JPH01100679A - Contour extracting method - Google Patents

Abstract

PURPOSE:To process at high speed by taking out three horizontal picture elements consisting of a noticed picture element and both the side adjacent picture elements as a bit train and similarly converting three vertical picture elements to a horizontal successive bit train and taking out the train. CONSTITUTION:When an original cell input part 20 is activated by a control part 10, a cell divided into grid forms in a transfer origin part image 120 is transferred to an original cell buffer 140. Then, an original cell rotating part 30 is activated to rotate the original cell held by the buffer 140 by 90 deg. rightward and transfer to a cell buffer 150 after the rotation. Then, an original cell control part extraction contour part 40 is activated to output the horizontal and the vertical three picture elements as the bit train from the buffers 140, 150, refer to a table memory 180, obtain an target contour picture element value and output to a contour cell buffer 200. Then, a counter cell output part 50 is activated to transfer the contents of a buffer 200 to a prescribed contour cell position in a transfer destination memory 210. Thereby, a bit train access for taking out plural bits once can be attained to process at high speed.

Description

[Detailed description of the invention] [Industrial application field] The present invention can be read by a scanner or the like and can be read by a computer.

Image data held in memory (seen in a document).

illustrations, stamps, 9 handwritten letters, photographs, etc.).

Concerning a method for extracting the contours of. 15 [
Prior Art] A conventional device is described in Jikai BE 60-” 1osos No. 1.

As shown above, the density of each pixel in the original image is called a pixel.

Additionally, prior art devices of this type require bit accesses of the memory. 2゜ [Problems to be solved by the invention] The above-mentioned conventional technology is aimed at speeding up bit access.

There was a problem that no consideration was given to this, and that a large amount of processing time was required.

The object of the present invention is to provide a high-speed contour extraction method that uses 5° bit string access instead of the bit access described above.

It is about providing.

[Means for solving problems]

The above purpose is obtained by dividing the original image into a grid.

Original cell rose 1, 1 that holds a cell that is a small rectangle
After rotation, holding the cell rotated by 90 @.

cell buffer and the contents of the original cell buffer above.

Rotate 90″ and transfer it to the cell buffer after one rotation,
Cell rotation means and specific bits within the original cell buffer.

1st bit string to extract consecutive bit strings at once + 5 Retrieval means and identification within the cell buffer after one revolution.

This is achieved by providing a second bit string number extraction means for successively extracting the bit strings at once.

[Effect]

For contour extraction, three horizontal pixels (the left adjacent pixel of the pixel of interest,
It is necessary to access the pixel of interest, the pixel adjacent to the right of the pixel of interest) and the three vertical pixels (the pixel adjacent to the bottom of the pixel of interest, the pixel of interest, and the pixel adjacent to the top of the pixel of interest).

The three horizontal pixels can be extracted at a time by the first bit string number extraction means. Also,.

The cell rotation means includes bits that are continuous in the vertical direction.

Since it converts a column into a horizontally continuous bit string.

The above two vertical pixels are the second bit string number].

[by means], it can be taken out at once. follow,.

This allows bit string access and speeds up processing.

Ru.

〔Example〕

An embodiment of the present invention will be described below with reference to FIG.

Ru. FIG. 1 is a block diagram showing the overall configuration of an embodiment according to the present invention.

In FIG. 1, the transfer destination memory 110 contains images.

- The data is retained. Transfer destination memo above.

extracting the outline of a portion of the image data 120 in the memory 210 and transferring it to a portion of the destination memory 210;
3.

However, at this time, it is necessary to specify the following parameters.

■ sadτ: Transfer destination partial image start address.

■ zw: Horizontal axis (in bytes) ■ zskpl: Transfer destination address/skip length (in 5 totes) ■ #A = Transfer destination partial image height (raster unit).

position) ■ dadr: Partial image (extracted contour) transfer.

Starting address 1゜■ dzkpl
: Transfer destination address/skip length (pi.

Figure 2 (6) (A) K is an explanatory diagram of each of the above parameters.

show.

The outline of the partial image 120 is extracted by 10 degrees horizontally.

Buto. It is divided into square areas of 10 vertical rasters (this is called an original cell'), and the process is performed for each original cell. That is, in FIG. 2, the contour of the original cell CJ1 is extracted, this is transferred to the contour cell Cd1, the contour of the 9th order ICCs2 is extracted,
This is transferred to the contour cell Cd2. Henceforth. 4.

