JPH0624020B2 - Method for detecting filled area in digital image - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial field of application> The present invention relates to a digital image filled area detection method for efficiently extracting a filled portion in an image from given digital image data.

<Conventional technology> OA (Office Automation), CAD (Computer Aide)
Due to the spread of d Design) etc., research and development for drawing recognition for the purpose of automatic reading of drawings have been actively conducted recently.

In the drawing recognition, it is general that the drawing data as a digital image is once converted into numerical data such as a vector structure and then analyzed. At this time, in a drawing typified by an electric circuit diagram, as shown in FIG. 13, black pixels such as a black circle a indicating a branch point of a normal line figure portion and a triangle portion b forming a part of a diode are displayed. Since the filled parts are mixed,
It is necessary to efficiently separate and extract both in advance.

It is known that a conventional method of extracting a filled-in portion of a digital image is based on a processing procedure as described below as a method involving thinning processing. That is, with respect to the original digital image as shown in 14 (a), it is described in documents such as "Considerations regarding thinning method"(Tamura; IEICE technical research report, PRL75-66, 1975). Using the Hildith thinning method, the thinning is performed until the standard line width becomes a figure of 1 pixel, and the thinned image shown in FIG. 14B in which the line figure part and the filled area are mixed is obtained.

Contour point extraction processing is performed on the obtained thinned image using a mask pattern of 3 pixels × 3 pixels,
The contour point extraction image shown in FIG. 7C is generated separately from the thinned image.

A line segment is traced on the obtained contour point extraction image to create vector data.

All the loops l _{1 to} l ₅ are detected by tracing the vectors from the positional relationship between the vectors.

All loops l ₁ to detected on the contour point extraction image
For l ₅ , the presence or absence of black pixels in each loop at the corresponding position on the thinned image shown in FIG. 14 (b) is checked to determine whether it is a filled area.

In accordance with the above processing procedure, the filled area in the original digital image is extracted.

<Problems to be Solved by the Invention> However, in the above-described conventional method, it is necessary to check the presence or absence of black pixels in the loop for all loops on the image. Therefore, loops such as gates and flip-flops are configured. It is troublesome for a drawing such as an electric circuit diagram in which a large number of symbols to be filled are included compared to the filled area, and the filled area cannot be efficiently extracted.

Further, in the conventional method, for holding the thinned image,
In addition to the existing image memory, it is necessary to provide another image memory for temporarily storing the contour point extraction image, which is not preferable because the hardware increases accordingly.

The present invention has been made in consideration of the above disadvantages, and in the thinned figure in which the line figure part and the filled area are mixed, which is obtained by performing an appropriate thinning process on the original image,
An object of the present invention is to realize a method capable of directly and efficiently extracting a filled area without detecting a loop.

<Means for Solving the Problem> In the present invention, which solves the above problem, a pattern for detecting a filled area of 3 pixels × 3 pixels is held, and this pattern is compared with the current mask pattern to determine the filled area. The detection method is as follows.

That is, an image memory for storing digital image data in which a line figure portion and a filled area obtained by thinning the target drawing are mixed, a pixel value of a target pixel on the image memory, and 8 pixels in the vicinity thereof. An address control unit that performs a read access for reading a value and a write access for writing a pixel value to a specified target pixel, and a pixel value of the target pixel read from the image memory and its vicinity 8
A serial-parallel conversion circuit that sequentially takes in serial data corresponding to a mask pattern of 3 pixels × 3 pixels, which is a pixel value, and converts the serial data into an address signal string, and 1 according to the address signal string.
A data table memory for generating a bit data signal, a plurality of registers and an arithmetic unit are provided to refer to the serial-parallel conversion circuit and the data table memory, and the coordinates of the pixel of interest to be tracked next. In a digital image filled area detecting method for detecting a filled area of a digital image, the digital image data is a mask of 3 pixels × 3 pixels. When the contents of this mask pattern match a predetermined fill area detection pattern by sequentially performing raster operations using the pattern, (a) the pixel of interest in the mask pattern is set as the fill boundary starting point, and The first black pixel is checked by checking the pixel value counterclockwise or clockwise with the pixel to the left of the pixel of interest as the search start point. A step of tracking pixel, (b) a mask of 3 × 3 pixels around the tracking pixel
When a pattern is set and the mask pattern coincides with the pattern for detecting the filled area, the tracing pixel is set as the filled boundary point and the previous filled boundary point is set as the search start point, and (c) When the mask pattern does not match the filled area detection pattern, the tracking pixel is used as a search start point, and the procedure (b) is repeated using the previous filled boundary point as a tracking pixel. This is a method for detecting a filled area of a characteristic digital pixel.

