JPS6220074A - Image processing method - Google Patents

Abstract

PURPOSE:To execute an image conversion processing at a high speed and also without requiring a page memory with large capacity, by executing the magnification, rotation and reduction of an original image data. CONSTITUTION:Magnification interpolation rotation interpolation reduction processings and a parallel movement processing during that time are executed with respect to a data of every 1 line from an image reader 3, etc. In such a way, the parallel processing can be executed at a high speed, even as a buffer memory 4 before a conversion, a memory capacity of a full page of one page portion is not required, and by only providing a memory capacity of only a several line portion, an image conversion processing can be executed. Also, the interpolation processing is completed, a smooth and natural interpolation can be executed, and an image processing of a high quality can be executed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image processing method in an image processing apparatus that performs data conversion processing on image data.

[Prior art] Conventionally, image data has been transformed by enlarging/reducing by 1M, rotation, translation, etc. (generally referred to as affine transformation).
The method to do this is to have full page memory before and after conversion, calculate the address before conversion that corresponds to the address after conversion, and read data into the page memory before conversion each time the conversion is performed. I went to

[Problems to be solved by the invention] Therefore, in the conventional method, even in the case of raster data obtained by directly reading the original image with a line sensor, etc., a full page memory is required to store the original image data before conversion. .

The capacity of the full page memory is 16 petS/mm for one A4 size page, and 16 Mbytes for image data with a depth of 8 bits.

In the case of color image data, a storage capacity that is at least three times this amount is required, and in this case, the capacity of the full page memory becomes 48 Mbytes.

Since image conversion processing requires such a large capacity memory, devices capable of image conversion processing have become complicated and expensive.

In addition, processing control is complicated, and a lot of time is spent on image transformation processing such as enlarging/reducing, rotating, and translating raster data, and there is no effective method to perform these processes at high speed. Ta.

[Means for Solving the Problems] The present invention has been made for the purpose of eliminating the problems of the above-mentioned prior art. For example, as a means of solving this problem, it is possible to The image forming apparatus includes a line buffer memory having a capacity, an enlarging means for enlarging original image data, a rotating means for performing an image rotation process, and a reducing means for performing an image reducing process.

[Operation] In such a configuration, by performing an enlargement process in which the original image data is enlarged, a rotation process in which the image rotation process is performed subsequent to the enlargement process, and a reduction process in which the image reduction process is performed subsequent to the rotation process, Image conversion processing can be executed at high speed and without requiring a large capacity page memory.

Embodiment 1 Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

Configuration of this embodiment] FIG. 1 is a block diagram of an embodiment according to the present invention, in which 1 is a CPU that controls the entire control of this embodiment, and 2 is a CPU.
This is the main memory in which control procedures and the like, which will be described later, are stored. 3 is an image reading device for reading image information; the image reading device 3 reads an original image with a line sensor, and sequentially outputs the read image data line by line as raster data. A buffer memory 4 temporarily stores raster data sent from the image reading device 3, and has a memory capacity for only several lines of raster data. 5
6 is an enlargement processing unit that performs necessary enlargement processing on a predetermined area (predetermined portion) in the image data from the buffer memory 4, and 6 is an enlargement processing unit that captures missing pixel data after enlargement processing and rotation processing are performed. and an interpolation processing unit that performs averaging operations on these pixel data, 7 a rotation processing unit that performs a predetermined rotation process as necessary on the image data, and 8 a reduction process as necessary on the image data. a reduction processing unit that executes
is a page memory that stores image data that has been subjected to image conversion processing and the like in units of pages. Further, 20 is a bus that connects each processing section.

[Operation of the Embodiment] FIG. 2 shows a process diagram of the image data conversion process in this embodiment, which consists of the following configuration.

Hereinafter, the image data conversion process shown in FIG. 2 in this embodiment will be explained in detail using the example shown in FIG. 3 as a specific example.

30 shown in FIG. 3(a) indicates the size of one page of the original read by the image reading device 3, and the portion shown at 31 in this one page of the original 30 is enlarged/1i! This is an example in which image data conversion processing such as small size, rotation, middle/line movement, etc. is performed, and the image data is stored in the area 33 in the area 32 of one page of the page memory 9 shown in FIG. 3(b).

The buffer memory 4 of this embodiment is the COD of the image reading device 3.
The pixel data (raster data) sent in the 1st hour system is sequentially transferred to a pipe according to the process shown in Figure 2. Image data that has been subjected to line processing and conversion processing is sequentially written to page memory.

As shown in Figures 3(a) and (b), when performing all conversion processes such as enlargement, rotation, translation, and reduction, all steps shown in Figure 2 are executed in sequence, and if there is any unnecessary processing, skips those processes and moves on to the next process.

