JPH09266531A - Inteprolative operation method for image data and device for executing the method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform interpolative operation for image data so as to an enlarged image which has neither of longitudinally nor laterally extending edge parts affected and makes an obliquely extending edge part smooth and sharp. SOLUTION: According to primary image data (source image data) Sorg stored in a storage means 10, a decision means 312 detects the edge part which extends slantingly and its extending direction, and an image data selecting means 32 for interpolation determines primary image data Sorg at three points nearby a point to be interpolated including two primary image data along the extending direction of the edge part as a base for calculating secondary image data (interpolated image data) to be interpolated. Then a triangular plane interpolative operation means 33 calculates secondary image data at the point to be interpolated according to the three determine primary image data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data interpolation calculation method and an apparatus for implementing the method.

[0002]

2. Description of the Related Art Conventionally, various methods have been known in which a radiographic image recorded on a radiographic film is photoelectrically read to obtain an image signal, the image signal is subjected to appropriate image processing, and then the image is reproduced and recorded. In the field. In addition, radiation image information of a subject such as a human body is temporarily recorded on a sheet-shaped stimulable phosphor, and the stimulable phosphor sheet is scanned with excitation light such as laser light to generate stimulated emission light. A radiation image recording / reproducing system that photoelectrically reads the stimulated emission light to obtain image data and outputs a radiation image of the subject as a visible image to a recording material such as a photographic photosensitive material or a CRT based on the image data is already available. Has been put to practical use. This system has the practical advantage of being able to record images over a very large radiation exposure area compared to conventional radiographic systems using silver halide photography.

In a system for reproducing a visible image based on the image data obtained as described above, when it is desired to observe the region of interest to be observed in the visible image in more detail, the region is enlarged and reproduced. I have something to do. This enlarged image is a secondary image data having a data number different from the original image data number by performing a predetermined interpolation calculation on the original image data of the sample points obtained by reading the original image. Is obtained, and the visible image is reproduced based on this interpolated image data. In this case, some interpolation points may overlap the sample points depending on the enlargement ratio.

By the way, sample points (pixels) for carrying respective image data, which are generally used from the viewpoint of ease of construction of the image input / output device, are arranged at predetermined intervals in a square lattice shape in the vertical and horizontal directions. In the case where the images are arranged to form an image, the interpolation operation in the image enlargement processing is performed by interpolating the interpolated image data, and original four points near the point (interpolation point) to be newly set (interpolation point). This is done by linearly interpolating the image data.

For example, as shown in FIG. 7 (A), with respect to the pixels S (points represented by the symbols ◯) of the original image arranged in a square lattice pattern, the intervals at which the pixels S are arranged are different. When it is desired to obtain the interpolated image data of the interpolated points S ′ (points represented by X symbols) arranged at intervals, for example, the interpolated point S ′ ₀ is obtained by the following procedure.

Image data S _A , S _B , S _{C of} four neighboring pixel images S _A , S _B , S _C , S _D (a unit lattice forming a square lattice) a surrounding the interpolation point S ′ ₀ , S _D (for simplification, the same symbol as the pixel symbol is used). This means that a square mask of the unit lattice a including the interpolation point S ′ ₀ is set and the original image data of the sample points in this mask is used.

Here, between the pixels S _{A and} S _B of the original image, S _C and
The pitches between S _D , S _{A to} S _C , and S _{B to} S _D are set to 1, respectively, and x from the pixel S _A (S _C ) of the interpolation point S ′ ₀
The distance in the axial direction (horizontal direction) is Tx (see FIG. 7B), and the distance in the y-axis direction (vertical direction) from the pixel S _A (S _B ) is Ty.
In case of, the interpolated image data Sm and Sn of the interpolated points Sm and Sn corresponding to the position of the interpolated point S ′ _{0 in} the x-axis direction are
It is obtained by the calculation of the linear interpolation of the following equations (1) and (2).

Sm = (1−Tx) S _A + TxS _B (1) Sn = (1−Tx) S _C + TxS _D (2) Next, interpolation image data Sm and Sn are obtained in the y-axis direction of the interpolation point S ′ _0. The interpolated image data S ′ ₀ is obtained by performing the linear interpolation calculation of the following equation (3) used.

S ′ ₀ = (1−Ty) Sm + TySn (3) The above calculation can be similarly applied to the other interpolation points S ′ to obtain the respective interpolated image data S ′.

The above-described interpolation method is not always used only when enlarging an image, and even when a high-resolution image is reproduced even when the image is not enlarged or reduced, the image is reproduced in more detail. It can also be applied to the case where it is desired to observe.

The sampling points that contribute to the interpolated image data are not limited to the four points in the vicinity of the interpolation point, but the sampling points in the square mask of 4 points × 4 points including the surrounding points are used. Good.