Here, the filled area detection pattern is a mask of 256 types of 3 pixels × 3 pixels in which the central pixel is a black pixel.
A mask pattern is obtained by excluding (a mask pattern in which the number of peripheral pixels of the central pixel is 2 or less and the number of peripheral pixels is the same as the number of 8-connected pixels) from the pattern.

<Operation> In the method for detecting a filled area of a digital image according to the present invention, when an image memory having a thinned image is raster-scanned with a mask pattern of 3 pixels × 3 pixels and a preset filled area detection pattern is obtained. The boundary tracking operation is started. Then, a search start point is set in this mask pattern, the master pattern is examined counterclockwise or clockwise, and black pixels are extracted as tracking pixels. A mask pattern of 3 pixels × 3 pixels is set with the tracking pixel at the center, and it is checked whether or not the pattern matches the pattern for detecting the filled area. If they match, the tracking pixel is set as the filled boundary point. If they do not match, the processing is continued with the current tracking pixel as the search start point and the immediately preceding filled boundary point as the tracking pixel.

<Embodiment> FIG. 1 is a structural block diagram showing an example for specifically carrying out the method of the present invention.

This figure shows a general configuration of a digital image processing apparatus, which is a main part of an image memory 1, an address control unit 2, a serial / parallel conversion circuit 3, a data table memory 4, a boundary line tracing unit 5, and a buffer memory 6. It is composed of The method of the present invention is stored as a processing algorithm in the boundary tracking unit 5.

Here, the operation of each unit in this image processing apparatus will be briefly described.

The image memory 1 is a pixel memory for storing digital image data, and prestores digital image data obtained by performing appropriate thinning on a target drawing and in which line drawing parts are mixed with regions.

The address control unit 2 is a block that performs read / write access to the image memory 1 at high speed according to an instruction from the boundary line tracking unit 5. Further, when the X and Y coordinates of the pixel of interest on the image memory 1 are given from the boundary line tracking section 5, in the read access, as shown in FIG.
The pixel value of the target pixel R specified by the Y coordinate (black or white,
“1” or “0”) and the values of 8 pixels in the vicinity of the pixel R shown in FIG. 3B are set to the pixels R ₀ , R ₁ , R ₂ , R ₃ , R ₄ , R.
₅ , ₅ , R ₆ , and R ₇ are successively read from the image memory 1 in order, and in the write access, the pixel value “0” or “1” is written to the target pixel R designated by the X and Y coordinates.

The serial-parallel conversion circuit 3 sequentially takes in 9-bit serial data corresponding to the mask pattern of 3 pixels × 3 pixels read from the image memory 1 and converts it into a 9-bit address signal sequence.

The data table memory 4 is a 9-bit address output AD (# 000 to # 1FF; # is 1) of the serial-parallel conversion circuit 3.
A one-bit data signal C of "0" or "1" which is predetermined according to the hexadecimal notation is generated.

A plurality of registers border following section 5 _RP S, RF, _RP
E , RP _i-1 , RP _i , RQ _j, and an arithmetic unit,
This is a block that executes the overall control and the extraction algorithm of the filled portion. Further, it refers to the data from the serial-parallel conversion circuit 3 and the data table memory 4 to determine the X and Y coordinates of the pixel of interest to be tracked next and generate the coordinates of the boundary line of the filled area.

The buffer memory 6 stores temporary point data necessary for the extraction processing of the filled portion in the boundary line tracing section 5 and the coordinate point sequence of the boundary line of the filled area sequentially extracted.

By the way, for the graphic object of the present invention, an appropriate thinning process is performed in advance until the line part of the original image becomes a line graphic.
Connected line figure part (Pixel 8
4) in which filled areas are mixed.
The digital image data is as shown in the figure. In this figure, the area indicated by the mesh line is the filled area.

Next, before specifically describing the method of the present invention, the principle of the method of the present invention will be described.

