JPH0228891A - Graphic rotation processor - Google Patents

Abstract

PURPOSE:To rotate a rectangular area at high speed without a picture element getting out by executing a rotational processing to a picture element unit based on an inclined pattern to show the rotational angle of a graphic after the rotation and one part of picture data. CONSTITUTION:The inclined pattern corresponding to an inputted rotational angle theta is generated in an inclined pattern generating part 3 and stored in an inclined pattern storage part 4. On the other hand, in parallel, the picture data including the color and attribute of the picture elements for (n) lines are transferred from a picture memory 1 to cache memories 21, 22...2n. Next, the inclined pattern storage part 4 is accessed and the attention-paid picture elements of a destination area are scanned one by one by the inclined pattern. The corresponding picture elements of a source area corresponding to the attention-paid picture elements are interpolation-processed and the color and attribute of the attention-paid picture elements are specified. The picture data of the destination area are transferred from an arithmetic part 5 to the picture memory 1. Thus, the rotational processing is terminated.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a figure rotation processing device that performs rotation at an arbitrary angle, and more specifically, the present invention relates to a figure rotation processing device that performs rotation at an arbitrary angle. The present invention relates to a figure rotation processing device that rotates a figure.

[Prior Art] Image rotation is a basic coordinate transformation function frequently used in image processing such as image editing and pattern matching, and the development of high-speed processing technology has become an important issue.

Conventional image processing devices that rotate figures are disclosed in Japanese Patent Laid-Open No. 57-121753 and Japanese Patent Laid-Open No. 57-117061, which limit the rotation angle to 90° or 45° and perform high-speed processing. There is something that does the processing.

However, if the rotation angle is limited, the applications thereof will also be limited.

Also, as one of the conventional techniques for rotating figures at arbitrary angles,
Some methods perform figure rotation using affine transformation and interpolation processing. This performs affine transformation on all pixels in the source region, which is the region before rotation, and moves the coordinates after the transformation to the nearest pixel by interpolation processing such as nearest neighbor interpolation or bilinear interpolation, and rotates the figure. This is what we do. In addition, as other conventional technology, "Transactions of the Institute of Electro-Communications Engineers J '86 Vol.
J69-D No.], which utilizes two-step coordinate transformation via an intermediate image to make it possible to collectively access memory for a plurality of pixels.

[Problem to be solved by the invention]

In the former method of conventional technology, which rotates a figure at an arbitrary angle,
Since the pixels in the rotation area, which is the area after rotation, and the pixels in the source area do not have a one-to-one correspondence, interpolation processing such as nearest neighbor interpolation or bilinear interpolation is required as described above. Since this interpolation process uses information in the vicinity of a position in the source area that corresponds to one pixel in the destination area, there is a problem in that the image data in the source area must be accessed many times.

Furthermore, even if the source area is a rectangle along the image coordinate axis, the destination area rectangle will be tilted with respect to the image coordinate axis, so if the boundary conditions of the destination area correspond to the pixels of the coordinate axis, the boundary Conditions become mathematically complex.

Therefore, there is a problem in that it is difficult to scan all pixels in the destination area as rotated pixels without omitting or overlapping them.

On the other hand, in the latter conventional technique, although the number of memory accesses is reduced, a medium for storing intermediate images is required, and the coordinates of the image must be transformed twice.
There is a problem in that the number of calculations is large.

This invention was made to solve the problems of these conventional techniques, and by performing rotation processing pixel by pixel based on a tilt pattern indicating the rotation angle of a rotated figure and a part of image data. An object of the present invention is to provide a graphic rotation processing device that rotates a rectangular area at high speed without missing pixels.

[Means to solve the problem]

A figure rotation processing apparatus according to the present invention performs mouth shape processing by raster scanning and rotates a rectangular area having image data, and the figure rotation processing apparatus rotates a rectangular area having image data. a tilt pattern generating means for generating the tilt pattern, a tilt back storage means for storing the tilt pattern, an image data storage means for temporarily storing a part of the image data in the rectangular area to be rotated, and the tilt pattern; The image processing apparatus is characterized by comprising a calculation means for rotating image data of a rectangular area pixel by pixel based on the image data stored in the image data buffer means.