By the way, in the reproduced visible image, there is an edge portion where the change in the density (luminance) is abrupt, such as a bone portion in a radiation image, and such an edge portion may be enlarged.

However, when such an edge portion extends diagonally with respect to the square lattice (unit lattice) in which the pixels of the original image are arranged (vertically and horizontally with respect to the rhombus lattice in which the pixels of the original image are arranged). The same applies when extending in the direction)
In addition, when the interpolation calculation is performed according to the above-described equations (1) to (3), the enlarged image of the edge portion extending in the diagonal direction (vertical and horizontal directions) becomes prominent with stepwise steps.

For example, in an image having an edge portion extending in an oblique direction as shown in FIG. 8A, microscopically, it is a region of high density points (shown by black circles) shown in FIG. 8B. When interpolation image data is obtained by applying the above-described interpolation calculation to a portion where a boundary line (edge) with a low-density point area (indicated by a white circle) extends in an oblique direction, the interpolation image data S ′ is obtained. _{Since 0} also depends on the low-density original image data S _D , the image data indicates intermediate density whose density is slightly lower than that of the high-density pixels S _A , S _B , and S _C (FIG. 9).
(B)). Therefore, the obtained interpolated image data S '
_When the enlarged image based on ₀ is reproduced, the edge part is shown in FIG.
As shown by the broken line in (B), the image has an enlarged stepped step. That is, in the entire image, the edge portion extending in the oblique direction as shown in FIG. 8 (A) is enlarged by the enlargement process, as shown in FIG. 9 (A). It will be.

Such a stepped portion of the edge portion may be an obstacle to image interpretation when observing the vicinity of the edge portion, and there is a possibility that the diagnostic performance of the image is deteriorated.

Note that this problem is not limited to an image in which sample points are arranged in a square lattice in the vertical and horizontal directions, but an edge portion extending in the vertical and horizontal directions when the sample points are arranged in a rhombus in an oblique direction. The same can occur in the edge portion extending in an oblique direction with respect to the arrangement direction of the unit lattices.

[0017] In addition, the interpolation image data of the interpolation point S _'0,
As described above, the pixels S forming the unit grid including the interpolation points
_{A, S B, S C,} S D image data S _A of (near four points of the interpolation point), S _B, S _C, not only obtained by using only S _D, generally square mask including the interpolation point The same problem as described above occurs when the data is obtained by using the data of the pixels inside.

As an example of a technique for solving this problem, Japanese Unexamined Patent Publication (Kokai) No. 4-268975 discloses a technique which uses three original image data among four sample points constituting a unit lattice. This technique uses four sample points that make up the unit cell
Divide into two sets of sample points that are diagonal to each other, find the difference between the original image data of the two sample points included in each set, and compare the differences obtained for each set, The unit lattice is divided into two triangular areas by a line connecting two sample points belonging to the smaller diagonal, and one triangular area including an interpolation point is selected from the obtained two triangular areas. The interpolation calculation is performed on the basis of the original image data of the three sample points forming the triangular area.

When this action is illustrated in a concrete example, the unit cell is divided into two triangular regions (region 1 and region 2 or region 3 and region 4) by the above comparison as shown in FIG. When an interpolation point exists in the area 1 shown in the figure, based on the original image data of the grid points (sample points) S _A , S _B , and S _C , when an interpolation point exists in the area 2, the grid point S _B , S _C , S _D based on the original image data, if grid points (sample points) S _A , S _C ,
Based on the original image data S _D, if there is an interpolation point in the region 4 lattice points S _A, S _B, on the basis of the original image data S _D, to obtain each interpolated image data S _'0.

[0020] Such interpolation image data S _'0 obtained in the 3 excluding the different sampling points are three points original image data is extremely other among the sample points of the four points constituting the unit lattice
Since it depends on the image data of the sampling points of the points, the data values are not affected by the extremely different sampling points, and therefore the interpolation points of the intermediate density suddenly occur in the high density area and the low density area. It can be prevented from appearing.

[0021]

As described above, according to the technique disclosed in Japanese Patent Laid-Open No. 4-268975, the four sample points forming the unit cell are located at diagonal positions with respect to each other.
The direction in which the edge part extends is specified by dividing the original image data of the two sample points included in each set and comparing the differences obtained for each set. are doing.

However, the extending direction of the edge portion cannot always be specified by the method disclosed in the above-mentioned Japanese Patent Laid-Open No. 4-268975. For example, when a very thin (one pixel width) low-density region extends to the upper right as shown in FIG. 10, if the above method is applied to the unit cell a,
The difference between the original image data of the two sample points included in each of the two sample points that are diagonally opposite to each other is substantially zero, and even if the differences obtained for each group are compared. The diagonal with the smaller difference cannot be specified, and therefore the extension direction of the edge cannot be specified.