The outline of the method of the present invention is, as described in <Measures for solving problems> and <Operation>, as described above, the image memory 1 storing the thinned image in which the 8-connected line figure portion and the filled area are mixed. Is sequentially raster-scanned using a mask pattern of 3 pixels × 3 pixels, and when the content of this mask pattern matches a predetermined filled area detection pattern, the center pixel of this mask is used as a starting point, for example, Boundary line tracking is performed pixel by pixel using 8-connection using a mask of 3 pixels x 3 pixels described in documents such as "Computer Image Processing" (Yasuin, Nakajima; Sangyo Shuppan, 1979), and a filled area is displayed each time. Only the tracking pixels that match the content of the detection pattern are sequentially extracted as the boundary points of the filled area.

Here, the contents of the filled area detection pattern are defined.

First, in the mask pattern of 3 pixels × 3 pixels as shown in FIG. 3B, the central pixel R is a black pixel (R =
"1"), "On topological properties of sampled binary figures" (Yokoi, Shimawaki, Fukumura;
The following values N _s and N _c are calculated according to the conditions described in IEICE Transactions D, 56-D, 11, 1973).

The total number of black pixels around the central pixel R, that is, the number of peripheral pixels N
_S And further, 3 pixels generated by erasing the center pixel R ×
The number of connected black pixel components in the mask of 3 pixels, that is, the number of connected 8 N _c Calculate as However, R ₈ = R ₀ , ▲ ▼ = 1-R
_{Let k} .

Then, among the mask patterns of 3 pixels × 3 pixels, those in which the central pixel R is a black pixel (R = “1”) are classified as follows.

(A) The number of peripheral pixels N _s and the number of 8-connections N _c are the same value (N _s = N _c ) (B) The number of peripheral pixels N _s and the number of 8-connection N _c are different values (N _s ≠ N _c ) The condition (A) is classified into five ways of eight connection numbers N _c = 0,1,2,3,4 as in the example shown in FIGS. 5 (a) to (e). That is, when the number of 8-connections N _c = 0 (a), the central pixel R is the arc starting point, and when the number of 8-connections N _c = 1 (b), the central pixel R.
Is an end point, when the number of 8 connections N _c = 2 (c), the central pixel R is a connection point, when the number of 8 connections N _c = 3 (d) is a center pixel R is a branch point, and the number of 8 connections N _c = 4 In the case of (e), the central pixel R becomes an intersection. When the number of 8-connections N _c = 0,1,2, the central pixel cannot form a filled area, and when the number of 8-connections N _c = 3,4, it is considered that the areas are concentrated at one point, and it is regarded as a filled area. Shall be treated.

The condition (B) is, as in the example shown in FIGS. 6A to 6C, (a) 8 number of connections N _c = 1 (2 ≦ N _s ≦ 7), (b).
8 number of connections N _c = 2 (3 ≦ N _s ≦ 6) (c) 8 number of connections N _c
= 3 (4 ≤ N _s ≤ 5). That is, these means a state in which there is one or more black pixel connected components around the central pixel, and it is considered that the central pixel constitutes a part of the filled area or the branch point.

From the above consideration, the pattern for detecting the filled area is defined as follows.

That is, the filled area detection pattern is defined as 256 (3 × 3) mask patterns in which the central pixel R is a black pixel, and the pattern (C) “the peripheral pixel number N _s is 2 or less and the peripheral pixel number N _c” is set. 8 that have the same number of connected N _c ”.

Specifically, as shown in FIG. 7, there are a total of 25 mask patterns that satisfy the condition (C), and therefore there are 256-25 = 231 patterns for the filled area detection.

Next, when the image memory 1 is sequentially raster-scanned using a mask pattern of 3 pixels × 3 pixels, and the contents of this mask pattern match the pattern for detecting the filled area and the filled boundary point Q _m is extracted. , Next filled boundary point Q
A method procedure for extracting _{m + 1} will be described with reference to FIGS. 8 and 9.

(A) As shown in FIG. 8A, a mask of 3 pixels × 3 pixels is set with the boundary point Q _m as the central pixel.

(B) Next, it is checked whether the black pixels or white pixels ,, ... sequentially from the previous boundary point Q _m-1 in the counter-clockwise boundary point Q _m, finds the first black pixel, the point Be a tracking point P _n .

(C) As shown in FIG. 8 (b), a mask of 3 pixels × 3 pixels is set around the tracking point P _n, and it is checked whether or not this mask pattern matches the filled area detection pattern. .

