JPH01251182A - Contour detector - Google Patents

Abstract

PURPOSE:To prevent the capacity of an image memory, the constitution of hardware and processing time from the influence of the size of an image or resolution and simultaneously and to reduce a noise component by making a table showing a coordinate and connecting relation based on a command column obtained from a contour extraction part and obtaining a data column having a meaning by making a contour element as a line or a graphic. CONSTITUTION:When the command column concerning to image data in the thick frame of a figure is fetched by a command column analysis part 2, the contour element Cj shown by a dotted line is registered to the contour element table, the direction and the coordinate of it are written and simultaneously, the numbers of the other contour elements which are respectively connected before and after the concerned contour element Cj are written in a connection element number column. In this case, since the contour element Cj is positioned at the front of the contour element Ci, it is written in a front connection column concerning to the column of the contour element Cj. Besides, the contour number Si is written in a contour number column. As for the direction of the contour element, (a1-a8) is specified corresponding with a top and bottom, a right and left and an oblique direction like the figure in case of 8 coupling and the four directions of a top and bottom and a right and left are specified in case of 4 coupling.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Application of Δ Industry-1- The present invention relates to a device for detecting the outline of a binary image.

B1 Summary of the Invention The present invention is an apparatus for detecting a contour corresponding to a black-and-white boundary based on binary pixel data corresponding to black and white obtained by raster scanning a subject, which detects pixel data of 2 pixels x 2 pixels. Based on the pixel data taken out one line before, for example, find the connection relationship between the contour pixels in the raster scan Y direction, create a command string containing this relationship and each piece of information about the arrangement pattern of the pixel data, In addition, based on these command sequences, a table in which contour elements are serialized is created, and short contour vectors are removed from the contour vector table, thereby reducing image memory capacity, hardware configuration, and processing time. The present invention aims to improve the performance of the apparatus by eliminating short vectors, reducing noise components, and eliminating short vectors.

C Conventional technology When processing patterns such as characters and figures, for example, converting objects such as documents and drawings into black and white binary image data (human cover pattern) obtained by operating a human-powered device such as an image scanner. Then, contour pixels of the object are extracted from this binary image data, and information compression processing and object recognition processing are performed.

Among these, the conventional method for extracting contour pixels of an object from binary image data exhibits an operation as shown in FIG. This can be described using the following process flow.

■First, store all binary image data in a dedicated single image memory M.
to be memorized.

(2) Next, from this memory M, a point, for example, point P, which is the starting point of the contour of T to be detected is searched.

■Sequentially track and extract the points adjacent to this point P1 and
``Extract the contour pixels of.

1) Problems to be Solved by the Invention As described above, the conventional contour pixel extraction method requires a dedicated memory for storing one screen's worth of binary image data. This takes into account the disadvantage that the larger the size of the manual drawings and drawings, and the higher the resolution, the larger the memory capacity.

From a hardware configuration point of view, since the image memory increases in proportion to the vertical and horizontal sizes of human-generated images, a configuration that takes into account the addition of a memory board may be necessary in some cases.

For example, an image memory for 〇 size requires 16 times the capacity compared to an A4 size image memory, and even when it is composed of a Δ4 size memory board or one AO size requires as many as 16 sheets.

Next, processing time also has a significant impact. Conventional methods require waiting time until one screen's worth is stored in the image memory, and in order to extract the contour, it is necessary to perform sequential tracking using software, which also depends on the image of the target image. The processing time will be affected in proportion to the size and resolution.

To summarize the above, in the conventional contour extraction method, simply increasing the image size or increasing the resolution to C11,
There are drawbacks that affect hardware configuration, processing time, etc.
Furthermore, it was a factor that affected the appearance and price of the product.

The present invention has been made to solve these problems, and it is not affected by the image memory capacity, hardware configuration, or processing time, and it also makes it easier to extract contours. It is an object of the present invention to provide a contour detecting device that can perform the following steps, remove short vectors, reduce noise components, and improve the performance of the device.

161 Means for Solving the Problems FIG. 1 is a diagram showing the configuration of the present invention, and 1 is a contour extraction section. As shown in Fig. 2, this contour extraction unit 1 takes in binarized pixel data corresponding to black and white obtained by raster scanning the subject, and extracts pixel data of 2 pixels x 2 pixels along the scan line. Based on this pixel data, a command sequence for contour detection is created and output to the contour analysis section 2 at the subsequent stage. The contour analysis section 2 updates the descriptions in the contour element table, contour management table, and contour connection table in the table storage section 3 based on the command sequence. The contour analysis section 2 also includes a macroscopic processing section 31 that uses the contour element table and the contour management table to extract the maximum point from the starting point among the points constituting one contour pixel series, and a macroscopic processing section 31 that extracts the maximum point from the starting point. microscopic processing unit 32 that extracts
Then, the contour vector of the point extracted by this processing section is written in the contour vector series table 33, and the short vector among these contour vectors is removed by the short vector removal section 34, and the description is updated.