The present invention has been made in view of the above circumstances, and provides an interpolation calculation method and apparatus for image data in which the detection accuracy of an edge portion extending obliquely with respect to the arrangement direction of the unit lattice of the original image is improved. The purpose is to do.

[0024]

According to the interpolation calculation method of image data of the present invention, when the interpolation calculation method using three points is applied when the edge portion extends obliquely with respect to the original image, the image is formed. The direction in which the edge portion extends is accurately determined based on not only the original image data of the unit lattice of pixels to be processed but also the original image data around the unit lattice.
Then, the unit lattice is divided into two triangular regions based on the accurate determination result of the extension direction of the edge portion, and the unit lattice is set based on the original image data of the three pixels forming the triangular region. By obtaining the interpolated image data of the interpolated points thus obtained, an image having sharp edges and no step is obtained.

That is, the image data interpolation calculation method according to the present invention is a unit lattice in which a large number of sample points, in which original image data representing an image are defined, are arranged in a lattice pattern at predetermined intervals, and are used as lattice points. 2 consisting of two sample points located diagonally to each other of the unit cell with respect to the array direction
For each set of sample points, the difference between the original image data of each two sample points forming each set of sample points is obtained, and the difference between the image data of each set of sample points is compared to obtain the original image data. The extending direction of the edge portion where the change is abrupt is specified, and the interpolation image data of the interpolation point included in the inside of the unit lattice that the edge portion traverses is defined as at least the edge among the four sample points constituting the unit lattice. In an interpolation calculation method of image data obtained based on original image data of three sample points in the vicinity of the interpolation point including two sample points diagonally located along the extending direction of the part, It is characterized in that the direction in which the edge portion extends is specified based on the original image data of the sample points around the unit grid.

Here, as a method of specifying the extending direction of the edge portion by comparing the differences of the image data of each sample point set, specifically, the samples at the diagonal positions obtained for each set. It is specified that the edge extends along the diagonal direction of the sample point set on the side where the difference between the image data of the points is small. However, since the width of the image data is usually about 8 bits (0 to 255), it cannot be concluded that the edge portion is always present when the difference between the sample point sets is about 1 or 2, or the exact edge portion is not present. Since it is not possible to specify the extension direction of
The difference between the sample point sets is compared with the threshold value corresponding to the data in which the existence of the edge is experimentally confirmed in advance, and the edge extension direction is determined only when the difference is equal to or larger than the threshold value. You can do it like this.

The image data interpolation calculation device of the present invention is a device for implementing the above-described interpolation calculation method of the present invention, in which original image data representing an image, which is arranged in a grid pattern at predetermined intervals, is defined. The two sample points, each of which is composed of two sample points and are diagonally located on the unit cell, with respect to the array direction of the unit cell having a plurality of sample points as grid points. The difference between the original image data of each two sample points forming the point set is calculated, and the difference between the image data of each sample point set is compared to determine the extending direction of the edge portion where the change of the original image data is sharp. Among the four sample points forming the unit lattice which the specified edge extension direction crosses, and the edge portion, the interpolation point including at least two diagonally located sample points along the extending direction of the edge portion. 3 neighborhoods Of image data having interpolation image data extraction means for extracting sample points and interpolation calculation means for obtaining interpolated image data of interpolation points included in the unit lattice based on the extracted three sample points. In the interpolation calculation device, the edge extension direction determination means further specifies an extension direction of the edge portion based on original image data of sample points around a unit lattice having the interpolation point inside. It is what

[0028]

According to the interpolation calculation method and apparatus for image data of the present invention, the interpolation calculation method using three points is applied when the edge portion of the image extends obliquely with respect to the original image. For the four sample points that form the unit cell including the inside, the difference in the image data between the two sample points diagonally located with respect to each other is obtained, and the difference is compared between the sample point pairs to determine the edge The extension direction is basically determined, but if the difference in the image data between the two sets of sample points is substantially equal, the extension direction of the edge cannot be determined from the magnitude relationship of the differences, so the interpolation point is The extension direction of the edge can be reliably determined by comparing the difference in the image data of the diagonal sample points similar to the above with respect to the unit lattice including the image data further surrounding the unit lattice to which it belongs.

[0029]

BEST MODE FOR CARRYING OUT THE INVENTION A specific mode for carrying out the image data interpolation calculation method of the present invention will be described below.

FIG. 1 is a schematic block diagram showing an image reproduction system including a specific interpolation calculation device 30 for implementing the image data interpolation calculation method of the present invention. The image reproduction system shown in the figure is adapted to the storage means 10 for storing the original image data representing the image, the input means 21 for inputting a desired reproduction format, and the desired reproduction format inputted from the input means 21, A multi-formatter 20 that performs predetermined signal processing on the image data (hereinafter referred to as primary image data or original image data) Sorg stored in the storage means 10, and an image that has been subjected to predetermined signal processing by the multi-formatter 20. A reproducing means 40 such as a CRT or a printer for reproducing the visible image in the desired reproduction format based on the data (hereinafter, referred to as secondary image data or interpolation image data) S'is configured.