(D) If the mask pattern centered on the tracking point P _n matches the content of the filled area detection pattern, the eighth
As shown in FIG. 6C, the tracking point P _{n is set} to the next boundary point Q.
_{Let m + 1} .

(E) On the other hand, as in the case of the tracking point P _n shown in FIG. 9A, when the mask centering on the tracking point P _n does not match the content of the filled area detection pattern, the boundary point is again detected. Returning to the mask centered on Q _m , this time, as shown in FIG. 9B, this time from the tracking point P _{n in the} counterclockwise direction, ...
The black pixel or the white pixel is checked in this order, and the first detected black pixel is set as the next tracking point P _{n + 1} .

(F) Next, as shown in FIG. 9 (c), the tracking point P _{n + 1}
When a mask of 3 pixels × 3 pixels is set centering on, and the mask pattern matches the pattern for detecting the filled area, as shown in FIG. 9 (d), the tracking point P _{n + 1 is set} to the next boundary point. Let Q _{m + 1} .

(G) When the mask pattern centered on the tracking point P _{n + 1} does not match the pattern for detecting the filled area, the mask pattern centered on the boundary point Q _m is returned again, and counterclockwise from the tracking point P _{n + 1.} Check whether it is a black pixel or a white pixel,
The first black pixel is set as the next tracking point, and the same operation as described above is repeated.

In this way, the filled boundary points are sequentially extracted.

The above operation has described the case where the boundary points of the filled area are extracted in the counterclockwise direction. However, when extracting the boundary points in the clockwise direction, as shown in FIG. 10, in the mask centered on the point Q _m , Clockwise from the previous boundary point Q _m-1 ,
The black pixels and the white pixels are examined in this order, and the first black pixel is set as the tracking point P _n .

The above is the principle of the method of the present invention. Next, using the apparatus shown in FIG. 1 for the digital image data shown in FIG. 4, refer to the flowchart of FIG. 2 and the procedure explanatory diagram of FIG. The method of the present invention will be specifically described.

In the data table memory 4 shown in FIG. 1, as shown in FIG. 12, the 3 × 3 mask pattern (FIG. 12 (a)) read from the image memory 1 is input data. 9-bit address “R (MSB), R ₀ , ..., R ₇ (LSB)” converted by the serial-parallel conversion circuit 3 as
Data "0" or "1" corresponding to (Fig. 12 (b))
Is stored. At this time, "1" indicates that the corresponding mask pattern matches the pattern for detecting the filled area, and "0" indicates that it does not match (FIG. 12 (c)).

In addition, the address corresponding to the value “0” includes the one in which the central pixel R is the white pixel (R = “0”) in the 3 × 3 mask pattern, and has 256 kinds of addresses # 000 to # 0FF. There are a total of 281 addresses including 25 mask patterns shown in FIG. Therefore, the address having the content “1” is the remaining 231 addresses.

Now, the method procedure of the present invention will be described with reference to the flowchart shown in FIG.

(1) The boundary line tracking unit 5 sequentially gives pixel coordinates in a digital figure as shown in FIG. 4 to the address control unit 2 for read access, and rasterizes the pixel memory 1 using a 3 × 3 mask. To scan. The 9-bit serial data “R, R ₀ , ..., R ₇ ” of the 3 × 3 mask pattern read out from the image memory 1 is once taken into the serial-parallel conversion circuit 3 and sequentially stored in the data table memory 4. Will be the address to. Then, the content of the mask pattern is checked each time with the output data C from the data table memory 4.

(2) If the output data C = “0”, that is, the mask pattern does not match the pattern for detecting the filled area, the procedure (1) is repeated.

(3) As shown in FIG. 11 (a), output data C =
“1”, that is, when the mask pattern M ₁ matches the pattern for detecting the filled area, the central pixel R of the mask M ₁
Is set as the boundary starting point P _s of the filled area, and its coordinates are stored in the register RP _s and the buffer memory 6 of the boundary line tracing unit 5.

(4) The boundary line tracking unit 5 uses the starting point P _s (= center pixel R)
The pixel R ₄ adjacent to the left of is the search start point P _F , this coordinate is stored in the register RF _F, and it is checked from the output data of the serial-parallel conversion circuit 3 whether it is a black pixel “1” or a white pixel “0”.

