the endless processing is similarly performed not only in the vertical direction but also in the horizontal direction.

[0005] Now, the film in which the image of the conduit is made endless as described above is set on the scanner and the film image is read. However, if the film size is large, the film cannot be set on the scanner as it is. There are cases. In other words, the size of the film used for photography varies, and a large one of about 930 mm x 460 mm may be used. However, there is currently a scanner that can set and read such a large film as it is. It does not.

Therefore, another film having the same image is prepared from the film in which the image of the conduit is made endless as described above, and one film A is shown in FIG.
9B, the lower portion is cut off to leave a portion longer than the half in the vertical direction by δ / 2, and for the other film B, the upper portion indicated by the hatched portion in FIG. Leave the part longer by δ / 2. With this size, it can be read by a scanner. Then, these two films are set on a scanner and read. 9A and 9B, the unit of the numerical value is mm, and the broken line indicates the center of the film in the vertical direction. Also, δ is generally set to about 20 mm.

In this way, since the δ portion of the image obtained by reading these two films is exactly the same image, if the δ portion of both images is superimposed and synthesized.
An image of the original size conduit of 30 mm x 460 mm is obtained.

[0008]

However, in practice, even when the two films are carefully aligned when they are set on the scanner, when the images of both images are superimposed on the δ portion, A shift may occur in the image of the conduit in the portion of δ. When such a shift of the conduit image occurs, it is apparent that the conduit image is also uneven, and the commercial value is significantly reduced.

When such a shift occurs, these images are displayed on a monitor and moved, rotated, deformed, etc. with respect to each conduit image manually using appropriate software capable of performing image processing. The retouching operation was performed to correct the misalignment between the two images, but the retouching operation was not only very troublesome and time-consuming, but also took a very large burden on the operator. .

Accordingly, the present invention provides a method for reading a large image by dividing it into two parts so as to include a part of the same image, and superimposing the same image part to obtain one image of the original size. It is an object of the present invention to provide an image processing method and an image processing apparatus which can automatically correct a shift of an image portion and greatly reduce the burden on an operator.

[0011]

In order to achieve the above object, an image processing method according to the present invention comprises a reference image consisting of an image of a conduit, and a lower end and an upper end of the reference image, respectively. An image processing method for synthesizing the same image and the correction target image made, wherein the position of the conduit at the upper end of the correction target image and the position of the conduit at the lower end are different from the position of the corresponding conduit in the reference image. Is determined, the amount of deviation in each part of the correction target image is determined based on the deviation amount, and the position of the image of the conduit of the correction target image is moved so as to correct the deviation amount. .

Further, the image processing apparatus according to the present invention synthesizes a reference image composed of an image of a conduit and an image to be corrected whose upper end and lower end are the same as the lower end and upper end of the reference image, respectively. An image processing apparatus for performing
The position of the conduit at the upper end of the image to be corrected and the position of the conduit at the lower end are determined from the position of the corresponding conduit in the reference image, and the amount of deviation in each part of the image to be corrected is determined based on the amount of deviation. And performing a process of moving the position of the image of the conduit of the correction target image so as to correct the shift amount.

[0013]

Embodiments of the present invention will be described below with reference to the drawings. First, an image processing method according to the present invention will be described. FIG. 1 is a flowchart showing an embodiment of the processing steps in the image processing method.

[Step S1] First, a film from which an image is read by a scanner is prepared. This is the same processing as that performed conventionally. First, the film obtained by photographing the grain is subjected to the endless processing described with reference to FIG. 8, and further, two divided films are created as described with reference to FIG. It is. In this case, the width △ for endlessness shown in FIG. 8 is the same as the width δ shown in FIG. 9, that is, the width when the same image portion is provided when the film is divided into two, and these widths are used. To the subsequent step S5
When each image is divided into blocks, it is desirable that the width of each image is set so that the vertical direction of each image can be divided into an integer number. Here, the width △ and the width δ are selected in this way.

FIG. 2 shows an example of these two films.
FIG. 2A shows a film A as one film,
FIG. 2B shows a film B as the other film. Here, △ = δ, the portion of the width A at the upper end of the film A and the portion of the width の at the lower end of the film B have the same image, and the width of the lower end of the film A is the same. δ
It is clear that the portion and the portion having the width δ at the upper end of the film B are the same image.