The multi-formatter 20, for example, divides one film into four small areas which are different from each other, and prints a reduced image of four different images in each area, one format on one film. The primary image data Sorg is adapted so as to be compatible with various formats for reproducing an image, such as a format for printing a large image as it is or a format for enlarging a part of the image and printing the enlarged part on the film. The signal processing is performed, and particularly when the image is enlarged or reduced, the interpolation calculation device 30 of the present invention for calculating secondary image data having a different number of data from the primary image data Sorg is included. .

Here, the primary image data Sorg (= {Sij; i, j = used in the image reproducing system of the present embodiment.
2, 1, ...}) is image data representing an image in which pixels (lattice points of white circles and black circles) are arrayed in the vertical and horizontal directions as shown in FIG. There is an edge portion where the density (the value of the primary image data) changes sharply in an oblique direction that rises to the right with respect to the lined direction. In the figure, those with high density are represented by black circles, and those with low density are represented by white circles.

The interpolation calculation device 30 uses the primary image data Sor
Edge extension direction determination means for detecting an edge extending in an oblique direction based on g and determining the extension direction
31, interpolation image data extraction means 32 for determining the three primary image data Sorg which are the basis for calculating the secondary image data S ′ of the interpolation point, and the unit grid of the unit grid based on the extracted three sample points. And an interpolating calculation means 33 for obtaining interpolated image data S ′ of an interpolating point included inside.

Here, the edge extension direction judging means 31 stores a judgment algorithm which will be described later and performs a predetermined judgment operation according to the algorithm.

Further, the interpolation image data extraction means 32 includes the neighborhood 3 of the interpolation point including the primary image data of two points which are adjacent to each other in the extending direction of the edge portion obtained by the determination means 31.
The primary image data of the points is extracted, and the triangular plane including the interpolation points is determined.

Further, the interpolation calculation means 33 stores a predetermined interpolation calculation algorithm which will be described later, and based on the input primary image data {Sij}, the secondary image data is obtained by the interpolation calculation according to this algorithm. Calculate S '.

Next, the operation of the image reproducing system of this embodiment will be described.

First, a desired enlargement ratio is input to the input means 21, and this enlargement ratio is input from the input means 21 to the multi-formatter 20.

The multi-formatter 20 stores the primary image data Sorg shown in FIG.
(= {Sij}) is read.

Further, the multi-formatter 20 inputs the read primary image data Sorg to the interpolation calculation device 30 in order to obtain secondary image data representing an enlarged image corresponding to the input enlargement magnification.

The primary image data Sorg input to the interpolation calculation device 30 is first input to the determination means 31. Judgment means 31
Is the image data Sij, S of each grid point of the smallest square grid (unit grid) of the input primary image data Sij.
The difference of the primary image data between the lattice points of (i + 1) j, Si (j + 1), and S (i + 1) (j + 1) at diagonal positions is calculated. That is, each calculated value | Sij-S (i + 1) (j + 1) |, | S (i + 1) j-S
i (j + 1) | Then these two calculated values | Sij
-S (i + 1) (j + 1) |, | S (i + 1) j-Si (j + 1) | are compared,
Of these, the one with the larger calculated value means that the edge part where the density change is sharp crosses between the two grid points for which the difference is obtained, and conversely, the one with the smaller calculated value means the density. It means a flat part with little change. Therefore, the determination means 31 determines that the edge portion exists along the direction orthogonal to the diagonal direction with the larger calculated value, that is, along the diagonal direction with the smaller calculated value.

Specifically, in the case shown in FIG. 2A, the value of | S (i + 1) j-Si (j + 1) | is substantially zero, and |
Since the value of Sij-S (i + 1) (j + 1) | shows a relatively large value, the grid points of the primary image data S (i + 1) j and the grid points of Si (j + 1) are It is determined that the edge portion extends along the connecting direction, that is, in the diagonally upward right direction.

Now, for example, the operation in the case where the primary image data Sij of the sample points forming a certain unit lattice a is as shown in FIG. 3A will be described.

When the same algorithm as above is applied to the unit grid of the primary image data Sij appearing in such an array, grid points Sij, S (i + 1) (j +) along one diagonal direction are applied.
1) difference of primary image data | Sij-S (i + 1) (j + 1) | and the other diagonal grid points S (i + 1) j, Si (j + 1) 1
The difference | S (i + 1) j-Si (j + 1) | of the next image data is substantially zero, and the comparison of these values shows that the magnitude relationship between the two is substantially the same and the edge portion extends. I cannot identify the direction in which I am in That is, as shown in FIG. 3 (B), there may be an edge in which image data having low density rises and rises to the right, or as shown in FIG. 3 (C), image data having high density rises to the left. It is not possible to determine whether there is an edge that rises to the left and continues.

