JPS6269387A - Method for extracting contour line of picture - Google Patents

Abstract

PURPOSE:To extract even the contour lines of a multi-value picture in the areas of the same color at a high speed and with high efficiency, by connecting raster segments to be detected accurately to one of the contour lines formed by the preceding scans and at the same time extracting and coding all contour lines with the screen scan of just a single time. CONSTITUTION:The raster segments 11-14, i.e., the pieces of contours obtained from a single main scan are coded into 0-3 by a 4-coupling direction code 15 with respect to a binary picture. Then raster segment detection is carried out together with the connection of raster segments, branch processing, meeting point processing and tracking end processing. This method can be applied easily even for a multi-value input picture by using the raster segment detecting process exclusive for multi-value pictures. In other words, the runs can be defined for pictures of the same color on a single scan line in the case of the multi-value input pictures.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Field of Application] The present invention relates to a method for extracting the outline of an image.

[Conventional technology]

The operation of converting conventional tester data into vector data will be explained with reference to FIGS. 9 to 11.

Figure 10 shows raster data vector data conversion (R-
41 is a 3X3 matrix pattern, each number flJ indicates the right and left pixels under the hill of the target pixel 41a, and each number "2" indicates the four diagonal directions of the target pixel 41a. Indicates a pixel.

FIG. 11 is a schematic diagram explaining the R-V conversion operation, and 5
1 indicates the original data (raster data), 52 indicates the above 3
The outline data (hatched area) extracted into the ×3 matrix pattern 41 is shown.

:fS FIG. 12 is a flowchart explaining the RV conversion operation. Note that (1) to (4) indicate each step.

First, the original data 51 is scanned using the 3×3 matrix pattern 41 to obtain the outline data 52 of the original blank image (1). Next, the start point of a black pixel of the contour line data 52 is picked up, for example, from the upper left using the 3X3 matrix pattern 41 (2) 0 Next, the contour line data 52 is picked up from the start point using the 3X3 matrix pattern 41 until it returns to the start point. , trace the outline, and set the traced outline pixels to white (3) 0. Next, determine whether raster extraction of the entire screen has been completed. 4) If YES, end the control;
If so, return to step (2).

In this way, the configuration is such that the outline of the raster data 51 can be extracted.

[Problem that the invention seeks to solve]

In this way, in the past, the original data 51 had to be scanned at random and the scanning had to be repeated until it returned to the starting point, which resulted in a problem in that it took a very long time to extract the contour line, significantly slowing down the processing efficiency. was there.

This invention was made in order to solve the problems described in -A.
It is an object of the present invention to provide a method for extracting contours of raster type images that can be extracted smoothly.

[L step to solve the problem]

The image contour extraction method according to the present invention includes a scanning line element detection process that detects a scanning line element corresponding to an image contour fragment within a scanning line width each time an image is horizontally scanned; A contour line element connection process that connects the scanning line element detected in the previous scan to the end of the scanning line element formed in the previous scan, a definition setting process that defines and sets a new contour line element, and two contour line elements. a contour line element group expression process that merges the two contour line elements and expresses it as a contour line element group, and an extraction process that determines whether or not a closed contour line element is formed according to the relationship between two contour line element groups and extracts a contour line. It consists of.

[Effect]

In this invention, each time an image is horizontally scanned, a scanning line element corresponding to a contour fragment of the image within the scanning line width is detected, and the detected scanning line element is used as the scanning line element formed in the previous scanning. Connecting the two contour elements to the end, merging the two contour elements to form a group of contour elements, and determining whether a closed contour element has been formed according to the relationship between the two groups of contour elements. - Extract the outline of the image. Furthermore, if the detected contour line element is a new one, the definition is set each time.

〔Example〕

FIG. 1 is a schematic diagram illustrating a method for extracting contours from a binary image, which is an embodiment of the present invention.

In this figure, 1a is a branch point, which branches into contour segments cs, , cs2. 1b is the branch point, contour segments C53, C5a
Branch into. 1c is a branch point, which branches into contour segments C55 and C56. 1d is a branch point, which branches to contour segment C57 and CSs. 2
a is the meet point (merging point) and contour segment C
32 and C33 are connected together. 2b is a meet point where contour segments C34 and C56 are connected together. 2c is the meet point, contour segment) C5
I, C3s are connected together. 2d is a meet point where the contour segments C51 and C5s are connected together. Note that L1 to L8 indicate each scanning line, and A,
B shows a binary image.

