JPS6220072A - Image information processor - Google Patents

Abstract

PURPOSE:To execute the processing at a high speed by constituting the titled processor so that a raster data can be brought to a deformation processing, without providing a random access memory on an input side, and constituting the circuit of only addition, subtraction, etc. CONSTITUTION:The titled processor is provided with a raster buffer 5 being a line memory, a parameter memory 11 for storing the first parameter, a data processor 2 being an interpolating means for interpolating a picture element data, an image converter 50 being a converting means, a parameter memory 27 for storing the second parameter, and an image memory 3 being a storage means. An inputted raster data 12 is stored in the raster buffer 5, and a control device 4 sets a parameter corresponding to the raster data, to parameter memories 11, 12. In accordance with the contents of this parameter memory 11, the raster data 12 is interpolated by the data processor 2. On the other hand, in the image memory 3, a picture element data 32 which is converted by the image converter 50 is written in a memory array 26 indicated by address counters 28, 29 which are operated, based on a parameter memory 27.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image information processing device that transforms digitized image information by digital signal processing.

[Prior Art] Affine transformation such as rotation, translation, reduction and enlargement of a two-dimensional image is performed at a point (X) in the image coordinates before transformation.

y) to the point (U, V) in the image coordinates after transformation, then (x, y) and (U
, V). Here, α, β, γ, δ, ε, and ζ are transformation rasters.

U = alpha
Missing pixels in the image after conversion (this is due to the mesh of (X, Y) and (U, V) due to digitization
In order to avoid the existence of a mesh of y), and generally (x,
y) does not match the previous mesh, a method has been used to calculate the converted pixels from surrounding mesh points. This inverse transformation can also be expressed by the third and fourth equations, which are similar to the first and second equations. Here, π, ρ, σ, τ, φ, and ω are parameters for inverse transformation, and are derived from α, β, γ, δ, (, ζ by a predetermined method.

X = πU + ρV + σ - + + φ3rd 3rd y = τU
+φV+ω...-4th equation and problem to be solved by the invention] In the prior art, such inverse transformation was calculated for each pixel, and therefore the transformation took time. In many cases, as shown in Figure 6, point 60 (x,
y), mesh points 61 to 64 of the image around the point 60(x, y) must be read out, and it takes time to read these out. In addition, [X] in the figure is
is a Gaussian symbol representing an integer not exceeding .

Furthermore, since images before conversion are read out in random order as described above, there are problems such as the need for expensive random access memory.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image information processing device that eliminates the drawbacks of the conventional example described above, can be easily implemented in hardware by reducing memory on the input side, and speeds up image conversion processing. .

[Means for Solving the Problem] As a means for solving this problem, for example, the image information processing apparatus of the embodiment shown in FIG. a data processing device δ2 as an interpolation means for interpolating pixel data, an image conversion device 50 as a conversion means, a parameter memory 27 for storing a second parameter, and an image memory as a storage means. 3.

[Operation] In the configuration shown in FIG. 1, input raster data 1
2 is stored in the raster buffer 5, and the control device 4 sets parameters corresponding to the raster data in the parameter memories 11 and 12. The data processing device 2 interpolates the raster data 12 according to the contents of the parameter memory 11. - In the image memory 3, the drawing data 32 converted by the image conversion device 50 is written into the memory array 26 indicated by the address counter 28, 29 operating on the basis of the parameter memory 27;

Embodiment An image information processing apparatus according to an embodiment of the present invention will be described in detail below with reference to the accompanying drawings.

[Block diagram of image information processing device (Fig. 1)] Fig. 1 shows an embodiment of the present invention, in which 1 is a reading sensor that reads a document;
2 is a data processing device; 3 is an image memory; 50 is an image conversion device that performs affine transformation on image data; and 4 is a control device that includes a CPU 4-1 and a ROM 4-2 containing control programs and the like; Controls the entire device. Image data that has been raster-scanned, sampled, and digitized by the reading sensor 1 is sent to the data processing device 2 one main scan at a time.

In the data processing device 2, the image data is subjected to at least two main scans, and according to the parameter memory 11 written by the control device 4, the area sandwiched between the previous two main scans is covered after being affine transformed. Process all pixels of the converted image. The pixel data 21 processed by the data processing device 2 is transferred to the image conversion device 50
converted by Converted pixel data 32 is written into image memory 3 according to parameter memory 27 written by control device 4 . Note that data is written into the parameter memory 11 and the parameter memory 27 by the control device 4 every main scan.

The data processing device 2 will be explained in detail below.

[Description of Data Processing Apparatus (Fig. 2) (Fig. 3) Fig. 2 is a detailed diagram of the data processing apparatus 2 of this embodiment.

5 is a raster buffer that stores multiple main scanning data; 6-
9 is a latch circuit that latches the pixel before conversion, 10 is an interpolation circuit, and 11 is a parameter memory.

