JPS6059474A - Picture converter - Google Patents

Abstract

PURPOSE:To avoid the intermittence of picture elements contained in an output picture with a device which performs a geometric conversion of pictues by means of a picture memory, by covering an input picture so that the adjacent minute areas have a part common with each other and approximating the conversion function with each minute area. CONSTITUTION:A program corresponding to the type of picture conversion is transferred to a processor 1 from a memory 3 by a command given from an input/output device 2. Then the parameter of the program is converted by the device 2 and transferred to a processor 4. The entire input picture is covered with a minute area so that the adjacent minute areas have a part common with each other. Then the conversion is given to the key point of each area, and an operation is carried out to obtain a position after conversion. This obtained position is stored to a buffer memory 5. The conversion information on each key point is used to obtain the adverse linear conversion made approximate by hardwares 6 and 7 for exclusive use. It is shown that a certain area A on an input picture IM1 is changed to a specific area B on an input picture IM2. Then the interpolation of a picture given from a memory 8 is performed to each picture given from a memory 8 is performed to each picture element within a processing area by means of an obtained address.

Description

DETAILED DESCRIPTION OF THE INVENTION [-Field of Industrial Application] The present invention relates to an image conversion device that realizes geometric deformation of an image using an image memory.

[Background Art and Problems Therewith] The applicant of the present application has previously proposed an image conversion processing device that performs the following processing.

(2) Divide the entire original image into minute regions (several pixels x several pixels), apply a predetermined transformation to the representative point of each region, and find the position after the transformation.

■ Linear approximation of transformations (including nonlinear ones) near representative points.

■ Calculate the inverse transformation of the approximated linear transformation (also a 1-dimensional linear transformation).

■ Apply this approximated inverse transformation to all pixels near the representative point after transformation.

This image conversion device will be described in further detail.

As shown in Figure 1A, each sample point of the input image (black circle “・
”) on the orthogonal orthogonal ordinal marker,
An area So of (mXn) pixels including other sample points is determined around the representative point (XI I Each sample point of the memory (indicated by a white circle "・J")
are arranged Y, Y,, on the complex jj1ζ mark,
A predetermined area S is defined. This area S is a predetermined area S.

The position (y+ +
3'z).

Thereafter, each of the sample points of the output image memory included in this area S is selectively associated with the sample points of the input image included in the area So.

The conversion formula is Yl-ψ' (XI, X2)...○Y2 =
cp2(XI + X2) --@, and XI
When the area So is defined in the shape of (m X n ) pixels at the X2 orthogonal coordinates 2 as shown in FIG. 2A, YI
The area S defined on the Y2 orthogonal coordinate is located at the position (y+, V2) ('/+='p' (X
I +X2) + y2-ψ2(xl + X2)
), the four numbers all + nu + revision 1
It becomes a parallelogram shadow area defined by and -.

4 numbers all + α12 + α21 and α22
is determined using the partial differential value or difference value at the sample point (XI, X2) of the input image for each of the conversion equations shown in equations (1) and 0 above. When using partial differential values, it is determined, and when using difference values, ■ a+i=[ψ' (xl+m, -9'' of x (XI +
X2 ) ) X XαI+ −0a+z−cψ'
(x+ , xz+n) −ψ' (x+ I X2
) :]×±X(2+2−■az+=Cψ2(xl−1
-m, X2) -cp2(XI, X2)]x
It is determined as y:a2v.

Here, Hei! , α12+α21. α22 is a coefficient for correcting the size of the parallelogram shadow area, α11. α1□, α2
1. α2°≧1. In other words, XlX2 rectangular coordinates"-
The area S defined by In order to eliminate gaps between the parallelogram shadow areas S defined on the YI Y2 orthogonal coordinates, the parallelogram shadow areas S are expanded using correction coefficients to eliminate gaps between the parallelogram shadow areas S. It is something.

In addition, the factors that cause the intermittency include all the areas S,
There are approximations using aI2°azl and azz and calculation errors.

In this way, the area S defined on the XlX2 orthogonal coordinates
. Corresponding to the rectangular coordinates y, y2, a parallelogram shadow region S is defined, and the region S. Among the sample points in the output image memory, sample points included in the parallelogram shadow region S are selected as those corresponding to the sample points of the human image included in the image memory.