FIGS. 2(a) to 2(d) are schematic diagrams explaining the configuration of each contour segment shown in FIG. ), and the direction code (raster segment type) of each raster segment 11 to 14 is ``O to 3J''.Raster segment 11.14 is Raster segment 12 indicates that raster segments are connected to each other, and raster segment 13 indicates that a new contour segment has been generated. That is, in FIG. 1, when scan line L1 is scanned, branch point )la of binary image A is extracted, and contour segment CS, is extracted.

C52 occurs. That is, the raster segment [3] at the branch point 1a is defined, and the raster segments 11 and 74 forming each contour segment CS+ and C52 are obtained every next scan. When the scanning line L2 is scanned sequentially, the branch point 1b of the binary image A is extracted, and the contour segment) C53, C
54 occurs. That is, the raster segment 13 at the branch point 1b is defined, and the raster segments 11 .
14 are obtained every next scan. Then, when the scanning line L3 is scanned, a meet point 2a where the contour segments C52 and CS3 are connected is defined, and a raster segment 12 is obtained.

When this process is performed up to the scanning line L8, one closed outline element group is formed.

The contour segment CSI ~C obtained in this way
The connection relationship of 36 is defined by the active chain AC, which will be described later, and is created in a storage means (not shown).
Stored in ABLE.

The active chain AC of the binary image A shown in FIG.
AC+ = (C3+), AC2=(C32,C33
dispute.

GSa, defined from (S6', C35),
Active chain A C= (AG+, AC:2”
). Note that the tree means that the contour segments C5I are connected in reverse order.

Next, the outline direction encoding processing operation of a binary image will be explained with reference to FIGS. 1 to 3.

Raster segments 11 to 14, which are contour fragments obtained by one main scan of a binary image, have a 4-connected direction code of 1.
5 is encoded into rO~3J, and includes rask segment detection processing, rask segment connection processing, branch processing,
Meet point processing and tracking end processing are executed. The order of each of the above processes will be explained below.

FIG. 3 is a schematic diagram illustrating a pointer stack that stores contour segment information. PS is a pointer stack, which is composed of a first-in command first-out memory, and information on each contour segment C3i (ID number, contour segment (address information, etc. of the last data of C3) are recorded. C5BUF is a contour segment buffer that stores contour segments C5i as a four-connected chain incoat row. The contour segment buffer C3BUF is set to 1, for example, on RAM.

Raster segment detection processing (scanning line element detection processing) is performed by 2
The current scanning line of the value image A is compared with the previous scanning line, and the raster segments 11 to 11 existing in that line are compared.
Extract 14. At this time, Rusk segment type,
In the case of the raster segment length, for example raster segment 11, the raster segment type is (a), the four-connected direction codes 15 are foj, fOJ, r3J, and the raster segment length is r2j.

Raster segment connection processing (contour line element connection processing) is performed using raster segment detection processing (raster segment) 11.1
4 is detected, the detected raster segment 11.14 is converted into a contour segment cs, (i=
1 to 8) is determined according to the contents of the pointer stack PS.

The detected raster segment type is (a).

In case (1), the first data of the pointer stack PS is taken out and the raster segment is connected to the end of the corresponding contour segment. After this, pointer stack PS
The end address of the data taken out from is updated and the data is bushed at the end of the pointer stack PS.

Branch point processing is a process that is performed when raster segment 13 is detected in raster segment detection processing, that is, when two new contour segments C5I are generated. Define ID number. Next, the first quadruple M direction code 15 of the two new contour segments) CSi is written into the contour segment buffer C3BUF. Subsequently, the information of the new contour segment) CS i is written to the last of the pointer stack PS. This ends the definition setting process.

The meet point processing (contour line element group representation processing) is performed when the rask segment 12 is detected, that is, when the two CS
In the process when i meets at one point (meet point), C3 information is extracted from the top of the pointer stack PS, the rask segment is connected to the corresponding contour segment C51, and then the top 2 of the pointer stack PS is
At the same time, the combination of active chains is updated.

The trace end process (extraction process) is a process that is performed when the branch points that are the origin of two active chains are the same after the meet point process is completed, and ACTABL
Check the contour segments that make up the contour line and their connection order with reference to E, and connect the necessary contour segments correctly to create one 4-connected in-code string. This means that the detection of one contour line has been completed.

Next, a method for extracting contours from a z-value image according to the present invention will be explained with reference to FIG.

FIG. 4 is a flowchart illustrating a method for extracting contours from two images according to the present invention. Note that (1) to (9) indicate each step.

Scan two images and execute the rask segment detection process 1) and determine the detected rask segment type 2) If the rask segment type is (a) or (1), the rask segment detection process Processing is performed to connect the rask segment 11.14 to the contour segment C51 corresponding to the direction code of the rask segment (3). Next, it is determined whether the connection process for 1217 minutes has been completed (4), and if NO, step (1) is determined.
), and if YES, the scan line is updated and the process returns to step (1) (5).