(5) When the pixel R ₄ is a black pixel, the pixel R ₄ is set as the end detection point P _E and its coordinates are stored in the register RP _E ,
Move to step (7).

(6) When pixel R ₄ is a white pixel (in the case of FIG. 11 (a))
Is clockwise (counterclockwise from the search start point P _F (pixel R ₄ ) specified by the register RP _F in the mask of 3 pixels × 3 pixels centered on the starting point P _s. In 8), the pixel is checked in the counterclockwise direction as the clockwise direction), the first black pixel is set as the end detection point P _E, and its coordinates are stored in the register RP _E.

(7) The starting point P _s is set as the previous tracking center pixel P _i−1 and its coordinates are stored in the register RP _i−1 .

(8) In the 3 × 3 mask (corresponding to the mask M ₁ in FIG. 11 (a)) centered on the previous tracking center pixel P _i−1 specified by the register RP _i−1 , R
Pixels are checked counterclockwise from the search start point P _F specified by P _F , and the first black pixel is the current tracking center pixel P _i (= P
₁ ) and store the coordinates in the register RP _i .

(9) Whether the previous tracking center pixel P _i-1 and the end detection point P _E , and the current tracking center pixel P _i and the starting point P _s match, are registered in the registers RP _i-1 and RP _E , The contents of the register RP _i and the register RP _s are compared to detect whether or not the tracking process is completed.

(10) The previous tracking center pixel P _i−1 and the end detection point P _E ,
If the current tracking center pixel P _i and the starting point P _s match, the procedure moves to step (18).

(11) Previous tracking center pixel P _i−1 and end detection point P _E ,
When the current tracking center pixel P _i and the starting point P _s do not match, the boundary line tracking unit 5 gives the coordinates of the current tracking center pixel P _i specified by the register RP _i to the address control unit 2 to read. Make access. Next, as shown in FIG. 11B, a 3 × 3 mask pattern M ₂ centered on the point P _i (= P ₁ ) is read out from the image memory 1 and taken into the serial-parallel conversion circuit 3, Depending on the value of the output data C from the data table memory 4, this mask pattern M ₂
Examine the contents of.

(12) If the output data C = “0”, that is, the mask pattern M ₂ does not match the filled area detection pattern, proceed to step (16).

(13) Output data C of mask M ₂ shown in FIG.
When “1”, that is, the mask pattern M ₂ matches the pattern for detecting the filled area, the current tracking center pixel P _i (P ₁ ) designated by the register RP _i is set as the boundary point Q _j of the filled area, The coordinates are stored in the register RQ _j and the buffer memory 6.

(14) In the mask M ₂ , the previous tracking center pixel P
_i-1 (corresponding to _{P s)} and the search start point _{P F,} and stores the contents of register _{RP i-1} to the register RP _F.

(15) of the current track center pixel _P i a _{(P 1)} and the tracking center pixel _{P i-1} of the previous stores the contents of register RP _i in the register RP _i-1. Then, the procedure moves to step (8).

In this way, steps (8), (9), (11), (13), (1
4) and (15) are repeated, and the filled area boundary points P ₁ , P ₂ ,
To extract the P _3.

(16) In step (12), the output data C corresponding to the 3 × 3 mask pattern M ₂ centered on the point P _i (= P ₁ ).
Is “0”, that is, when the mask pattern M ₂ does not match the pattern for detecting the filled area, the tracking center pixel P _i (P ₁ ) at this time is set as the search start point P _F , and the register R
Storing the contents of the P _i in the register RP _F.

(17) Then, the boundary point Q _j (P _s ) of the immediately previous filled area
Be the previous tracking center pixel P _i−1, and register RQ _j
Store the contents of the register into register RP _i-1 . Register RP
The P _i-1 of the coordinate point specified by the _i-1 is given to the address control unit 2 performs a read access, the point from the image memory _{1 P i-1 (P s} ) 3 × 3 mask around the The content of the pattern (corresponding to the mask M ₁ ) is read out and taken into the serial-parallel conversion circuit 3. Next, the procedure returns to step (8).

Here, steps (12), (16), and (17) are steps for detecting the boundary between the line figure portion and the filled area in the digital image, and a specific example thereof is shown in FIGS. Will be explained.