Below, process C8 → Cds, Cza → CtLa (
Principle 2 Contour cells move from left to right and top to bottom).

Now, the principle of one contour extraction will be explained here. The image is assumed to be black and white binary, and the contour pixels are selected from among the black pixels. Further, the color of one selected contour pixel is black, and the color of pixels other than one contour pixel is white.

do. Figure 3 (α) is an example of the original image from which contours should be extracted.

Figure (b) shows the extracted contour. It can be seen from Figure 3.

Like, 10 (1) A white pixel in the original image is one contour pixel.

do not.

(2) Among the black pixels in the original image, all four adjacent pixels on the top, bottom, left and right sides are black pixels.

In some cases, the black pixel is not treated as one contour pixel. , 5(3)
Among the black pixels in the original image, on top.

If there is a white pixel in any one of the four adjacent pixels on the bottom, left and right sides, that black pixel is treated as one contour pixel.

The above is an explanation of the principle of zero contour extraction. In the first part 0, the table memory 180 stores the above rules as a table. FIG. 4 shows an example of the configuration of the table memory 180. Figure (a) shows the pixel of interest and its four adjacent pixels.

bl or b5 is the pixel of interest, hl is the left adjacent pixel, b
5 is the two adjacent pixels on the right, b4 is the upper adjacent pixel, and h6 is the upper adjacent pixel. In the same figure (h), the table memory 18
In the configuration example of 0, it is composed of 8 bytes (=64 bits). Me.

What's in the sled?

It is. Please refer to the hatched parts.

There isn't, so it's optional. of the book table memory 180.

The reference method is to first use 5 bits of bl, bl, and hs.

The numerical value bytadr (0
<,.

hytadr≦7), bytes of tape A/-memory 180.

-7 deL/, K bytadr is calculated and the first order is 44.
A numerical value expressed by a combination of 3° bit values of As and A6.

The in-byte no-hit address hitadτ is determined by bitadr (Q≦hitadr≦7). Furthermore, the pit in the table memory 180 located at the above bytαdr2°hitadr is taken out. Now, hl,
bl, bs=1. 1. 1. ha-.

As, A6: 0. 1, 0. If so, bytadr.

=7. bitadr=2. Then, bit b (bit value=1) shown in outline in FIG. 4(A) will be extracted.

In FIG. 1, the table memory 1aoic is.

The values shown in FIG. 4(A) are set. Original cell...
Buffer 140 is the original of the destination partial image.

The cell buffer 150 for temporarily holding cells is the original cell buffer 150.

The original cell held in the file buffer 140 is shown on the right.

90@Cell for temporarily holding rotated cells.

It is a buffer. Also, one contour cell buffer 200 degrees is used to temporarily hold the extracted contours.

- It is a buffer.

The control unit 10 controls the entire transfer destination partial image 120.

This is the part that controls the extraction. The control unit 10 moves slightly.

10 is the original cell input section 20. Original cell rotating section 60.
.

Original cell contour extraction control section 401 contour cell output section 50 is set to "2"
0,7.

The processes activated in this order are repeated for the number of cells 150 in the transfer destination partial image 120. Figure 5 shows processing) and low.

When the original cell input unit 20 is activated, the corresponding cell in the transfer destination/partial image 120 is transferred to the original cell buffer 5a1.
Transfer to 40. FIG. 6 (α) shows the transfer destination memory 120.
The original cell inside is transferred to the original cell input section 20 in Ch) of the same figure.

Indicates the original cell that has been sent. Since 1 byte = 8 bits, each cell is arranged horizontally and vertically according to this number.

Both have the same 8-bit pitch, 10 bits, 10 las,
.

Input one by one. Therefore, there are some pixels that input the same pixel multiple times to the original 9-cell buffer 14 OK.

exists within the image. Hatching is shown in Figure 6 (α).

This is the pixel that was applied. Also, the width of the cell is 10 bits.

Since it is a capacitive layer of the original cell buffer 140.

Allocate 2 bytes per 1 raster of the original cell, 10 rasters.