On the other hand, if the rask segment type is (one) in step (2), that is, if rask segment 13 is detected in the rask segment detection process, two new contour segments CS i have occurred. Therefore, ID numbers for new contour segments, active chains, and branch points are defined. Then,
Write the first four connected direction needs 15 of the new contour segment) C3; to the contour segment buffer C3BUF. Next, branch point processing is executed to write information on the new contour segment CSi to the pointer stack PS6), and the process returns to step (4). On the other hand, if the rask segment type is (A) as determined in step (2), the C8 information is extracted from the top of the pointer stack PS, the rask segment is connected to the corresponding contour segment CSI, and then the top of the pointer stack PS is Execute meet point processing that discards the two pieces of information and also updates the combination of active chains 7).
Determine whether the branch points that are the sources of the two active chains are the same (8). If NO, return to step (4); if YES, it is determined that one contour has been detected. Therefore, ACTABL
Check the contour segments that make up the contour line and their connection order by referring to E, and execute the trace termination process to correctly connect the necessary contour segments and create one 4-connected chain input code string9). Return to step (4).

Next, the rask segment detection process will be further explained with reference to FIG.

FIG. 5 is a flowchart illustrating the rask segment detection process according to the present invention. In addition, (1) to (13)
indicates each step.

Obtain the white/black change position of the current scan line from scanning (
1), sorting is performed in descending order of position change based on the data of the previous scanning line (2). Then, extract two pieces of white/black change position data from the beginning of the sorted data 3), determine whether the two data are extracted from the current scanning line and the previous scanning line, respectively 4), if YES For example, it is determined whether the change position value of the previous scanning line is smaller than that of the current scanning line (5), and if YES, the rask segment type is determined as (a) and the process proceeds to step (11) and subsequent steps (8).
, if No, the rask segment type is determined to be (work) and the process proceeds to step (11) and subsequent steps (7).

On the other hand, if the determination in step (4) is No, it is determined whether both data are from the previous scanning line8).
If YES, the rusk segment type is determined to be (a) and the process proceeds to step (11) and subsequent steps (9); if no, the rusk segment type is determined to be (one) (10).

Then, steps (f3), (7), (9
), Processing for contour extraction, that is, branch point processing, meet point processing, contour segment processing, is performed according to the rask segment type determined in (10) (see FIG. 4) (11). Next, it is determined whether all data processing for one line has been completed (12). If No, return to step (3) and perform the same processing on the remaining data; if YES, update the scanning line. Return to step (1) (13). This operation is repeated until the scanning line is completed.

In the above embodiment, the case where the image information is a binary image has been described, but the present invention can be easily applied even when the input image is a multi-value image by replacing the rask segment detection process with one for a multi-value image. In other words, if the input image is a multivalued image, a series of images of the same color is created on one scanning line.
You can treat it as Hereinafter, a method for extracting a rask segment from a multivalued image will be explained.

FIG. 6 is a schematic diagram illustrating a subsequent run set, and 21 is the current run, which is the run to be processed. 22 is the next run, which is the run to the right of the current run. 23a-23
e is a run. Note that 21a and 21b are subsequent runs,
Among the runs indicated by the subsequent run pointer P and the runs to the right, those connected to the current run but not connected to the next run, and those connected to both the current run and the next run have the same color as the current run.

However, if the run 23c has the same color as the current run 21, it becomes a member of the subsequent run set. Note that the subsequent ten set may be an empty set.

FIG. 7 is a schematic diagram illustrating a continuous run set, and the same parts as in FIG. 6 are given the same reference numerals.

In this figure, 25a to 25h are runs, and each run 25
a, 25c, 25e, 25g are currently run 21
It shows that it is a new orchid with a different color. Each run 25b, 25d, 25f, 25h has the same color as the current run 21. According to the definition of a subsequent run set, runs 25a to 25g are subsequent runs to the current run 21, and among these, a run set that has the same color as the current run 21 is defined as a continuous run set. That is, runs 25a, 25c, 25e
, 25g is a continuous run. However, i! ! In the case of a continuation lasolan set, the color of the current run 21 is a stop run that stops at that spot. Furthermore, a merged run having the same color as the runs on the left and right of the current run 21 may occur directly below the current ten that has a subsequent run set of the empty set.

Next, a method for extracting contour lines from a multivalued image will be described with reference to FIGS. 8 and 9.

Figure 8 is a schematic diagram showing the relationship between runs, and 3]
indicates the left and right R3 of the continuous run, 32 indicates the new ten RS pair, and 33 indicates the left and right RS pair of the intermediate continuous run.