Therefore, the judging means 31 sets the unit cells b, c, d and e adjacent to the unit cell a as shown in FIG.
Similar to the above-described operation for each of the unit lattices b, c, d, and e, the difference in the image data between the lattice points at the diagonal positions is obtained, and the difference is compared between the diagonal points to determine the direction in which the edge extends. Specify. With respect to the one shown in FIG. 4, it is possible to specify the presence of an edge portion that is rising upward to the right from any of the unit cells b, c, d, and e.

Next, the operation of the interpolating image data extracting means 32 for extracting the three sample points which are the basis for calculating the secondary image data S'of the interpolation point will be described.

The interpolation image data extraction means 32 determines the secondary image data necessary for reproduction in accordance with the enlargement ratio of the image input from the input means 21 and the size of the output medium used for reproduction by the reproduction means 40. The number of data is determined, and interpolation points S'ij for the grid points representing the primary image data (FIG. 2 (B))
Is determined.

Here, the interpolation image data extraction means 32 is a grid point of three points near the interpolation point S'ij, and extends in the direction in which the edge determined by the determination means 31 extends (oblique direction rising to the right). A grid point including two grid points adjacent to each other is extracted. That is, regarding the interpolation point S′ij shown in FIG. 2B, three neighboring points including two grid points S (i + 1) j and Si (j + 1) which are adjacent to each other along the direction in which the edge extends. Grid points Sij of
S (i + 1) j and Si (j + 1) are extracted. Generally, as shown in FIG. 5, a primary image corresponding to three lattice points corresponding to the vertices of any one of the triangles 1 to 4 depending on the direction in which the edge extends and the position of the interpolation point. A set of data will be extracted. For example, when the interpolation points are present inside the triangle 1 whose edge extends in an upward-sloping diagonal direction, lattice points S _A (= Sij), S _B (= S (i + 1) j), S _C
If (= Si (j + 1)) is extracted and interpolation points are present inside the triangle 3 whose edge extends obliquely upward to the left, grid points S _A (= Sij) and S _C (= Si ( j + 1)), S _D
(= S (i + 1) (j + 1)) is extracted.

The extracted grid points Sij, S (i + 1) j, Si (j +)
The primary image data Sij, S (i + 1) j, Si (j + 1) of 1) are input to the interpolation calculation means 33, and the interpolation calculation means 33 inputs the three primary image data Sij, S ( The secondary image data S'ij of the interpolation point S'ij is calculated based on i + 1) j and Si (j + 1).

The algorithm for calculating the secondary image data S'ij will be specifically described with reference to FIG. 2B as an example.

Here, the grid points Sij to S (i of the primary image data are
+1) j and Sij to Si (j + 1) have unit length 1, and the distance of the interpolation point S'ij from the grid point Sij in the x-axis direction (horizontal direction) is Tx, and from the grid point Sij. Of the virtual interpolation point Sm corresponding to the position of the interpolation point S'ij in the x-axis direction, and the grid point S of the interpolation point S'ij. The image data Sn of the virtual interpolation point Sn corresponding to the position in the direction connecting the (i + 1) jx and the Si (j + 1) (the direction in which the edge extends) is expressed by the following equations (4) and (5). Calculated by linear interpolation.

Sm = (1-Tx) Sij + TxS (i + 1) j (4) Sn = (1-Tx) Si (j + 1) + TxS (i + 1) j (5) Next, the interpolation point S'ij In the y-axis direction of, the secondary image data S′ij is obtained by performing the linear interpolation operation of the following equation (6) using the image data Sm and Sn.

S′ij = (1−Tx−Ty) Sij + TySi (j + 1) + TxS (i + 1) j (6) Secondary image data of the interpolation point S′ij obtained according to the equation (6). Since S′ij is not affected by the grid points S (i + 1) (j + 1) whose density values are extremely different in the direction crossing the edge, this interpolation point exists on the edge of the image. Also,
The secondary image data at the interpolation point depends only on the primary image data Si (j + 1), S (i + 1) j and Sij in the edge extending direction. Therefore, interpolation points having extremely different data values do not suddenly appear on the edge.

The calculation formula of the secondary image data S'ij shown in the formula (6) is for the case where the interpolation point exists inside the triangle 1 shown in FIG. Even when there are interpolation points inside, the secondary image data can be calculated by the same procedure as described above.

