JPS62279477A - Image data rotating system - Google Patents

Abstract

PURPOSE:To decrease a distortion generated in an image, and to improve a recognition rate by generating an interpolating black picture element in an intermediate position, when continuous black picture elements exist in one direction, and thereafter, converting a real number coordinate value which has been obtained by a rotation processing, to an integer coordinate value. CONSTITUTION:An inclination angle detecting part 15 segments an image data of a graphic to be recognized, from an image memory 14, and detects an inclination angle theta of its graphic. On the other hand, when continuous black picture elements P and Q exist in one direction of vertical, horizontal and diagonal directions on a main lattice of an image, an interpolating picture element generating part 11 generates an interpolating black picture element S in its intermediate position. A rotation arithmetic part 12 is rotated in the reverse direction by magnitude of the inclination angle thetawhich is detected by the detecting part 11, namely, by -theta against each black picture element and the interpolating picture element for constituting an image on the main lattice. A rounding arithmetic part 13 rounds a coordinate value of each black picture element and the interpolating picture element, which is derived by the arithmetic part 12 and executes a processing for converting a real number value to an integer value. In such a way, even after the rotation processing, as for the picture element P and Q, the adjacent relation is held and no interruption is generated.

Description

[Detailed Description of the Invention] 3. Detailed Description of the Invention [Summary] In an image data rotation method that generates a rotated image by rotating an image by a predetermined angle, the image data is rotated vertically on the main grid of the image.

When there are continuous black pixels in either the horizontal or diagonal direction, an interpolated black pixel is generated at the intermediate position, and then the rotation process and the real number coordinate values obtained by this rotation process are converted to integer coordinate values. Perform the processing to do. This eliminates the occurrence of discontinuities, irregularities, holes in black pixels, etc. in the image after rotation processing, and improves the figure recognition rate.

[Industrial application field]

The present invention relates to an image data rotation method for rotating a tilted figure to the correct position when recognizing a figure, such as character recognition or drawing reading.

When recognizing figures such as characters and symbols, a pattern matching method is often used in which an image of the figure to be recognized is matched with a standard pattern registered in a dictionary. In that case, it is necessary to completely adjust the inclination of the image of the recognized figure and the standard pattern, so if the figure to be recognized is written at an angle, the process of rotating the image data to the correct position is necessary as preprocessing. It will be done.

[Conventional technology]

FIG. 7 shows a conventional image data rotation method.

In FIG. 7, reference numeral 21 denotes a rotation calculation unit that performs rotation calculation processing on image data. A rounding unit 22 performs a rounding operation so that the real X and Y coordinate values of the rotated image data become integer values.

Next, the operation shown in FIG. 7 will be explained with reference to the explanatory diagram of the image data rotation method shown in FIG.

Now, suppose that the character C is tilted by θ with respect to the X coordinate, as shown in FIG. 6 (al).
The image of the letter C tilted by 0 is rotated clockwise by θ according to the following equation (1).

However, regarding the direction of θ, the counterclockwise direction is positive.

Here, X and y are the X of each black pixel of character C before rotation processing.
and Y coordinates, and X and Y are the X and Y coordinates after rotation processing.

By the way, since each pixel of the image is generated on the main grid, its coordinates are represented by integer values. However, since the X and Y coordinates determined in (1) above are generally real values,
It is necessary to convert it to an integer value.

The rounding calculation unit 22 converts the real-valued X and Y coordinate values obtained by equation (1) into integer-valued X and Y coordinate values by performing rounding such as rounding.

In the manner described above, it is possible to obtain a rotated image of the letter C in the correct position without any inclination, as shown in FIG. 6 (bl).

[Problem that the invention seeks to solve]

As mentioned above, the conventional image data rotation processing method is (11
The X and Y coordinate values obtained in step 2 were rounded to generate integer X and Y coordinate values.

However, when the X and Y coordinate values are converted into integer values by such rounding processing, a shift occurs in the position of the converted pixel. For this reason, phenomena such as pixels that were adjacent to each other before conversion are no longer adjacent to each other after conversion, or holes are created among the black pixels that make up the image occur.

For example, as shown in Figure 8(a), the x and y coordinate values are (5
, 6) and (6, 5),
When rotated counterclockwise by 30 degrees, the X of point P after rotation
And the Y coordinate values (Xp, Yp) are set as θ=3001x-5 and y=6 in the above equation (1), so that XP=
1.33,y,=7.70. Also, the X and Y coordinate values of point Q after rotation (Xo.

y,) is θ=30', x=6, Y-5 in equation (1)
By setting, Xo=2.70, 'y', =7.3
It becomes 3.