[Step S2] The images of the two films prepared in step S1 are read by a scanner and binarized. As a result, an image obtained by binarizing the image of the conduit is obtained. This binary image is raster data. In the following, the 2 obtained from the film A shown in FIG.
The value image is called a reference image, and the film B shown in FIG.
Will be referred to as a correction target image. FIG. 3 shows an example. FIG. 3A shows an example of the reference image A, and FIG. 3B shows an example of the correction target image B. The size of the reference image A and the size of the correction target image B are the same. Hereinafter, as shown in FIG. 3A, the origin of the reference image A is set at the upper left corner, and the horizontal direction is the x axis and the vertical direction is the y axis. The same applies to the correction target image B.

[Step S3] In this step S3,
The sizes of the reference image and the image to be corrected in the x direction and the y direction, that is, the number of pixels are checked.
That is, although the film A and the film B are obtained by dividing the original film into two, when reading with a scanner,
The size may differ from several pixels to several tens of pixels. However, since these two images are combined in a later step as described later, it is not desirable that the image sizes are different. So, in this step,
The size of these two images is made equal in advance.

An appropriate technique can be adopted as a technique for making the size of the reference image and the size of the image to be corrected the same, but a simple technique is to delete pixels of a large-sized image. Good. For example, x of the reference image
The number of pixels in the direction is n, and the number of pixels in the x direction of the image to be corrected is m
When (> n), it is sufficient to delete only (mn) pixels in the x direction of the correction target image to make n pixels. The same applies to the y direction.

[Step S4] In this step, each conduit is vectorized in order to obtain position information and shape information. That is, the raster data of the binary image obtained in step S2 is converted into vector data. Then, serial numbers from number 1 are assigned to the vector data of the conduit. Here, the form of the conduit vector data may be any form as long as information on the position and shape of each conduit can be obtained. Here, the form is as follows.

First, the positions of all the contours on the left side of the conduit are determined, and the distance from the contour position to the contour on the right side of the conduit is determined. However, only one point is determined for the upper vertex and the lower vertex of the conduit. That is, if the upper vertex of the conduit has only one pixel, the position of that pixel is taken as the upper vertex position, and if there are multiple pixels at the upper vertex, a predetermined position in those pixels is used. , For example, the leftmost pixel, the center pixel, or the rightmost pixel is the position of the vertex. Further, one representative point of the conduit is determined. How this representative point is determined is optional,
For example, the center of gravity of the image of the conduit may be determined as the representative point, or the center position of a line connecting the upper vertex and the lower vertex may be set as the representative point. Here, the latter is adopted.

Thus, for the image of one conduit, the coordinates of the representative point, the coordinates of the upper vertex, the coordinates of the lower vertex, and the coordinates of the left contour and the right to Will be obtained.

The following is a description of an example. Assuming that there is a conduit image as shown by hatched portions in FIG. 4, with respect to the conduit, the upper coordinate vertex P _1, the coordinates of the lower apex P _7, contour P ₂ on the left side of the middle coordinates to P _6, the distance from these contours P ₂ to P ₆ to the right of the contour of the conduit, and the data of the coordinates of the representative point Q obtained,
These data collectively become vector data that defines the position and shape of the conduit.

The above-described vectorization processing of the conduit image is performed on all the conduit images of the reference image and the correction target image.
Give each one a serial number. Note that the coordinates in this step are coordinates with the origin at the upper left end of the reference image and the correction target image as shown in FIG.

[Step S5] In this step, the reference image and the image to be corrected are divided into blocks of the same size. In the y direction of each image, the width may be divided by δ (= △). The width in the x direction is arbitrary. However, the x direction may be divided into integers. Here, the correction target image B is shown in FIG.
As shown in (b), 2 of 4 (x direction) × 5 (y direction)
It is assumed that the block has been divided into 0 blocks. Incidentally, B ₁₁ as shown in convenience figure for each block of the correction target image B ~
B ₅₄ is assigned.

The reference image A is similarly divided into blocks, but as will be apparent from the later description, the reference image A has only the upper end block and the lower end block as shown in FIG. Therefore, only the upper end and the lower end need to be blocked as shown in FIG. Each block at the upper end of the reference image A is denoted by reference numerals A _{U1 to} A _U4 for convenience, and similarly, blocks at the lower end are denoted by reference signs A _{D1 to} A _D4. .

Due to such blocking, the images of the block A _U1 and the block B ₅₁ are in principle the same. Similarly, images of block A _U2 and block B ₅₂ , block A
_U3 and images of the block B _53, the block A _U4 and Block B ₅₄
Are in principle the same. Block A
Image _D1 and the block B _11, the block A _D2 and the block B ₁₂
, The image of blocks A _D3 and B ₁₃ , and the images of blocks A _D4 and B ₁₄ are in principle the same.