Since the example shown in FIG. 3 includes an enlargement process, the enlargement processing section 5 first executes an integer times enlargement process of 110. If the magnification factor is α, then for integer times expansion of 1, if α is an integer, it will be multiplied by 6, and if α is not an integer, it will be multiplied by 6 ([α+1
) times ([ is a Gaussian symbol). That is, when 1 x α x 2, the image is expanded twice, and when 2 x α x 3, it is expanded three times.

In this way, in this embodiment, in order to once enlarge the magnification beyond the actual magnification, the reduction processing step 150 described later is executed to correct the magnification. That is, if α is not an integer, the reduction processing step 150 /([α]+1).

When the integer multiple enlargement processing step 110 is executed, pixels with no data (missing processing) inevitably occur, so the next interpolation step 1 is performed.
20 performs data interpolation.

Then, a rotation processing step 130 is performed using a known method. Here, too, as a result of the rotation process, there are pixels with no data, so the interpolation process step 140 is subsequently executed.

Then, the reduction processing step of 150 is executed.

In enlargement mode, if the enlargement rate α is not an integer, α/
Execute reduction by a factor of ([α]+1).

Furthermore, when the mode is not the enlargement mode but the reduction mode, the light integer times enlargement processing step 110 and the interpolation step 120 are not executed, and the reduction processing is performed according to the reduction ratio given in the reduction processing step 150.

Of course, even when rotation processing is not necessary, the rotation processing step 130 and the interpolation step 140 are skipped and not executed.

Note that the parallel movement process can be handled only by address conversion processing for the page memory 9 without generating pixels with no data. Therefore, it may be inserted and executed before any processing step of the integer-fold enlargement processing step 110, the rotation processing step 130, or the reduction processing step 150, and may be selected and executed in consideration of efficiency. Thereby, operability can be improved.

Next, the necessity of separately performing the interpolation process 120 in addition to the interpolation process 140 will be explained.

Without the two interpolation steps 120 and 140, all transformations such as enlargement/reduction, translation, and rotation can be performed only by coordinate transformation, and moreover, they can also be coordinate transformed as one combined transformation (affine transformation). So it looks like it's very convenient. However, these two interpolation steps 12
For example, when the image data shown in FIG. 4(a) is enlarged twice and rotated by 45 degrees, if all interpolation is performed in one step (interpolation step 140), As shown in FIG. 4(b), the number of pixels for which data must be interpolated increases, and interpolation that naturally connects pixel data becomes extremely difficult. For this reason, the integer times enlargement step 110
The interpolation step 120 is also executed after . This allows for smooth and natural interpolation.

Furthermore, instead of performing enlargement and interpolation completely, image processing can be performed in combination.

This example will be explained below with reference also to the timing chart of Figure 7FS5.

In the integer multiple enlargement step 110, first consider the enlargement step in the main scanning direction (X direction). Here, the magnification rate α=1.
If it is 2, the enlargement process in the integer-fold enlargement step 110 will be [α]+1=2, and a two-fold enlargement will be executed. For example, "A", "B" shown in FIG. 5(a),
Each data of "C'" is at address "0" of buffer memory 4.
°°, 41''.

In this case, the read clock from the buffer memory 4 is twice the pixel clock (read address increment clock) (Fig. 5(b)
) l. As a result, the data read from the buffer memory 4 becomes as shown in FIG. 5(C), and as a result, the data is expanded twice. At the same time, interpolation between the enlarged data is also performed.

Note that the interpolated data here consists of the same pixel data consecutively corresponding to the enlargement magnification, but for example, after this, the interpolation processing section 6 performs an averaging operation on two main scanning pixels (pixels corresponding to the enlargement magnification). It can also improve data connectivity.

Similarly, for the enlargement process in the sub-scanning direction (Y direction), 2
When enlarging the data twice, it is sufficient to perform control such that data for one line is read out repeatedly. Here, if you want to change the readout line using the water mode synchronization signal, a counter that receives the water mode synchronization signal as input is provided, and what happens when it is twice as high? When the horizontal period signal is doubled, it is used as a decimal counter, and as a ternary counter when the horizontal period signal is doubled (as a ternary counter when it is tripled), and once every third time when the horizontal period signal is expanded to 3 times. ).

In this case as well, if the interpolation processing section 6 performs an averaging operation on the pixels in the sub-operation direction, the connection of data can be made clear.

In the manner described above, the integer multiple enlargement processing step 110 and the interpolation step 120 can be executed.

Next, the rotation processing step 130 and the interpolation step 14.6 will be explained.

In the rotation processing step 130, the rotation processing section 7 first performs coordinate transformation, and performs rotation processing by making the data correspond to the pixel (address) closest to the coordinates after the transformation. However, rotation processing may result in pixels having no data. However, since the rotation here is a same-magnification rotation, two or more pixels without data do not occur consecutively. Therefore, the interpolation processing unit 6 converts the data of one pixel into 2×2 pixels adjacent to the coordinates transformed by rotation processing.
write to the pixel. For example, the address of the nearest pixel (i
, D, then (++1.0.