20 bytes. The remaining 6 bits of 1 raster 2 byte (=16 bits) are not used. FIG. 7 shows a processing flow of the original cell input section 20. At block 2010, one rusk in the original cell. .

φ8・ (=2 bytes) are transferred continuously from one transfer destination to another.

At block 2020, the transfer destination start address csadr is determined.

(according to the following formula).

csadr = czadr −)−zzkpl +
(3W-2) -----...(1).

Further, in block 2050, all rasters (=5
10 rasters) has been transferred, and all rasters.

If it has not been transferred, the process returns to block 2010 and is completed.

If the cell has already been transferred, the original cell input process ends. .

When the original cell rotation unit 30 is activated, 50 is the original cell.

The original cell held in the buffer 140 is rotated 90 times to the right and transferred to the cell buffer 150 after 0 rotations.

In FIG. 8(α), the original cell is held in the buffer 140.

The arrangement of each bit of the original cell is shown in Cb) of the same figure.

The rotated cells are held in the rotated cell buffer 150.

This shows the arrangement of each bit in the file. This rotating section 50 is known.

method (for example, Japanese Patent Application Laid-Open No. 146367/1983)
The processing flow is not particularly described here.

When the original cell contour extraction control unit 40 is activated, the process starts.

The contour surface of the part of the original cell excluding one raster on the top and bottom and one dot on the left and right (the part shown in the outline in FIG. 8 (α)).

Outline the element (8 dots horizontally = 1 byte, 8 rasters vertically).

5 horizontal pixels to output to cell buffer 200.

Elementary output section 60. Vertical 3-pixel output unit 709 outputs a contour image.

Part 80. Shift register 902 contour byte output section 10
(1,,.

control the behavior of The processing flow is shown in FIG.

vinegar. The contour cell is 8 bits horizontally x 8 rasters vertically (64 bits in total).

), each of which has a bit value.

Find each bit. Scan from left to right, top to bottom.

Proceed. In the example of FIG. 8 (α), bit 12→13→・
・：・・・1゜→19→22→23→・・・・・・ Bits from bit 12.

Find the bit value of each contour pixel up to 89. Now, note.

If the pixel we are looking at is bit 12, then the image shown in FIG.

In block 4010, 3° pixels (bits 11.12.1g) from the left adjacent pixel of the pixel of interest are transferred to the original cell buffer 1.
40 to the bytαctr register 160. In block 4020, three pixels (bits 22.12.2) from the lower adjacent pixel of the pixel of interest are output from the rotated cell buffer 150 to the hit adr register 170. At this time.

The above three pixels are output from the cell buffer after rotation.

Do not output from the original cell buffer. The reason is that when outputting from the original cell buffer, bit access is required for three pixels (three times).

When outputting from the cell buffer after one rotation.

The target three pixels are arranged consecutively horizontally, and the five pixels (
Figure 8 (h>) Three pixels can be output at once.

This is for the purpose of At block 4060, the output has occurred.

The above hytadr, hitadr, yoshi, tape roof +-T-1J.

180, the target contour pixel value of 1 bit is obtained.

Ru. In block 4040, this contour pixel value of 1 pixel is taken into the shift register 90. block.

At 4050, the contour pixel values are stored in the shift register 90.

Determine whether 8 bits (=1 byte) have been formed.

If it is less than 8 bits, move the pixel of interest to the right.

- Advance by a cumulative number of pixels (if the pixel you are currently paying attention to is bit 121°, advance to bit 13), and go to block 4010 again.
Proceed to. When 8 bits of contour pixel values are formed.

Proceeding to block 4060, the contents of shift register 90 (
1 byte) to one contour cell buffer 200.

Further, in block 4070, 1 contour cell 8 ras, 11.

If the number of transferred rasters is less than 8 rasters, move the target pixel one pixel to the right (the current pixel is bit 19). If so, proceed to block 22) and proceed to block 4010 again. When the number of transferred rasters reaches 8 rasters, the process ends.

When the contour cell output unit 50 is activated, 50 is two contours.

The contents of cell buffer 200 are transferred to a predetermined contour cell location in destination memory 210°. Figure 10 shows the ring.

5 shows a processing flow of the cell output unit 50. Block 1. .

At 5010, 1 lanuta (1
゜bytes) are transferred from destination to destination and blocked.