After the reference image A and the correction target image B are divided into blocks as described above, it is determined to which block the vectorized image of each conduit belongs in step S4. Which block a certain conduit belongs to may be determined based on the coordinates of a representative point of the conduit. Specifically, in the correction target image B, a representative point of a certain conduit is a block B
_ij (i = 1,..., 4: j = 1,..., 5), the conduit is connected to block B _ij even if the end of the conduit protrudes into an adjacent block. Shall belong. That is, in FIG.
_Although it extends over two blocks _{i-1, j} and B _ij , since the representative point Q is inside the block B _ij , this conduit belongs to the block B _ij . This makes it possible to determine which block belongs to all the conduits of the correction target image B. The same applies to the reference image A.

At this time, the origin is determined for each block, and the coordinates of the vector data of each conduit are converted into coordinates based on the origin of the block to which it belongs. The origin may be selected at the upper left corner of each block. Obviously, this coordinate transformation can be easily performed by parallel translation of the coordinates.

[Step S6] In this step, for the blocks having the same image in principle, the shapes of the images of the conduits belonging to those blocks are matched to search for the corresponding conduit. That is, as described above, the block A _U1 of the reference image A and the block B
_Since the images of ₅₁ are in principle the same, the respective conduits belonging to the block B ₅₁ are searched for the same conduits belonging to the block A _U1 by shape matching. In this matching, the position and shape of each conduit may be checked one by one based on the vector data of the conduit. For example, paying attention to one of the conduit blocks B _51, recognizes the position and shape from the vector data within the block B _51, by referring to the vector data of the conduit belonging to the block A _U1, in the block A _U1 It is sufficient to check whether or not there is a conduit having the same shape as the conduit.

By doing so, the block B
_The number of the conduit belonging to ₅₁ and the number of the conduit belonging to the block _AU1 can be searched for, and the conduit can be associated with both blocks. In addition,
Since the reference image A or the correction target image B is likely to contain also an image based on dust, always all conduits belonging to the block B ₅₁ of the correction target image B, the block A _U1 of the reference image A It is not always possible to associate the pipes to which they belong with one-to-one correspondence. Thus, when there is no corresponding conduit in both blocks, there is no need to perform the correspondence.

This processing is performed by _combining block A _U1 and block B ₅₁
Image, image of block A _U2 and block B ₅₂ , image of block A _U3 and block B ₅₃ , block A _U4 and block B
₅₄ images, the image of the block A _D1 and the block B _11, the image of the block A _D2 and the block B _12, the image of the block A _D3 and the block B _13, the image of the block A _D4 and the block B ₁₄ performs.

[Step S7] In this step, how much the pipes associated with each other in the corresponding blocks of the reference image A and the correction target image B are shifted is calculated, and the shift amount is calculated based on the shift amount of the pipes. To determine the amount of deviation as a block. At this time, with reference to the position of the representative point of the conduit in the reference image A, the deviation amount of the position of the representative point of the conduit in the correction target image B associated with this conduit is determined.

[0033] For example, if the conduit D _B1 of the conduit D _A1 and the block B ₅₁ of the block A _U1 is associated, the position of the vector data of these conduits, in block B ₅₁ of the representative point of the conduit D _B1 Is the block A _{U1 at} the representative point of the conduit _DA1.
It is calculated how much it deviates from the position at.

The calculation of such a shift amount is performed by the block A _U1
And performed for all conduits correspondence in both blocks of the block B _51. Obviously, it is not necessary to calculate the amount of deviation for conduits that are not associated.

[0035] Next, determine the amount of deviation of the block B _51, which takes the histogram of the shift amount of the conduit may be the shift amount as the block B ₅₁ with the displacement amount thereof maximal frequency.

The above processing is performed by adjusting the upper edge of the image B to be corrected.
Lock B₁₁~ B₁₄, And block B at the lower end₅₁~ B₅₄To
About it. Thus, block B₁₁, B₁₂,
B₁₃, B ₁₄, B₅₁, B₅₂, B₅₃, B₅₄About
The shift amount of these blocks is calculated.

It should be noted that only the shift amount in the x direction and the y direction is required as the shift amount. Of course, there is a possibility that the image of the conduit of the correction target image B is rotated with respect to the image of the corresponding conduit of the reference image A. However, according to experiments by the inventors, even if the image is rotated, The amount of rotation is very small, and it has been confirmed that even if only the amount of displacement in the x direction and the y direction is calculated, it can be sufficiently handled.