FIG. 9 is a flowchart 1 for explaining a method for extracting contours from a multivalued image, showing another embodiment of the present invention. In addition,(
1) to (14) indicate each step. Among these steps, steps (1) to (9) correspond to the rask segment detection process shown in FIG.

First, the current run 21 is set (1). Next, a subsequent run set and a continuous run set are determined for this current run 21 (2). Next, it is determined whether the subsequent run set is an empty set (3), and if YES, the current run is a stop run, so the current run is set as a stop run and that information is stored in the RS buffer, which is a first-in first-out memory. Record (5). Next, it is determined whether or not a merged run exists (8), and if NO, proceed to step (10) and subsequent steps; if YES, merge run processing is performed to correct the corresponding rask segment data, and proceed to step (10) and subsequent steps ( 7).

On the other hand, if the judgment in step (3) is No, that is, if the subsequent run set is not an empty set, it is determined whether the continuous run set is empty or not. Record the R3s existing between the current run and the subsequent run set in the RS buffer in order from the left.8)
, if YES, the current run is stopped, and the subsequent ten performs raster segment detection on the new ten set, and the result is stored in the -EIRS bar.

Next, one piece of RS data is taken out from the beginning of the RS buffer (10), and branch point processing, meet point processing, contour segment connection processing, etc. are performed depending on the raster segment type (11). Next, it is determined whether the RS buffer is empty (12), and if No, the process returns to step (10), and if YES, the current run pointer and subsequent tempo pointer are updated (13), and then the previous screen is updated. 14.
), if YES, end the control; if NO, step (
Return to 1).

In this way, even in a multi-valued image, contour line information of the multi-valued image can be extracted at high speed by one image scan.

In the above embodiment, the case where binary image data is directly processed has been explained, but if similar processing is performed on compressed data obtained by compressing binary image data in the scanning direction,
It goes without saying that contour lines can be extracted faster.

〔Effect of the invention〕

As explained hereafter, the present invention includes a scanning line element detection process that detects a scanning line element corresponding to a contour new moon of an image within a scanning line width every time an image is horizontally scanned, and A contour line element connection process that connects to the end of the detected scanning line element, a definition setting process that defines and sets a new contour line element, and a contour line that merges two contour line elements and expresses them as a group of contour lines. It consists of prime group expression processing and extraction processing that determines whether a closed contour element is formed according to the relationship between two contour line elements and extracts the contour. All contour lines can be extracted and encoded in one screen scan while accurately connecting to one of the contour elements formed in the previous scan, and the contour lines of the same color area of not only binary images but also multi-value images can be extracted and encoded. It has an excellent effect of being able to extract quickly and efficiently.

[Brief explanation of drawings]

FIG. 1 is a schematic diagram illustrating a method for extracting contour lines from a binary image showing an embodiment of the present invention, FIG. ff12 is a schematic diagram illustrating the configuration of each contour segment shown in FIG. FIG. 4 is a schematic diagram illustrating a pointer stack for storing segment information; FIG. 4 is a flowchart illustrating a method for extracting contours of a z-value image according to the present invention; FIG. 5 is a flowchart illustrating a raster segment detection method according to the present invention; FIG. 6 is a schematic diagram illustrating a subsequent run set, FIG. 7 is a schematic diagram illustrating a continuous run set, FIG. 8 is a schematic diagram showing the relationship between runs, and FIG. 9 is a schematic diagram illustrating the relationship between runs. A flowchart illustrating a method for extracting contour lines from a multi-valued image showing an example, FIG. 10 is a schematic diagram illustrating a matrix pattern of raster data vector data conversion (R-V conversion), and FIG. 11 is a schematic diagram illustrating the R-V conversion operation. Schematic diagram to explain, 12th
The figure is a flowchart explaining the RV conversion operation. In the figure, 1a to 1d are branch points, 2d to 2d are meetpo India, 11 to 14 are raster segments, and 15
is a 4-connected direction code, and 25a to 25h are tens. Figure 1 Figure 4 Figure 5 Figure 6 Figure 10 Figure 11 Figure 12

Claims (1)

[Claims] Each time an image is horizontally scanned, a scanning line element detection process that detects a scanning line element corresponding to a contour fragment of the image within the scanning line width, and a scanning line element detected by this scanning line element detection process is detected in the previous scan. A contour line element connection process that connects to the end of the scanning line element formed by , a definition setting process that defines and sets a new contour line element, and a contour that merges two contour line elements and expresses it as a group of contour line elements. An image contour characterized by comprising a line element group expression process, and an extraction process of determining whether a closed contour line element is formed according to the relationship between two contour line element groups and extracting a contour line. Line extraction method.