FIG. 3 shows the relationship between the mesh of the image before conversion and the mesh of the image after conversion. In this figure, lattice points X1...x6 drawn with solid lines are the meshes of the image before conversion, and lattice points ul...US drawn with broken lines are the meshes of the image after conversion.

In FIG. 3, each grid point represents the center of a pixel.

The explanation will be given below according to FIGS. 2 and 3.

Raster data 12 of the image digitized by the reading sensor 1 is stored in a raster buffer 5. Two pieces of raster data stored in the raster buffer 5 and made readable, namely, the first main scanning data 15 and the second main scanning data.

It is latched into data latches 8 and 6, and further latched into data latches 7 and 9 in the next cycle. The data latched in the data latches 6.7 and 8.9 at this time will be specifically described.

In FIG. 3, the first main scan is X1-x2-X5. The second main scan is from X3 to X4 to x6.

The second main scan is scanned temporally later than the first main scan, and X2 is a pixel transferred temporally later than Xl. Now, data latch 6 has pixel
Data of pixel x3 is stored in data latch 7. Similarly, data latch 8 has x2. The data latch 9 stores xl.

These latched data 17, 18, 19° 20 are input to the interpolation circuit 10. On the other hand, parameter memory 11
From lattice point u3 of the image after conversion (FIG. 3) and lattice point X1+x2+X3 of the image before conversion. Parameters 24 such as weights determined by the control device 2 based on the positional relationship with X4 are input to the interpolation circuit 10.

The interpolation circuit 10 extracts image data 21 corresponding to the grid point u3 from the image data 17, 18, 19° 20 and the parameter 24.
is calculated and output to the image memory 3. Interpolation circuit 10
The interpolation method by
An arithmetic average of corresponding pixel data using a reciprocal of the distance from 4 as a weight can be adopted, and in this case, the weight can be set as the parameter 24. For example, in the interpolation circuit 10, from the coordinate u3 after transformation to each coordinate X+ lX2 before transformation,
The pressure paths up to X3 1X4 are connected to I+, 17 . 13
．． 14, and the pixel density at the coordinate position is expressed by the symbol [
], the arithmetic averaged pixel density is [u 3
1= ([x+]・P(1+)+hr2]・P(12)+
[xal・P(h)×[x41・P(14)1 =(Phi[x+]+P2・[x2]+P3・[x31+
Pa・[xal lwhere P() is the weight function and 21
~P4 are parameters actually output from the memory 11.

The data latched in the data latches 6, 7, 8°9 in the next cycle are X6, X4, and X4 in FIG. 3, respectively.
X5 and X2, the parameter 24 output from the parameter memory 11 is also updated to an appropriate value, and the image data corresponding to the grid point u2 is output.

Thereafter, image data is similarly calculated for grid points of the converted image included between the first main scan and the second main scan, and is output to the image memory 3. When all the calculations for the first and second main scans are completed and the data for the third main scan, which includes point X7 in FIG. The contents are also updated, and the same processing sequence as described above is performed for the second main scan and the third main scan, and this continues until the last one scan.

This time order is based on the contents of the parameter memory 11, and in the control device 4 in FIG. 2, Xl.X2. X3. X
4. From the coordinate position (U, V) of the image after conversion in step 4, move to the summer mountain.

There is no need to use floating point arithmetic in the actual calculation; fixed point arithmetic is sufficient.Furthermore, since the calculation involves moving to the next grid point in the main scanning direction, only addition/subtraction or table reference is required. minutes and is fast. Furthermore, since this parameter has a certain period, even faster processing is possible by patterning it.

Next, the image memory 3 in FIG. 1 will be mainly explained.

Description of Image Memory (Fig. 4) (Fig. 5)] Fig. 4 is a detailed diagram of the image memory section of the embodiment according to the present invention.

In FIG. 4, 25 is a buffer memory for temporarily storing data from the data processing device, 26 is a memory array, and 2
7 is a parameter memory that holds parameters created by the control device 4 for one main scanning, and 28 is a main scanning address counter that generates addresses in the main scanning direction. 29 is a sub-scanning address counter in the sub-scanning direction, and 30 and 31 are starting value address registers for setting starting values in memory address counters 28 and 29, respectively.

The operation will be explained below according to FIG. The image conversion device 50 described above outputs the converted image slip data 32 corresponding to the start of a certain main scan, and stores it in the backup memory 25. At the timing when this pixel is transferred from the buffer memory 25 to the memory array 26, 5. The main scanning address counter 28 and the sub-scanning address counter 29 take in the starting value 39.40 from the starting value address register 30.31, and according to the instructions from the parameter memory 27, address in the memory array 26 corresponding to the address data 37.38. Write pixel data 33 at the position.