The blocks B ₁₁ , B of the image B to be corrected
_12, for a block in the _{B 13, B 14, B 51} , B 52, B 53, B 54, if it can not calculate a shift amount of the block by some reason, the blocks adjacent to the block The shift amount may be calculated based on the shift amount. For example, for some reason, the conduit belonging to the block B ₅₃ of the correction target image B and the block A _{U3 of the} reference image A
Although it is not possible to calculate the shift amount of the block B ₅₃ in the case where correspondence conduits belonging to could not be at all, in this case, the deviation of the block B ₅₂ adjacent to the right side △ x _52, △ y _52, the amount of deviation △ x ₅₄ block B ₅₄ adjacent to the left side, when a _{△ y 54, = △ x 53} = (△ x 52 + △ x 54) / 2 ... (1) △ y 53 ( Δy ₅₂ + Δy ₅₄ ) / 2 (2) The deviation amount as the block B ₅₃ may be calculated.

In this way, the shift amount is calculated for the blocks at the upper end and the lower end of the image B to be corrected.
Next, the shift amount for the intermediate block is calculated. The shift amount of the intermediate block is calculated by proportionally distributing the shift amount of the upper block and the shift amount of the lower block in the row of the block.

For example, the block B of the image B to be corrected now
_{Assuming that} the shift amounts of the blocks ₁₁ are △ x ₁₁ and 11y ₁₁ and the shift amounts of the block B ₅₁ are △ x ₅₁ and △ y ₅₁ , the intermediate blocks B ₂₁ , B ₃₁ and B ₄₁ of these blocks are The shift amount is as follows. Shift amount △ x ₂₁ block B _21, △ y ₂₁ is _{△ x 21 = △ x 11 +} (△ x 51 - △ x 11) / 4 ... (3) △ y 21 = △ y 11 + (△ y 51 - △ y ₁₁₎ / 4 is calculated by (4), the deviation amount △ x ₃₁ block B _31, △ y ₃₁ is _{△ x 31 = △ x 11 +} 2 * (△ x 51 - △ x 11) / 4 ... ( _{5) △ y 31 = △ y} 11 + 2 * (△ y 51 - △ y 11) / 4 ... ( calculated in 6), the deviation amount △ x ₄₁ block B _41, △ y ₄₁ is △ x ₄₁ = △ x _{11 + 3 * (△ x 51} - △ x 11) / 4 ... (7) * △ y 41 = △ y 11 + 3 (△ y 51 - △ y 11) is calculated / 4 ... (8). The same applies to the shift amount of the middle block in other columns.

[Step S8] After the deviation amount of each block of the image B to be corrected is calculated as described above, the deviation amount of the block to which the conduit belongs next for the conduit belonging to each block of the image B to be corrected The position is translated so as to cancel. That is, when the deviation amount of a certain block B _ij is △ x _ij , △ y _ij , this block B _ij
Are translated in the x-direction _by- △ x _{ij and} translated in the y-direction _by- △ y _ij .

This means that the position of each conduit in the image B to be corrected has been corrected in accordance with the amount of displacement of the block to which the conduit belongs.

[Step S9] In this step, the vector data of the conduit of the image B to be corrected whose position has been corrected in step S8 is converted into raster data. by this,
The correction target image is returned to the binary image.

Further, also for the reference image A, the vector data of the conduit is converted into raster data to obtain a binary image. This binary image is the same as the raster data before converting the image of the conduit of the reference image A into vector data in step S4.

The image corresponding to the block at the upper end of the reference image A converted to the binary image and the image corresponding to the block at the lower end of the correction target image B converted to the binary image are as follows. Reference image A after matching and converted to a binary image
The image corresponding to the block at the lower end of the image corresponds to the image corresponding to the block at the upper end of the correction target image B after conversion to the binary image.

[Step S10] Here, step S9
The reference image A and the correction target image B of the binary image obtained in the above are synthesized. At this time, it is natural that the lower-end block portion of the reference image A and the upper-end block portion of the correction target image B are overlapped and synthesized. Since the positions and shapes of the conduits in these portions also match, it is apparent from the above description that the images of the conduits in the superimposed portions of the two portions clearly match without shifting.

As a result, when reading with a scanner, it is possible to obtain a large image equivalent to the original size before dividing the film into two.

[Step S11] In this step, the composite image obtained in step S10 is output by an appropriate output device. This output device may be a film output machine, or a so-called direct printing machine that directly engraves a cylinder, or may be another device.

The embodiment of the image processing method according to the present invention has been described above. Next, an embodiment of the image processing apparatus according to the present invention will be described.

This image processing apparatus executes the processing of the above-described image processing method, and can be constituted by using a personal computer or a work station, as schematically shown in FIG. 7, reference numeral 1 denotes a scanner, 2 denotes an input port, 3 denotes a control device, 4 denotes a display device, 5 denotes a ROM, 6 denotes a RAM, 7 denotes an input device, 8 denotes a storage device, and 9 denotes an output device.