That is, when an interpolation point exists inside the triangle 2, S'ij = (1-Tx) Si (j + 1) + (1-Ty) S (i + 1) j + (Tx + Ty- 1) S (i + 1) (j + 1) (7) When an interpolation point exists inside the triangle 3, S'ij = (1-Ty) Sij + (Ty-Tx) Si (j + 1) + TxS (i + 1) (j + 1) (8) When an interpolation point exists inside the triangle 4, S'ij = (1-Tx) Sij + (Tx-Ty) S (i + 1) j + TyS ( The secondary image data S′ij at the interpolation point can be calculated by applying i + 1) (j + 1) (9), respectively.

The interpolation calculation means 33 includes the above equations (6) to
(9) is stored, and the above formula is calculated according to the edge extension direction input from the determination unit 31, the position of the interpolation point input from the image data extraction unit 32, and the primary image data of the grid point. The calculation formula to be applied is selected from (6) to (9), and the secondary image data S′ij at the interpolation point is calculated according to the selected calculation formula.

The secondary image data S'of all the interpolation points thus obtained are output to the reproducing means 40.

When the primary image data of the four pixels forming the unit grid is arranged as shown in FIG. 3 described above, the interpolation point belongs to the direction in which the edge extends. The triangle corresponds to any one of the four types of triangles 1 to 4 shown in FIG. 11, and in any of these cases, the calculation may be performed in the same manner as in the case of belonging to the triangle 2 or 3 shown in FIG. .

The reproducing means 40 reproduces an image based on the input secondary image data S'as a visible image. As described above, in the reproduced visible image, the secondary image data value in the edge extending direction is not extremely different from the original primary image data value, and therefore the diagonally extending edge portion is also smooth. And, the image becomes sharp.

The interpolation calculation device 30 used in the image reproduction system of the present embodiment has been described using the primary image data stored in advance in the storage means 10. However, the interpolation calculation device of the present invention has this form. The present invention is not limited to this, and may be a form in which image data representing an image obtained by reading with an image reading device is directly received.

In the system of this embodiment, a linear interpolation coefficient (interpolation coefficient f
_{Although 0} (t) = 1-t and f ₁ (t) = t) have been described, the present invention is not limited to linear interpolation.
Interpolation calculation in which the interpolation coefficient is expressed by a polynomial of third order or higher,
The interpolation calculation represented by a function may be used.

An example of using these interpolation calculation formulas will be described below.

In the linear interpolation, the graph of the interpolation coefficient is as shown in FIG. 6 (A), but the graph of the interpolation coefficient represented by a polynomial of third or higher degree is as shown in FIG. 6 (B), for example. There is.

Hereinafter, the interpolation coefficient shown in FIG.
The case of the interpolation calculation formula using the following polynomials will be described. The interpolation coefficients f ₀ (t) and f ₁ (t) are f ₀ (0)
= 1, f ₀ (1) = 0, f ₁ (0) = 0, f ₁ (1) =
1 and f ₀ (t) + f ₁ (t) = 1 (where 0 ≦ t
The third order may be satisfied, for example, f ₀ (t) = 2t ³ −3t ² + 1f ₁ (t) = − 2t ³ + 3t ² as long as the condition of ≦ 1) is satisfied. The fifth order may be, for example, f ₀ (t) = − 6t ⁵ + 15t ⁴ −10t ³ + 1f ₁ (t) = 6t ⁵ −15t ⁴ + 10t ³ .

First, the case where the interpolation point belongs to the inside of the triangle 1 in FIG. 11 will be described. In this case, the direction in which the edge extends is the direction in which the straight line connecting the grid points B and C extends.

As shown in FIG. 12A, the desired interpolation point P
, A straight line parallel to the direction in which the edges extend (the direction in which the straight line connecting the grid points B and C extends) is obtained, and the point where this straight line intersects the line connecting the grid points A and B in the original image is virtually interpolated. A point α and a point that intersects with the line connecting the grid points A and C are virtual interpolation points β
Let the secondary image data of the virtual interpolation point α be the grid point A,
The secondary image data of the virtual interpolation point β is obtained by a high-order interpolation calculation formula based on the image data of the grid points arranged in the direction connecting B, and the image data of the grid points arranged in the direction connecting the grid points A and C is obtained. The secondary image data of the interpolation point P on the straight line connecting the obtained virtual interpolation points α and β is obtained by a high-order interpolation calculation formula based on Obtained by linear interpolation based on the image data.

That is, the image data α and β of the virtual interpolation points α and β are expressed by the following equations (10) and (11). Here, for simplification of the equation, it is assumed that the lattice point names and the symbols A, B, C, D representing the image data thereof are used as they are.

Α = f ₀ (Tx + Ty) · A + f ₁ (Tx + Ty) · B (10) β = f ₀ (Tx + Ty) · A + f ₁ (Tx + Ty) · C (11) Therefore, the secondary image data P of the interpolation point P Can be obtained by P = (Tx.alpha. + Ty.beta.) / (Tx + Ty) (12).