In FIG. 11 (c), it is assumed that the pixels P ₁ , P ₂ , and P ₃ are points determined to be filled boundaries, and the black pixel P ₄ is extracted as the current tracking center pixel P _i .

When a mask pattern M ₄ of 3 pixels × 3 pixels centered on the pixel P ₄ is set, this mask pattern M ₄ does not match the filled area detection pattern, and the current tracking center pixel P ₄ is set next. A mask with the search start point P _F and the boundary point P ₃ of the immediately previous filled area as the tracking center pixel P _i−1.
Return to pattern M ₃ . When a black pixel is detected counterclockwise from the search start point P _F (P ₄ ), the pixel P ₅ is extracted.

Next, by setting the mask pattern M ₅ as the tracking center pixel pixel P _5, does not match the mask pattern M ₅ is filled area detection pattern, as shown in FIG. 11 (d), the current The tracking center pixel P ₅ is set as the search start point P _F, and the process returns to the mask pattern M _{3 in} which the immediately preceding filled boundary point P ₃ is the center pixel. Then, the black pixel P ₆ is centered counterclockwise from the pixel P ₅ . The mask pattern M ₆ centered on the black pixel P ₆ matches the filled area detection pattern, and the pixel P ₆ is regarded as the filled area boundary point.

By this sequence, examine its contents by setting the mask pattern M ₅ of a central pixel mask pattern M ₄ and the pixel P ₅ around the pixel P ₄ from the pixels P _3, pixel P ₃ is a digital It is determined that it is the boundary between the line figure portion of the image and the filled area.

Now, returning to the flowchart of FIG. 2, the description of the procedure will be continued.

(18) In step (9), the previous tracking center pixel P
_{When i−1} and the end detection point P _E match the current tracking center pixel P _i and the starting point P _s , for example, in FIG. 11 (e), the previous tracking center pixel P ₁₄ and the end detection point P E _{When E} is the same and the current tracking center pixel P ₁₅ and the starting point P _s are the same, the current tracking center pixel P _i (P ₁₅ ) is set as the boundary point Q _j of the filled area, and is specified by the register RP _i. Boundary point Q _{j of the} filled area with the current tracking center pixel P _i
(In this case, the coordinates of the pixel P ₁₅ are the same as the starting point P _s ) and the coordinates are stored in the buffer memory 6.

FIG. 11 (e) shows the tracking center pixel P _i (i = 1, ..., 15).
FIG. 11 (f) shows the tracking state of the boundary point Q _j (j = 0, ..., B) of the filled area obtained by the above operation.
9) is represented.

(19) The boundary line tracking unit 5 uses the coordinates of the boundary point sequence Q _j of the filled area stored in the buffer memory 6 to define the boundary line of the point sequence Q _j on the image memory 1 and the pixels inside the boundary line. Are all converted from black to white.

The black pixel / white pixel conversion process is a process performed so that the boundary point Q _j of the filled area once obtained by the current operation is not detected again in the next raster scan. The boundary of the filled area that is drawn is regarded as a polygon, and for example, "Fundamentals of Interactive Computer Graphiccs"
(J. D. Foley / A. Van Dam; Addison-Wesley, 198.
It is performed by using the polygonal filling method used in computer graphics described in the literature such as 2).

(20) The boundary line tracking unit 5 checks whether or not the raster scanning of the 3 × 3 mask pattern in the image memory 1 is completed by the coordinates of the point P _s designated by the register RP _s .

(21) If raster scanning has not ended, the procedure returns to step (1), and raster scanning is performed from the point P _s on the image memory 1.

(22) When the raster scanning is completed, the extraction processing of the filled area is completed.

As described above, the image memory 1 storing the digital image data (thin line figure) in which the 8-connected line figure part and the filled area are mixed is sequentially raster-scanned using the mask of 3 pixels × 3 pixels, and this mask When the pattern content matches a predetermined pattern for detecting a filled area, the boundary pixel is traced pixel by pixel using the mask of 3 pixels × 3 pixels starting from the center pixel of this mask and filling each time. Only the tracking pixels that match the content of the area detection pattern can be sequentially extracted as the filled boundary points.

<Effects of the Invention> As described above, according to the method for detecting a filled area of a digital image of the present invention, a boundary line of a filled area can be directly formed from a thinned figure in which eight connected line figure parts and filled areas are mixed. Can be extracted.