However, each of the sample points of the output image memory included in this parallelogram shadow area S is the area S. It is unknown which point of the input image contained within the corresponding point corresponds. Therefore, next, a predetermined area S corresponding to each of the sample points of the output image memory included in the parallelogram shadow area S. The image of human power contained within is highlighted. That is, area S. It is determined to which sample point in the output image memory each of the sample points of the human image included in the image should correspond.

Now, area S. The point (ylo, y2o) in the parallelogram shadow region S corresponding to the point (XIO, Xzo) in can be expressed by linear approximation as follows. In this equation, in this example, from this, each of the sample points of the output image memory included in the parallelogram shadow area S corresponds to which point of the input image included in the area S by χ In order to decide what to do, the inverse function of the first-order approximation formula shown in the above-mentioned equation is used. The inverse function of this linear approximation is the inverse matrix.

Here, among the sample points of the output image memory included in the parallelogram area 100, sample points (y++, yz+)
Corresponding to this, the point (xII+X21) is the area S. For example, the point (Xll, X
21) The region S closest to. One number point (XI
+'+X21') is determined. In this case, for example, the point (
The image information corresponding to the point (Xll l X21) may be interpolated from the four sample information of the human image surrounding x++, X2+). Furthermore, other samples of the output image memory included in the parallelogram shadow area S correspond to the area S in the same manner. The points of the IrinoJ image contained within are determined.

In this way, a predetermined area S is defined on the input image. Sample points of the output image memory are determined for each point of the input image included in the sentence JL.

The above-mentioned area SO is input image '-: +9 and Iy
6ζ is taken, and there is no image conversion process for the entire input image 5
Ru.

” In two series of image conversion devices, adjacent areas SA after conversion
And the sieve is as shown in FIG. Due to the linear approximation, the boundaries of areas SA and SB do not coincide with each other, and there is a shift. Therefore, in order to prevent image intermittency, a correction coefficient al larger than 1 is set as described above.
l+αl 2 + a2 1 1 a2□ is used. If the transformation is performed using a first-order slow number without correction, the area after the transformation will be reduced as shown by SA and 100B' in the 3rd 1EI IZ. Therefore, an interruption occurs at a summing point P that is not included in either of the two.

As described above, although the previously proposed image conversion device can prevent the occurrence of pixel discontinuity, it has the following problems.

First, by multiplying by the correction coefficient, the linear approximation function becomes far different from the original conversion function. Second,
When linear conversion is performed, since first-order approximation is not performed, there is no need to multiply by a correction coefficient, and therefore, processing cannot be unified between linear conversion and nonlinear conversion.

``Object of the Invention'' Therefore, the present invention provides an image conversion device that can prevent pixel intermittency in an output image without performing an operation of multiplying by a correction coefficient when giving coefficients of a linear approximation function. The purpose is to

“Summary of the Invention” The present invention is an image conversion device that performs conversion using a conversion function approximated for each minute region. This invention divides an input image into minute regions so that adjacent minute regions have common parts. A conversion function is approximated for each minute area, and the minute area is transformed using the approximated conversion function to define the corresponding minute area in the output image. The approximated transformation function is inversely transformed for each pixel, and the input image information at the position after the inverse transformation is assigned as an output pixel.

``Example'' An example of the present invention will be explained in detail with reference to the drawings.

FIG. 4 shows the configuration of an embodiment of the present invention, in which reference numeral 1 indicates a first processor. In association with this processor 1, an input/output device 2 and a large capacity memory 3 are provided. Large-capacity memory 3
A program corresponding to the type of image conversion is transferred from the processor 1 to the high speed memory of the processor 1, and parameters of the program are changed using the input/output device 2. This program is then transferred to the second processor 4.

In this case, it is also possible to use a method in which the second processor 4 carries out the output of the large capacity memory 3 and the block change, and the first processor 1 carries out only the transfer of the ζ parameters.

The second processor 4 executes the programmed program. The entire input image is covered with minute areas so that adjacent images have common parts, the representative points of each area are transformed, and the positions after the transformation are determined by calculations and other necessary operations. information is created and stored in the buffer memory 5.