The original image data and the secondary image data P are shown in FIG. 12B when expressed in a three-dimensional space.

The calculation formula of the secondary image data P shown in the formula (12) is for the case where the interpolation point exists inside the triangle 1 shown in FIG. Even when the interpolation point exists inside 11), the secondary image data can be calculated by the same procedure as the above-mentioned procedure.

That is, when the interpolation point P exists inside the triangle 2 in FIG. 11 as shown in FIG. 13A, the interpolation point P
The secondary image data P of can be obtained by P = {(1-Ty) .alpha. + (1-Tx) .beta.} / (2-Tx-Ty) (13).

Next, the case where the interpolation point belongs to the inside of the triangle 3 in FIG. 11 will be described. In this case, the direction in which the edge extends is the direction in which the straight line connecting the grid points A and D extends.

As shown in FIG. 13B, the desired interpolation point P
, A straight line parallel to the direction in which the edges extend (the direction in which the straight line connecting the grid points A and D extends) is obtained, and the point where this straight line intersects the line connecting the grid points A and C in the original image is virtually interpolated. A point that intersects with the line connecting the point α and the grid points C and D is a virtual interpolation point β.
Let the secondary image data of the virtual interpolation point α be the grid point A,
The secondary image data of the virtual interpolation point β is obtained by a high-order interpolation calculation formula based on the image data of the grid points arranged in the direction connecting C, and the image data of the grid points arranged in the direction connecting the grid points C and D is obtained. The secondary image data of the interpolation point P on the straight line connecting the obtained virtual interpolation points α and β is obtained by a high-order interpolation calculation formula based on Obtained by linear interpolation based on the image data.

That is, when the interpolation point P exists inside the triangle 3 in FIG. 11 as shown in FIG. 13B, the interpolation point P
The secondary image data P of can be obtained by P = {(1-Ty) .alpha. + Tx.beta.} / (1 + Tx-Ty) (14).

Further, when the interpolation point P exists inside the triangle 4 of FIG. 11 as shown in FIG.
The next image data P can be obtained by P = {(1-Tx) .alpha. + Ty.beta.} / (1-Tx + Ty) (15).

The interpolation calculation using a high-order polynomial is
Good interpolation is possible for edges extending diagonally to the grid points, but not suitable for edges extending vertically and horizontally.

Therefore, it is appropriate to obtain the secondary image data for the edges extending in the vertical and horizontal directions as follows.

As for the vertical edge as shown in FIG. 14A, it first passes through the desired interpolation point P and is parallel to the extending direction of the edge (the extending direction of the straight line connecting the grid points A and C). A straight line is obtained, and a point at which this straight line intersects the line connecting the grid points A and B of the original image is a virtual interpolation point α, and a point at which the straight line intersects the line connecting the grid points B and C is a virtual interpolation point β, and virtual interpolation is performed. The secondary image data of the point α is obtained by a high-order interpolation calculation formula based on the image data of the lattice points arranged in the direction connecting the lattice points A and B, and the secondary image data of the virtual interpolation point β is calculated as the lattice point. B, C
The secondary image data of the interpolation point P on the straight line connecting the virtual interpolation points α and β obtained by the high-order interpolation calculation formula based on the image data of the grid points arranged in the direction connecting It is obtained by linear interpolation based on the secondary image data of the virtual interpolation points α and β.

That is, the secondary image data P of the interpolation point P
Can be obtained by P = {(1-Tx-Ty) .alpha. + Ty.beta.} / (1-Tx) (16). When the secondary image data P is expressed in a three-dimensional space, it becomes as shown in FIG.

The secondary image data P of the interpolation point P is similarly processed in the case of having a lateral edge as shown in FIG.
Can be calculated by P = {(1-Tx-Ty) .alpha. + Tx.beta.} / (1-Ty) (17). When the secondary image data P is expressed in a three-dimensional space, it becomes as shown in FIG.

In the case of such an edge in the vertical direction or the horizontal direction, the image data of the interpolation point is not limited to the case where the image data of the interpolation point is obtained based on the primary image data of the three points of the grid point. It may be obtained based on the four primary image data.

For example, in the case of an edge extending in the vertical direction, FIG.
As shown in, a straight line that passes through the desired interpolation point P and is parallel to the direction in which the edge extends (the direction in which the straight line connecting the grid points A and C extends) is obtained, and this straight line is the grid points A and B of the original image. The point that intersects the line connecting the lines is a virtual interpolation point α, the point that intersects the line connecting the grid points C and D is a virtual interpolation point β, and the secondary image data of the virtual interpolation point α is the grid points A and B. The secondary image data of the virtual interpolation point β is obtained by a high-order interpolation arithmetic expression based on the image data of the grid points arranged in the connecting direction, and the grid point C, D
The secondary image data of the interpolation point P on the straight line connecting the virtual interpolation points α and β obtained by the high-order interpolation calculation formula based on the image data of the grid points arranged in the direction connecting It is obtained by linear interpolation based on the secondary image data of the virtual interpolation points α and β.