The scanner 1 reads an image photographed on a film and outputs digital image data.
The input port 2 is for taking in digital image data from an external device. The control device 3 is composed of an appropriate processing unit. The display device 4 is composed of appropriate display means such as a color CRT.

The ROM 5 stores a program for processing executed by the control device 3, and particularly stores a program for executing image processing by the above-described image processing method according to the present invention. The RAM 6 is a memory used by the control device 3 as a work area. The input device 7 includes a keyboard, a mouse, and the like, and forms a man-machine interface together with the display device 4. Storage device 8
Is composed of a large-capacity storage device such as a hard disk or a magneto-optical disk. The output device 9 outputs an image composed of the extracted image of the conduit in a form according to the purpose of use, and when outputting to a film, a film output machine is used. Further, the configuration may be such that the data is output to an external storage device, a direct printing machine, or the like.

First, the operator prepares two films for reading by the scanner 1 (step S1). These two films, as described above,
What is necessary is just to carry out similarly to the prior art. Then, the operator sets these two films on the scanner 1 and enters the input device 7.
To start reading the image. If the scanner 1 is separate, the image data read by the scanner 1 may be taken in via the input port 2.

When the control device 3 captures image data from the scanner 1 or via the input port 2, the control device 3 converts the image data into R data.
The data is developed in the AM 6 and temporarily stored in the storage device 8.

Next, the operator operates the input device 7 to start execution of image processing by the above-described image processing method. Thereby, the control device 3 reads the program for the image processing from the ROM 5 and starts the processing.

First, the control device 3 performs a binarization process on the fetched reference image and the correction target image, and automatically and continuously performs the processes from step S3 to step S10 described above.

Thereafter, when the operator performs an operation for outputting an image of the conduit synthesized by the input device 7,
The control device 3 reads out the image data of the image of the conduit from the storage device 8 and outputs it to the output device 9.

As described above, according to the image processing method and the image processing apparatus, it is possible to automatically correct the displacement of the same image portion between the reference image and the image to be corrected.
The burden on the operator at that time can be greatly reduced.

Although the embodiment of the present invention has been described above, the present invention is not limited to the above embodiment.
It will be apparent to those skilled in the art that various modifications are possible.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating an embodiment of an image processing method according to the present invention.

FIG. 2 is a diagram illustrating an example of two films read by a scanner.

FIG. 3 is a diagram showing a reference image and a correction target image.

FIG. 4 is a diagram illustrating vectorization of an image of a conduit.

FIG. 5 is a diagram for explaining blocking of a reference image and a correction target image.

FIG. 6 is a diagram for explaining, based on the coordinates of a representative point Q of the conduit, determining a block to which the image of the conduit belongs in step S5 of FIG. 1;

FIG. 7 is a block diagram illustrating an embodiment of an image processing apparatus according to the present invention.

FIG. 8 is a diagram for explaining a process of making a film image of a wood grain endless.

FIG. 9 is a diagram for explaining a process of dividing the film into two when setting a film obtained by capturing a grain of wood in a scanner.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Scanner, 2 ... Input port, 3 ... Control device, 4 ... Display device, 5 ... ROM, 6 ... RAM, 7 ... Input device, 8 ...
Storage device 9, output device.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Naoki Kawai 1-1-1, Ichigaya-Kaga-cho, Shinjuku-ku, Tokyo Inside Dai Nippon Printing Co., Ltd. (72) Inventor Tetsuo Shinki 1-chome, Ichigaya-cho, Shinjuku-ku, Tokyo No. 1 Inside Dai Nippon Printing Co., Ltd.

Claims (2)

[Claims] An image processing method for synthesizing a reference image composed of an image of a conduit, and a correction target image whose upper end and lower end are the same as the lower end and upper end of the reference image, respectively. The position of the conduit at the upper end of the image to be corrected and the position of the conduit at the lower end are determined from the position of the corresponding conduit in the reference image, and the deviation in each part of the image to be corrected is determined based on the amount of deviation. An image processing method comprising: obtaining an amount; and moving a position of an image of a conduit of a correction target image so as to correct the shift amount. 2. An image processing apparatus for synthesizing a reference image composed of an image of a conduit and a correction target image whose upper and lower ends are the same as the lower and upper ends of the reference image, respectively. The position of the conduit at the upper end of the image to be corrected and the position of the conduit at the lower end are determined from the position of the corresponding conduit in the reference image, and the deviation in each part of the image to be corrected is determined based on the amount of deviation. An image processing apparatus for calculating an amount and moving a position of an image of a conduit of a correction target image so as to correct the shift amount.