Further, according to the method of the present invention, it is sufficient to prepare only the image memory for holding the digital image data in which the 8-connected line figure portion and the filled area are mixed.
Further, it is not necessary to provide another image memory for work.

[Brief description of drawings]

FIG. 1 is a configuration block diagram showing an example of an image processing apparatus for specifically implementing the digital image filled area detecting method of the present invention, and FIG. 2 is a flow chart specifically showing the procedure of the method of the present invention. 3 (a) and 3 (b) are diagrams showing a pixel of interest R on the image memory and a mask pattern of 3 pixels × 3 pixels centered on the pixel of interest R, and FIG. 4 is a digital diagram of the present invention. Figure showing image data, No. 5
FIGS. 6A to 6E are diagrams showing an example of a mask pattern of 3 pixels × 3 pixels in which the number of peripheral pixels N _s and the number of 8-connected pixels N _c are the same (condition A), FIG. a)-(b)
Is a mask pattern of 3 pixels × 3 pixels in which the number of peripheral pixels N _s and the number of 8-connected pixels N _c are different values (condition B)
7 is a mask pattern of 3 pixels × 3 pixels in which the number of peripheral pixels N _s is 2 or less and the number of peripheral pixels N _s and the number of 8-connected N _c are the same (condition C). 8A to FIG. 8C and FIG. 9A to FIG.
(D) and FIG. 10 are diagrams showing a procedure for extracting the next filled boundary point when the filled boundary point is extracted, and FIG. 11 (a).
.About. (F) are diagrams showing state transitions when the method of the present invention is applied to the digital image data shown in FIG. 4, and FIGS. 12 (a) to (c) are serial-parallel conversion circuit 3 and data table. FIG. 13 is a diagram for explaining the function of the memory 4, FIG. 13 is a diagram showing a part of an electric circuit diagram as an object of the present invention, FIG.
14 (a) to 14 (c) are diagrams showing a case where the original image is subjected to a thinning process. 1 ... Image memory, 2 ... Address control unit, 3 ... Serial-parallel conversion circuit, 4 ... Data table memory, 5 ... Boundary line tracing unit, 6 ... Buffer memory.

Claims (2)

[Claims] 1. An image memory for storing digital image data in which a line drawing portion and a filled area obtained by thinning a target drawing are mixed, and a pixel value of a target pixel on the image memory and its vicinity. An address control unit for performing a read access for reading a value of 8 pixels and a write access for writing a pixel value to a designated pixel of interest, and a pixel value of the pixel of interest read from the image memory and its neighboring 8 pixels Serial-to-parallel conversion circuit that sequentially takes in serial data corresponding to the mask pattern of 3 pixels × 3 pixels, which is the value of, and converts it into an address signal sequence, and a data table that generates a 1-bit data signal according to the address signal sequence. It has a memory, a plurality of registers and an arithmetic unit, and should be traced next with reference to the serial-parallel conversion circuit and the data table memory. And a boundary line tracing unit for generating the coordinates of the boundary lines of the determination and fill regions of the coordinates of the target pixel,
In the method for detecting a filled area of a digital image for detecting a filled area of the digital image, the digital image data is masked with 3 pixels × 3 pixels.
When the contents of this mask pattern match a predetermined fill area detection pattern through sequential raster operations using the pattern, (a) the pixel of interest in the mask pattern is set as the fill boundary starting point and A procedure of checking the pixel value counterclockwise or clockwise with the pixel next to the left of the pixel as a search start point and setting the first black pixel as a tracking pixel, and (b) a mask of 3 pixels × 3 pixels centered on the tracking pixel.・
When a pattern is set and the mask pattern coincides with the pattern for detecting a filled area, the traced pixel is set as a filled boundary point, and the previous filled boundary point is set as a search start point, and (c) If the mask pattern does not match the filled area detection pattern, the tracing moth is used as a search start point and the procedure (b) is repeated using the preceding filled boundary point as a tracing pixel. A method for detecting a filled area in a digital image, characterized by. 2. The pattern for detecting a filled area is selected from 256 types of mask patterns of 3 pixels × 3 pixels in which a central pixel is a black pixel (the number of peripheral pixels of the central pixel is 2 or less and the number of peripheral pixels is 2. The method for detecting a filled area in a digital image according to claim 1, wherein the master pattern is a master pattern excluding a mask pattern having the same number of 8 connected pixels.