In addition to the above-described interpolation operation in which the interpolation coefficient is linear and the interpolation operation in which the interpolation coefficient is expressed by a polynomial of degree 3 or higher, for example, interpolation coefficients f ₀ (t) and f ₁ (t )
May use the interpolation calculation represented by the function shown in the graph of FIG.

[Brief description of drawings]

FIG. 1 is a schematic block diagram showing an image reproduction system including a specific interpolation calculation device 30 for carrying out an image data interpolation calculation method of the present invention.

FIG. 2 (A) Pixels (sample points) forming original image data
And (B) a diagram showing the operation of obtaining interpolated image data.

FIG. 3 is a diagram showing a unit cell in which diagonal edges having substantially equal diagonal image data exist.

FIG. 4 is an explanatory diagram illustrating an operation of an interpolation calculation device.

FIG. 5 is a diagram showing the types of triangular areas (No. 1).

FIG. 6 is a graph showing interpolation coefficients

FIG. 7 is a conceptual diagram showing pixels forming original image data and pixels forming interpolation image data.

FIG. 8 is a diagram showing an original image and pixels forming the original image data.

FIG. 9 is a diagram showing an interpolated image and pixels forming the interpolated image data.

FIG. 10 is a view showing a unit cell in which diagonal edges having substantially equal diagonal image data exist.

FIG. 11 is a diagram showing the types of triangular areas (No. 2).

12A is a diagram showing an operation of obtaining interpolated image data (triangle 1), and FIG. 12B is a diagram showing original image data and interpolated image data in a three-dimensional space.

13A is a diagram showing an operation for obtaining interpolation image data (triangle 2), FIG. 13B is a diagram showing an operation for obtaining interpolation image data (triangle 3), and FIG. 13C is a diagram showing an operation for obtaining interpolation image data. (Triangle 4)

14A is a diagram showing an operation of obtaining interpolated image data (vertical edge), and FIG. 14B is a diagram showing original image data and interpolated image data in a three-dimensional space.

15A is a diagram showing an operation of obtaining interpolated image data (horizontal edge), and FIG. 15B is a diagram showing original image data and interpolated image data in a three-dimensional space.

FIG. 16 shows three sets of four original image data and interpolated image data.
Diagram in dimensional space

[Explanation of symbols]

 10 storage means 20 multi-formatter 21 input means 30 interpolation calculation device 31 edge extension direction determination means 32 interpolation image data extraction means 33 interpolation calculation means 40 reproduction means Sorg primary image data (original image data) S'secondary image data ( Interpolated image data)

Claims (2)

[Claims] 1. The unit grids are arranged in a grid pattern at a predetermined interval, and each of the unit grids is arranged with respect to the array direction of the unit grids, each grid point being a plurality of sample points in which original image data representing an image is defined. For two sets of sample points, each set of two sample points, located diagonally, the difference between the original image data of each two sample points forming each sample point set is calculated, and the image of each sample point set is obtained. The direction in which the edge portion where the change of the original image data is steep extends is specified by comparing the difference between the data, and the interpolated image data of the interpolation points included in the unit grid traversed by the edge portion is set to the unit. An image obtained based on original image data of three sample points in the vicinity of the interpolation point including at least two sample points located diagonally along the extending direction of the edge part among the four sample points forming the lattice Data interpolation calculation method In the above method, the interpolation calculation method of image data is characterized in that the extending direction of the edge portion is specified based on the original image data of the sample points around the unit lattice having the interpolation points inside. 2. The unit grids are arranged in a grid pattern at a predetermined interval, and each of the unit grids is arranged with respect to the array direction of the unit grids, each grid point being a plurality of sample points in which original image data representing an image is defined. For two sets of sample points, each set of two sample points, located diagonally, the difference between the original image data of each two sample points forming each sample point set is calculated, and the image of each sample point set is obtained. Of the four sample points constituting the unit lattice crossed by the edge extension direction determining means for identifying the extension direction of the edge portion where the change of the original image data is sharp by comparing the difference between the data, Interpolation image data extraction means for extracting three sample points in the vicinity of the interpolation point including at least two sample points located diagonally along the direction in which the edge portion extends, and the extracted three sample points. Based on the unit case In an image data interpolation calculation device having interpolation calculation means for obtaining interpolation image data of interpolation points included in a child, the edge extension direction determination means further includes: An interpolation calculation device for image data, characterized in that a direction in which the edge portion extends is specified based on original image data of a sample point.