The same data is also written to (i, ++1) and (++1, ++1). This process is performed on raster data sequentially; if data is written to the same address more than once, the data written later becomes valid. For this reason,
Pixels with no data do not occur, and data connections do not become unnatural.

In this manner, the rotation processing step 130 and the interpolation step 140 can be performed.

Next, the reduction processing step 150 by the reduction processing section 8 will be explained.

In the case of the integer magnification mode, no reduction processing is performed, but in the reduction mode, the reduction processing is performed as is with the set reduction ratio β, and in the enlargement mode and the enlargement ratio α is not an integer, the reduction processing is performed as is. =α/([α]+1) Reduction processing is performed.

This reduction process is performed by thinning out the data when writing to the page memory 9. As this thinning method, a method using a rate multiplier or the like may be considered, but in this embodiment, the following method is used.

First, reduction in the main scanning direction (X direction) will be explained.

Here, the reduction magnification is β, the address of the input data to the reduction processing unit 8 is m, and the write address to the page memory 9 is n.
Then, input data at address m that satisfies n=[βm] is written to address n of page memory 9. Therefore, although a plurality of pieces of manual data may be written to address n, the data written last becomes valid, and the data written first is thinned out.

For example, the address m shown in FIG. 6(a) is stored in the reduction processing unit 8.
When the input data is sent corresponding to the expansion rate α
When = 1.2, reduction rate β = α / ([α] + 1) = 0
．． 6, and the data written to the address n of the page memory 9 becomes the data shown in FIG. 6(b). Therefore, n
=o is written with ``D'', n=1 is written with ``D'', and n=2 is written with ``E''.

The relationship between the input data to the address m to the reduction processing unit 8 and the write address n and write data to the page memory 9 is shown in the table below.

Regarding the reduction processing in the table sub-scanning direction (Y direction), the M lines of input data may be associated with the N lines of data to be written to the page memory 9, and the reduction processing may be performed in the same manner as described above.

The above processing is performed completely independently in each processing section,
For example, when raster data from the image reading device 3 is sent to the buffer memory 4, the data previously stored in the buffer memory 4 is enlarged by the enlargement processing unit 5 and the interpolation processing unit 6 at the same time (pipeline processing ) to be done.

The converted image data stored in the page memory 9 is sequentially read out and output by the image output device IO. The image output device 10 is, for example, a device that outputs image data onto a CRT display device or permanently displays image data using an image printer.

As explained above, according to this embodiment, the data for each line from the three image reading devices is enlarged → interpolated → rotated →
Since we decided to perform the process of reducing → reduction and the f-stroke movement process in between, we were able to perform parallel processing at high speed, and
Even as the buffer memory 4 before conversion, a full page memory capacity for one page is not required, and image conversion processing can be performed with only a memory capacity for a few lines.

Further, since the processing is performed according to the above-described steps, the interpolation processing is also perfect, and smooth and natural interpolation can be performed, allowing high-quality image processing to be performed.

[Effects of the Invention] As described above, according to the present invention, image information conversion processing can be performed at high speed and without requiring a large capacity memory, and with a simple configuration. Furthermore, since the necessary interpolation processing is performed each time image conversion is performed, a very natural, high-quality image conversion process has been realized.

[Brief explanation of the drawing]

Figure 1 is a block diagram of an embodiment of the present invention, Figure 2 is an image conversion process diagram of an embodiment of the present invention, and Figures 3 (a) and (b) are diagrams of image conversion of this embodiment. Figures 4(a) and 4(b) are diagrams illustrating an example of image enlargement/rotation processing when no interpolation processing is performed, and Figures 5(a) to 5(b) are diagrams explaining examples.
C) is an enlargement/interpolation processing timing chart of this embodiment, and FIGS. 6(a) and 6(b) are reduction processing timing charts of this embodiment. In the figure, 3... Image reading device, 4... Buffer memory, 5... Enlargement processing section, 6... Interpolation processing section, 7...
Rotation processing section, 8: Reduction processing section, 9: Page memory, 10: Image output device. Patent applicant Canon Co., Ltd. Representative Patent attorney Yasunori Otsuka ---; -' Figure 3 (G) A 3゜Figure 4 (0) Figure 3 (b) Figure 4 (b)

Claims (4)

[Claims] (1) In an image processing method in an image processing device that performs data conversion processing on image data, an enlargement step of performing enlargement processing on original image data as necessary; a rotation step of performing image rotation processing subsequent to the enlargement step;
An image processing method comprising a reduction step of performing image reduction processing subsequent to the rotation step. (2) The image processing method according to claim 1, further comprising an interpolation step for interpolating missing image data between the enlargement step and the rotation step and between the rotation step and the reduction step. (3) The image processing method according to claim 1 or 2, wherein the enlarging step is an integer-fold enlarging process. (4) The image processing method according to any one of claims 1 to 3, wherein the original image data is raster data sent line by line.