At 5020, the transfer destination start address ctiad, r is determined by the following formula.

skip.

cd, atLr = cdadr + dzkpl
--・-・----・” (2) 1° Furthermore, in block 5050, it is determined whether all the tasks (8 rasters) within the contour cell buffer 200° have been transferred.

If all rasters have not yet been transferred, the process returns to block 5010°, and if all rasters have been transferred, the one contour cell output process 6 is completed.

・12・ The above is the explanation of the embodiment. In addition, in this embodiment, one outline extraction was explained, but the pixel values of the extracted outline and pixels are all white, and the pixel values other than the one outline pixel are transferred as they are from the original image. If you use K, you can narrow down the original image. 5 again. It is also possible to thicken the original image by inverting the black and white of the original image and applying the thinning process described above.

Needless to say, all of these can be realized simply by changing the contents of the table memory 180.

1゜[Effects of the Invention] As explained above, according to the present invention, each pixel.

- multiple bits at a time instead of bit accesses.

It is possible to access the bit string to be extracted and perform 1L processing.

This has the effect of increasing speed.　　　　　　1゜

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the overall configuration of an embodiment of the present invention. The lock diagram, Figure 2, is when extracting the outline of a partial image. An explanatory diagram of the specified parameters, Figure 3 is the source of contour extraction. Figure 4 is an explanatory diagram of the structure of the table memory. An explanatory diagram, FIG. 5 is a processing flow diagram of the control unit 10, and FIG. 6 is an input range of the original cell by the original cell input unit 20. After the original cell is input, it is held in the original cell buffer 140. An explanatory diagram of the original cell, #Iz diagram is the part of the original cell input section 20. Process flow diagram, Figure 8 shows the original cell and cell 5 after being rotated 90 degrees to the right.
An explanatory diagram of the bit arrangement, and Fig. 9 is an illustration of the original cell contour extraction. A processing flow diagram of the control unit 40, FIG. 10 shows outline cell output. 5 is a processing flow diagram of the unit 50. FIG. 10... Control unit, 20... Original cell input unit, 30...
・Original cell episode. conversion section, 40... Original cell contour extraction control section, 50... Contour cell. Output section, 60...Horizontal 3 pixels Output section, 70...Vertical 6 pixels. Output section, 80... Contour pixel output section, 90... Shift register. 100: Outline byte output section, 110: Transfer destination. MoIJ, 120...Transfer destination partial image, 130...
・Original cell. 140...Original cell buffer, 150...Cell after rotation, 5 buffer, 180...Table memory, 1
90... Carry flag, 200... Contour cell buffer, 210... Transfer destination memory, 220... Outline partial image, 250... Ring, 15. 2nd 3rd 13θ (b) Fig. (b) H tenth flat example + ← 6th ((IL) (b) " 1 + u Fig. 7 Fig. 8 (b) q 1F) to Fig.

Claims (1)

[Claims] 1. In the contour extraction method using four-pixel reference, in which the contour of a black-and-white binary image is obtained by referring to the pixel values of four adjacent pixels on the top, bottom, left, and right of the pixel of interest, a small rectangle obtained by dividing the original image into a grid pattern is used. An original cell buffer that holds cells that are rotated by 90 degrees, a rotated cell buffer that holds cells that have been rotated by 90 degrees, a contour cell buffer that holds cell outlines, and a cell that imports cells into the original cell buffer. an input means, a 90° cell rotation means for rotating a cell in the original cell buffer by 90° and transferring the rotated cell to the cell buffer; a pixel adjacent to the left of the pixel of interest, a pixel of interest, a pixel adjacent to the right of the pixel of interest; The pixel values of the three horizontal pixels are taken in at once from the original cell buffer horizontally, and the pixel values of the three vertical pixels of the pixel adjacent below the pixel of interest, the pixel of interest, and the pixel adjacent above the pixel of interest are After rotation, take in horizontally from the cell buffer at once, and from each pixel value of the above three horizontal and vertical pixels,
Contour extraction means for determining the pixel value of the pixel of interest and outputting it to the contour cell buffer, and contour cell output means for outputting the contents of the contour cell buffer, are provided, and the three vertical pixels are extracted from the rotated cell. - A contour extraction method characterized by capturing horizontally from a buffer at once and extracting contours of the entire original image by repeating contour extraction for each cell.