At the timing when the next pixel data 33 is transferred from the backup memory 25 to the memory array 26, the main scanning address counter 28 and the sub-scanning address counter 29 receive and take the control parameters 35 and 36 from the control parameter memory 27, and increase or decrease the counters. In response to the decrement or neither operation, the correct memory location address is generated and pixel data 33 is written to memory array 26.

This movement will be explained using FIG. In Figure 5,
The grid indicated by solid lines is the grid representing pixel positions before conversion, and the grid drawn by broken lines is the grid of pixels after conversion. Now, the processing for the area indicated by the arrow is calculated by the data processing device 2 described above, and is transferred to the buffer memory 2.
Suppose that it is stored in 5. Assuming that it is 1 in FIG. 5 that corresponds to the start of the main operation before conversion, the start value address register 30.31 contains the main scanning component h1. A sub-scan component 15 is stored by the controller. That's a signal! ! 42.43.

When vl is written to the memory array 26 and V2 is written, a signal 35 is sent from the control parameter memory 27 to the main scanning address counter 28 instructing it to be incremented by 1.
However, a signal 36 instructing not to increase or decrease is output to the sub-scanning address counter 29, the counter 28 increments, and the counter 29 sends the same value as the address 37.38 of the memory array 26. As a result, item (2) in FIG. 5 is accessed.

Below, V3. V4. By sequentially outputting the increase/decrease parameters for main scanning and sub-scanning as V5..., the correct writing position can be known. The contents of this parameter memory 27 are created in the control device 4 based on the first equation and the second equation, and are stored in the parameter memory 27.

Similar to the calculation described above, this calculation can be realized using fixed-point arithmetic, and no multiplication is required. Furthermore, since this parameter also has periodicity, it is possible to store patterns and perform batch processing. Further, the parameters stored in the starting value address registers 30 and 31 are also calculated by the control device.

[Explanation of operation flowchart of rj control device (Fig. 1) (Fig. 7)] Fig. 7 shows an operation flowchart of the control device. Parameters are determined based on conversion information and the like, and are set in the parameter memory 11 in step S2. Similarly, parameters are also set in the parameter memory 27 in step S3. In step S4, image conversion information is provided to the image conversion device. Next, in step S-5, it is checked whether data input in one main scanning direction has been completed, and when it is finished, it is checked in step S6 whether the last line has been read, and if it is not the last line, the process returns to step S1 and the parameters are reset. I do. This operation is repeated sequentially in the first and second main scans until the final line is read out.

The image is transformed and stored in the memory array 26 according to the procedure described above. The image in the memory array 26 is transferred 34 to an output device (not shown) and used for hard copying or the like.

(1) In the above embodiments, a reading sensor was used as the input device, but this may be a recording device such as a magnetic disk or magneto-optical disk, or a network communication device.

(2) Using multiple data processing units in parallel.

By using multiple control devices in parallel, the parameter memory 11
, 27, it is possible to improve the intended processing speed.

(3) When writing to the memory array 26, the parameter memory 27 and the starting W address 30.31 are used to create the desired deformed image on the memory array 26 as a bit image. The data output to the output device is structured so that it becomes the desired deformed image, the processing is distributed, and the apparent processing speed is improved.

"Effects of the Invention" As described above, according to the present invention, it is possible to transform raster data without having a random access memory on the remote side, and the circuit can be configured only by addition and subtraction. This has the effect of increasing speed.

[Brief explanation of drawings]

FIG. 1 is a schematic diagram of the configuration of an image information processing apparatus according to an embodiment of the present invention. FIG. 2 is a block diagram of the data processing device. Figure 3 is a diagram showing the relationship between coordinate meshes before and after local transformation for interpolation, Figure 4 is a block diagram of the image memory device, and Figure 5 is the relationship between coordinate meshes before and after global transformation for address transformation. 6 is a diagram showing the relationship between the image coordinate mesh before conversion and the image coordinate mesh after conversion, and FIG. 7 is a flowchart of parameter setting of the control device. In the figure, 2... data processing installation, 3... image memory,
4...Control device, 5...Raster eight noise, 6-91
4. Latch circuit, 10... Interpolation circuit, 11.27...
・Parameter memory, 26...Memory array, 28...
・Main scanning address counter, 29...Sub-scanning address counter, 30.31...Starting value address register, 5
o... Image conversion circuit. Patent applicant: Canon Co., Ltd. f:', L Agent: Patent attorney Yasunori Otsuka ・・・°,
21 Figure 1 Figure 2

Claims (2)

[Claims] (1) A line memory that stores at least two lines of raster data, an interpolation means that interpolates the corresponding pixel data in the line memory according to a first parameter, and a conversion that converts the pixel data interpolated by the interpolation means. An image information processing apparatus comprising: means; and storage means for storing pixel data converted by the converting means according to a second parameter. (2) The first parameter provides interpolation information of the pixel data converted by the conversion means, and the second parameter indicates a storage address of the converted pixel data. The image information processing device described in .