[Effect]

In this invention, when a rectangular area is rotated by an arbitrary angle, a tilt pattern corresponding to the angle is generated by the tilt pattern generating means, and is stored in the tilt pattern storage means. Then, some of the image data is temporarily stored in the image data buffer means from the image memory in which the image data is stored, and the slope pattern stored in the slope pattern storage means and the image stored in the image data buffer means are Based on the data, the calculation means rotates the rectangular area.

〔Example〕

The present invention will be explained below based on the drawings showing one embodiment thereof. FIG. 1 is a block diagram showing the configuration of a figure rotation processing device according to the present invention. In the figure, reference numeral 1 denotes an image memory that stores images of a source area and a destination area, and images read from the image memory 1 are shown in FIG. Data is provided to the selector 6 via the data bus. The selector 6 selects n cache memories 2L 22...2n, which are image data buffer means.
One line of the image pig is stored in each of the selected cache memories 21, 22, . . . 2n. The image data for lines 1 to n stored in the cache memories 21, 22, . cache memory 2122
...2n image data are given.

On the other hand, the slope pattern generation section 3 is configured to use, for example, a DDA (Digi
tal Differential Analyzer
A tilt pattern is generated by inputting the rotation angle from the relationship between the rotation angle and the tilt pattern generated using a digital differential analyzer. The generated slope pattern is stored in the slope pattern storage section 4, and is given to the calculation section 5 in response to an access request from the calculation section 5. That is, the calculation unit 5 determines the color and attribute of the pixels in the destination area by referring to the slope pattern and the n lines of image data stored in the cache memories 2122 . . . 2n, and writes them into the image memory 1.

The second region is a schematic diagram showing the relationship between the source region and the destination region, and the horizontal direction of the diagram is the X direction, and the vertical direction is the X direction.

Here, for simplicity, a rectangular area of 16 pixels in the X direction and 10 pixels in the X direction is centered at point 0 (0, 0), and the rotation angle θ -
Let us consider the case of jan-'6/16. Also p (1
5, O), Q (0, 9) is the vertex of the source region, P'
, Q' are vertices of the destination area corresponding to vertices P and Q.

Letting the coordinates after the inverse transformation be (χ, y), it can be expressed by the following formula.

At this moment, pixel D'2 in the destination area
．． This is position D2 in the source region of (2,1).

Next, the operation will be explained.

FIG. 4 is a flowchart of the rotation process of this invention,
First, in step 1, the rotation angle θ is determined by the tilt pattern generation unit 3.
Enter. When the rotation angle θ is manually set, the tilt pattern generation unit 3 generates a tilt pattern corresponding to the rotation angle θ. For example, the slope pattern is created using DI)A and stored in a numerical table in the slope pattern generation unit 3. Here, the slope pattern is 1010010010100.
100".

As shown in Fig. 3, the relationship between the slope pattern and the straight line of the destination area is "1° is assigned to
The coordinates of ' can be determined by affine transformation f.

The affine transformation f can be expressed by the following formula, where the coordinates before transformation are (x, y) and the coordinates after transformation are (x', y').

Therefore, the vertex P' is similarly the vertex Q' is Q'(x' Q+ y' q )=
(-3, 8).

FIG. 3 is a schematic diagram showing the correspondence between pixels in the source region and pixels in the destination region, and black dots indicate positions in the source region when the region pattern is subjected to affine inverse transformation f-'. The affine inverse transformation j-1 is such that a straight pixel is shifted by "1" in the X direction at coordinates obtained by (x', y') from the coordinates before the inverse transformation. Next, the slope pattern generated in step 2 is stored in the slope pattern storage section 4. On the other hand, in parallel, in step 4, image data including the colors and attributes of 1947 pixels is transferred from the image memory 1 to the cache memories 21, 22, . . . , 2n.

The subsequent steps are processed by the calculation unit 5. Next, in step 5, the coordinates of vertices P' and Q' are determined by affine transformation f. This determines the scanning range of the inclined pattern. Next, in step 6, the tilt pattern storage section 4 is accessed, and the target pixel in the re-destination area is scanned pixel by pixel using the obtained tilt pattern. In step 7, the corresponding pixel corresponding to the pixel of interest is converted to n of the cache memory 2L 22...2n in the source area by inverse affine transformation r-1.
Access by referring to line image data.