Reference numeral 6 indicates first dedicated hardware, which reads out information regarding the conversion for each representative point performed by the processor 4 from the buffer memory 5, and uses this information to read out information regarding the conversion performed by the processor 4 for each representative point. The area to be processed by the software 7 is specified.

That is, as shown in FIG. 5, the dedicated hardware 1 is informed whether a certain area A on the input image IM changes to an area B such as on the output image IMZ. The process performed by the first dedicated hardware 6 is a process of performing inverse transformation of the approximated linear transformation.

The second dedicated hardware 7 is supplied with input image data read from the image memory 8, and the output data of the dedicated hardware γ is sent to the image memory 9.
be included. The dedicated hardware 7 stores read addresses for all pixels within the processing range. First, the dedicated hardware 7 approximates all pixels in the vicinity of the representative point after conversion! This method performs an inverse transformation. Then, based on the requested read address, image data is read from the image memory 8, and after processing such as interpolation, it is stored in the image memory 9.

In the embodiment of the present invention described above, a case will be described in which an output image IM2 is formed from an input image IM+, as shown in FIG. In FIG. 6, broken lines drawn in the input image IMI indicate minute regions of equal size, for example, minute regions of (8×8) pixels. Of course, the area divided by this broken line is actually, for example, (64X9
6) Although there are a large number of them, they are shown in a simplified manner in FIG. 6 d1. In one embodiment of the present invention, adjacent minute regions serving as units of image conversion processing are arranged to have common parts.

In Fig. 6, 1, (two microregions 1.
We are focusing on SΔ and sB, and the area SA&;
I1, area SB The area SB is enlarged by, for example, l-rifsol from the position indicated by the broken line on the O side, and conversely, area SB is similarly enlarged by one sample from the position indicated by the broken line on the area sA side.

Therefore, both have a common part with a width of two samples. Although not shown in the drawings, the minute area is defined so as to have a common portion with adjacent areas on the other side in the lateral direction and on each of the upper and lower sides, as described above. These areas SA and sB are each transformed by a linear approximation function without correction, and the seventh
As shown in the enlarged figure, the area in image IM2 becomes rice. This conversion is performed in the same manner as described above, except that it is not multiplied by a correction coefficient.

As can be understood from Fig. 7, even if some errors occur due to the approximation of the conversion function, the converted image can be converted by converting each minute region where adjacent objects have common parts. There will be no interruptions.

Note that when approximating the conversion function, the approximation may be performed using a polygonal area other than the parallelogram shadow area.

1. Effects of the Invention According to the present invention, when providing an approximation function, even if the difference value or partial differential value between the representative point of the minute area of the input image and the conversion point of the output image is used as its coefficient as the coefficient, the output It is possible to prevent pixel gaps from occurring in an image. Therefore, it is possible to prevent the approximation function from becoming far different from the original conversion function, as when multiplying by a correction coefficient, and
This eliminates the problem that processing methods cannot be unified when performing linear transformation that does not require multiplication of complementary iF coefficients.

[Brief explanation of the drawing]

1, 2, and 3 are route maps used to explain an image conversion device to which the present invention can be applied, FIG. 4 is a block diagram of the configuration of an embodiment of the present invention, and FIG. FIGS. 6 and 7 are route maps used to explain one embodiment of the present invention. 1M, ......Manual image, 1M2 ・Output image, So. SA, SB・・・・・・Minute area of input image, S, S
A, S13 - Minute area of output image. Agent Sugi i'ili IE Knowledge Figure 1 Figure 3 Figure 2 Figure 4

Claims (1)

[Scope of the Claims] An image conversion device that performs geometrical conversion of an input image using a predetermined conversion function. The input image is covered with the above-mentioned microregions so that adjacent microregions have common parts, a transformation function is approximated for each of the microregions, and the transformation of the microregions described in 2 is performed using the approximated sawtooth transformation function. Responding to whatever happens 1.
Define a minute area in the output image, perform inverse transformation of the approximated transformation function for each pixel included in the minute area in the output image, and convert the input image information at the position after this inverse transformation to the output pixel. An image conversion device that is assigned as an image conversion device.