These P (XP, Y,) and Q (xo , YQ
) are rounded off and converted to integer values, and the coordinates are P' (XP ', Yp
) and Q'(Xa',Ya'), then (Xp').

YF') = (1, 8), (xo', yo') = (3, 7
). These P' and Q' form two discontinuous points that are not adjacent to each other, as shown in FIG. 8 (bl).

In this way, with conventional image data rotation methods, two points that were adjacent before the rotation process become disconnected due to the rotation process, or a hole appears in a blackened area after the rotation process. etc., there is a problem in that the distortion of the figure due to the rotation process is large, resulting in a reduction in the recognition rate.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image data rotation method that reduces distortion occurring in images after rotation processing and improves recognition rate.

[Means for solving problems]

The means taken by the present invention to solve the above-mentioned problems in the conventional image data rotation method will be explained with reference to FIG.

FIG. 1 is a block diagram showing the basic configuration of the present invention.

In FIG. 1, 11 is an interpolation pixel generation unit, which generates an interpolation black pixel at an intermediate position when there are continuous black pixels in any of the vertical, horizontal, and diagonal directions on the main grid of the image. . - Black pixels are pixels that make up an image.

Reference numeral 12 denotes a rotation calculation unit that performs calculation processing to rotate the coordinates of the black pixels on the main grid and the interpolation black pixels generated by the interpolation pixel generation unit 11 by a predetermined angle.

A rounding calculation unit 13 performs a process of converting the coordinate values of black pixels and interpolated black pixels obtained by the rotation calculation unit 12 from real values to integer values.

[For production]

The operation of FIG. 1 will be explained with reference to FIGS. 3 and 4.

FIG. 3 is an explanatory diagram of interpolation pixel generation processing of the present invention, and FIG. 4 is an explanatory diagram of an example of image data rotation processing and rounding processing of the present invention.

As shown in FIG. 3(a) to +dl, when there are continuous black pixels P and Q in any of the vertical, horizontal, and diagonal directions of the main grid of the image, the interpolation pixel generation unit 11 An interpolated black pixel S is generated at an intermediate position, for example, on a 1/2 grid point.

The rotation calculation unit 12 performs calculation processing to rotate the black pixels (P, Q) on the main grid and the interpolated black pixels (S) by a predetermined angle (θ). This rotation calculation process is, for example,
This is performed using the above-mentioned equation (1).

The rounding calculation unit 13 performs a process of converting the coordinate values of the black pixels (P, Q) and the interpolated black pixel (S) obtained by the rotation calculation unit 12 from real values to integer values.

Now, taking the case of Fig. 4 as an example, a concrete explanation will be given. Fig. 4 (al is the black pixel p (XP=51
yp = 6) and Q (XQ = 6- yo =
5) and interpolated black pixel S (Xs = 5.5, )'
s −5,5).

After the coordinates of these P, Q, and 8 points are rotated, for example, by 30 degrees counterclockwise by the rotation calculation unit 12, the respective X and Y coordinate values are rounded off by the rounding calculation unit 13, for example, and converted from real values. When converting to integer values, P, Q
and 8 points are P' (XP
' = 1, YP ' = 8), Q'(xo'
=3, Yo' =7) and S'(xs'=2
, Ys' = 8). Therefore, points P and Q, which were adjacent to each other before the rotation process, will maintain their adjacent relationship after the rotation process, and no discontinuity will occur. (Explained in more detail in Section 1).

As described above, when there are continuous black pixels in any of the vertical, horizontal, and diagonal directions on the main grid of the image, after generating interpolated black pixels at intermediate positions, rotation and rounding processing is performed. As a result, it is possible to reduce the distortion that occurs in the figure after rotation processing and improve the recognition rate.

〔Example〕

Embodiments of the present invention will be described with reference to FIGS. 2 to 5.

FIG. 2 is an explanatory block diagram of the configuration of an embodiment of the present invention, and FIG. 5 is an explanatory diagram of an example of a rotated image generated by the embodiment. 3 and 4 have already been described.

(A) Configuration of the embodiment In FIG. 2, an interpolation pixel generation section 11, a rotation calculation section 12
and the rounding calculation unit 13 are as already described.

Reference numeral 14 denotes an image memory in which image data of the figure to be recognized is stored.

Reference numeral 15 denotes an inclination angle detection section which takes out image data of the figure to be recognized from the image memory 14 and detects the inclination angle of the figure to be recognized.

(B) Operation of the embodiment The operation of the embodiment will be explained with reference to FIGS. 3 to 5.

The tilt angle detection section 15 takes out the image data of the figure to be recognized from the image memory 14, and detects the tilt angle θ of the figure to be recognized using various known methods.