Next, in step 8, the accessed corresponding pixels are interpolated to determine the color and attributes of the pixel of interest. Also step 6~
In parallel to step 8, in step 9, the image data of the next line is transferred from the image memory 1 to the empty cache memories 2L 22...2n. Then, in step 10, it is determined whether or not scanning has been completed for all pixels in the destination area. If it has not been completed, the process returns to step 6. If it has been completed, the image data of the destination area is transferred from the calculation unit 5 to the image memory 1. Transfer to. This completes the rotation process.

Further, when the affine inverse transformation f-I is applied, the corresponding points of the pixel of interest indicated by black dots gather near the straight line in the direction of the line segment plane, that is, the χ direction. Therefore, when accessing the corresponding pixel, line segment OP
By transferring the image data of several lines in the vicinity from the image memory 1 to the cache memories 21, 22, . This speeds up interpolation processing.

For example, when co-temporal interpolation is used as the interpolation process, in order to determine the color and attribute of the pixel D'21 in the destination area, the point in the source area obtained by inversely transforming the pixel of interest D'21 (2,1) is Four corresponding pixels S, , (2,0), S, (3,0) near D21(2,2,0,2)
), S.

It is necessary to refer to the colors and attributes of (2, ILS, , (3, 1), but store the image data of lines l and I!z in the cache memory 2L 22...2n and refer to it. Therefore, the corresponding pixels 520+3301 3
211 33 can be accessed at high speed.

As mentioned above, the dispersion of the inversely transformed points is expressed by the line segment surface,
That is, since the cache data is gathered near line lo, the required cache memory capacity is only for several lines. and the third
The starting point of scanning the destination area in the figure is Q
’ while shifting the cache memory 21, 22,
...The line data stored in 2n is sequentially updated and rotated.

In addition, line l is connected to cache memory 2L 22...2n.
.

When the line data of ~e is stored and the pixel cannot be determined by the rotation process, the next thing that is needed is the line data of line lli+1, so prepare a cache memory for one more line, While the calculation unit 5 is performing rotation processing using the line data of lines l, to lJ, line! ,. If line data of 1 is stored in a prepared cache memory, processing time can be further reduced.

Also, since this figure rotation processing device uses Ryofin transformation, it is possible to rotate with enlargement or reduction by changing the parameters, but in the case of rotation with reduction, the variation in the inversely transformed points is , increases as the reduction rate increases, so in order to access the reference pixels in the source area with a high hint rate, it is necessary to store more line data in the cache memory.

〔Effect of the invention〕

According to the present invention, when performing rotation processing, by obtaining a tilt pattern and using it, the pixel of interest can be scanned at high speed along the rectangular sides of the destination area without performing complicated calculations.

Furthermore, since the destination area is scanned by following the tilt pattern fixed to the coordinates, missing pixels will not occur.

Furthermore, by storing part of the image data in the source area in a cache memory that can be accessed at high speed with a relatively small capacity, an equivalent effect can be achieved in which the corresponding pixels in the source area can be accessed at a high hit rate.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a figure rotation processing device according to the present invention, FIG. 2 is a schematic diagram showing the relationship between the source area and the destination area, and FIG. 3 is a schematic diagram showing the relationship between the pixels of the source area and the destination area. FIG. 4 is a schematic diagram showing the correspondence relationship with pixels after affine inverse transformation, and is a flowchart of rotation processing. 1... Image memory 21. 22... 2n... Cache memory 3... Slope pattern generation section 4...
Inclined pattern storage unit - operation unit patent Applicant: Sanyo Electric Co., Ltd. Agent Patent attorney: Noboru Kono

Claims (1)

[Claims] 1. In a figure rotation processing device that performs figure processing by raster scanning and rotates a rectangular area having image data, a tilt pattern is generated that represents the inclination of a straight line forming the rectangular area after rotation. tilt pattern generation means; tilt pattern storage means for storing the tilt pattern; image data buffer means for temporarily storing a part of image data in a rectangular area to be rotated; and the tilt pattern and the image data. 1. A graphic rotation processing device comprising: arithmetic means for rotating image data of a rectangular area pixel by pixel based on image data stored in a buffer means.