On the other hand, the interpolation pixel generation unit 11 generates (al to (dl)
As shown in , when there are continuous black pixels P and Q in any of the vertical, horizontal, and diagonal directions on the main grid of the image,
An interpolated black pixel S is generated at the intermediate position. In this embodiment, interpolated black pixels S are generated on 1/2 grid points as shown.

The rotation calculation unit 12 calculates, for each black pixel and interpolation pixel constituting the image on the main grid, the magnitude of the tilt angle θ detected by the tilt angle detection unit 15 in the opposite direction, that is, (-θ).
only rotates.

This rotation calculation is performed at θ in equation (1) above.
−〇).

Rounding operation unit 3 performs a process of converting the coordinate values of each black pixel and interpolated black pixel obtained by the rotation operation unit 12 from real values to integer values by rounding them off in this embodiment.

Now, assuming that the tilt angle θ detected by the tilt angle detection unit 15 is 30 degrees, and specifically explaining the case of FIG. 4 as an example, in FIG. 4(a), black pixels P and Q are as follows. These are black pixels of the figure to be recognized existing on the main grid. The X and Y coordinates of black pixels P and Q are (xpsyp) and (xo
+3'o), then X9=5,)'p=6 and X0=
6, yo-5.

Since the black pixels P and Q exist consecutively in the diagonal direction, the interpolated black pixel S is generated at an intermediate position.

In this embodiment, the 1/2 grid point is the intermediate position, so if the X and Y coordinates of the interpolated black pixel S are X5 and y, then x = 5.5 and yi -5,5.

The rotation calculation unit 12 performs calculation processing to rotate each black pixel P, Q, and S by θ (=30 degrees) using equation 11. After rotation processing, each black pixel P, Q and each of the 8 points Y coordinate (
Assuming that XP, Y, ), (XQ, YQ) and (Xs, Ys), these are calculated as follows.

XP=1.33, YP=7.70xo=2
．． 70, Yd = 7.33Xs = 2.01
, y, =7.51 When the rounding unit 13 performs rounding and converts these real number coordinate values into integer coordinate values, the points corresponding to P, Q, and S after conversion are converted to P'
, Q', S', and their X and Y coordinates are (XP'
, YP'), (xa'tYo') and (Xs
', Ys'), these are calculated as follows.

X,'=], , YP'=8X,'=3
, Y,'-7 The black pixels P and Q, which were adjacent to each other, maintain their adjacent relationship even after the rotation process, and no discontinuity occurs. In the case of black pixels P' and Q' that are formed by performing rotation processing and rounding processing on the same black pixels P and Q using the conventional method, the adjacent relationship is not maintained and discontinuities occur, which is the eighth problem. This is as already explained in the figure.

FIG. 5 shows examples of rotated images obtained by the conventional method and the image data rotation method according to the present invention. (c) shows the rotated image of the letter A obtained by the method of the present invention. (As is clear from comparing the rotated images in bl and (C), According to the present invention, unevenness and holes in the image are reduced compared to the conventional method, and a clean rotated image can be generated.

〔Effect of the invention〕

As explained above, according to the present invention, there are no interruptions, irregularities, holes in black pixel areas, etc. in the image after rotation processing, and the recognition rate of characters, symbols, etc. written at an angle is improved. It can be improved.

[Brief explanation of drawings]

Fig. 1...Explanatory diagram of the basic configuration of the present invention, Fig. 2...
FIG. 3 is an explanatory diagram of the configuration of an embodiment of the present invention. FIG. 4 is an explanatory diagram of an example of the image data rotation process and rounding process of the present invention. Fig. 5: An explanatory diagram of an example of rotating image generation using the method of the present invention and a conventional method. Fig. 6: An explanatory diagram of an image data rotation method. Fig. 7...
- An explanatory diagram of a conventional image data rotation method, FIG. 8: An explanatory diagram of an example of conventional image data rotation processing and rounding processing. 1 and 2, 11... interpolation pixel generation unit, 12... rotation calculation unit, 1
3... Rounding calculation unit, 14... Image memory, 15...
・Inclination angle detection section.

Claims (1)

[Claims] In an image data rotation method that generates a rotated image by rotating an image by a predetermined angle, (a) continuous black pixels exist in any one of the vertical, horizontal, and diagonal directions on the main grid of the image. (b) For the black pixel on the main grid and the coordinates of the interpolated black pixel generated by the interpolated pixel generator (11), (c) converting the coordinate values of the black pixels and interpolated black pixels obtained by the rotation calculation unit (12) from real values to integer values; An image data rotation method characterized by comprising a rounding calculation unit (13) that performs processing.