Here, the contour element corresponds to a vector connecting two adjacent black pixels, and the contour is constructed by connecting these elements. As shown in Figure 3, the contour element table is a contour element table in which a unique code is attached to each contour element and a contour made up of a group of these contour elements, and its coordinates and direction are indicated for each contour element. This is for describing the code of the contour to which the contour element belongs and the codes of other contour elements connected before and after the contour element, respectively. The contour management table is used to record the codes of contour elements located at the tip and end of each contour, as shown in Figure 4, and the contour connection table is used to record the codes of contour elements located at the tip and end of each contour, as shown in Figure 5. Coordinates arranged in the raster scan direction, codes of unconnected contour elements whose front or rear ends exist at those coordinates, and the distinction between the front and rear ends of the terminal connection ends of the contour elements are described in correspondence. It is. The contour vector table is used to record the code of the starting point coordinates for each contour vector series, as shown in FIG.

The command string output from the F4 action contour extraction unit l includes information on the connection relationship between pixels and contour elements related to the 2 pixel x 2 pixel pixel data extracted at that time, and the black and white arrangement of the pixel data. It is a combination of a connection/pattern code containing pattern information and a coordinate code indicating the X coordinate of the pixel data "j".

An example of the connection relationship is shown in FIG. 6. When pixel data of 2 pixels x 2 pixels surrounded by the large frame in FIG.

This information indicates that the front end of the contour element located at is connected. In this example, the coordinates of pixel data are the coordinates of pixel 1) 1 located on the right-hand side when facing the page. In addition, when the pixel data surrounded by the large frame in Figure 1.11 (b) is imported, the rear end of the contour element located at the coordinate X n -1 before the X coordinate of this pixel data is connected. This will give you information that there is. In FIG. 6, frames with O marks indicate black pixels, and frames without O marks indicate white pixels.

For example, if the main frame in FIG. 6(a) is entered, the connection/pattern hood in this case includes the above-mentioned connection information and information on the black and white array pattern within the large frame, and the command string is as follows:
This connection/pattern code is combined with a coordinate code indicating the X coordinate of pixel 1). Note that this coordinate code can also be generated on the contour analysis section 2 side in synchronization with the timing of fetching the command sequence.

When the command string obtained in this way is taken into the contour analysis section 2, the following processing is performed. Assuming that a command string related to the pixel data of the large frame in Figure 7 is imported, the contour element Cj shown by the dotted line is registered in the contour element table, its direction and coordinates are entered, and the contour element Cj is
Enter the numbers of other contour elements connected before and after j in the connected element number column. In this case, the contour element Cj is located in front of the contour element Ci, so enter Ci in the rear connection column for the column for the contour element Cj, and enter 01 in the front connection column for the column for the contour element Cj. do. In this example, Si, which is the number of the contour to which the contour element Cj belongs, is entered in the contour number field. Regarding the directions of the contour elements, for example, in the case of 8 connections, a1 to a8 are defined corresponding to the l-down, left/right, and diagonal directions, as shown in Figure 8, and in the case of 4 connections, the directions are defined in the upper, lower, left, and right directions. Four directions are defined. Then, for the contour weight Si in the contour management table, change the contour element number column at the tip from C1 to C.
At the same time, regarding the X coordinate of the pixel data in the contour connection table, the forward connection column of the unconnected contour element number is updated from Ci to Cjl::. By the way, in actual processing, since the contour element to be connected to the contour element Ci is clarified by the contour connection table, Cj in the front connection column related to the contour element number Cj column in the contour element table is The description is made with reference to the contour connection table. In the above, when two contours generated separately as the raster scan progresses are connected to form one contour, the associated contour numbers are integrated into one contour and become the same.

- To,? L! In the process described above, the point that is the maximum when viewed from the starting point is extracted from each point constituting one contour pixel series, and then the point that is the maximum distance between the two consecutive extracted points is extracted. The contour vectors extracted in this way are stored in a table. The Nyoto vector is removed from the stored contour vectors.

G. Embodiment In an embodiment of the present invention, an internal bus 4.
A contour extraction section 43, a contour analysis section 43, and an internal memory 44 serving as a table storage section are connected to the contour extraction section 43, a contour analysis section 43, and an internal memory 44 as a table storage section. Further, the internal bus 41 is configured to be connected to a bus interface 4. The contour detecting device 4 is connected to the main bus 46 via the contour detecting device 4, thereby providing information obtained by the contour detecting device 4 to external equipment.

Next, the contour extraction section 4. A specific example of a command sequence generated from the following will be described. Fig. 10 and Fig. 1+ are diagrams showing the relationship between the black and white arrangement pattern of pixel data of 2 pixels x 2 pixels and the connection relation of the contour elements to this pixel data, and Fig. 1O shows the 8-connected , and FIG. 11 correspond to the case of four connections. In these figures, the four vertically arranged windows at the left end indicate pixel data (d, -d3) of 2 pixels x 2 pixels as shown in Fig. 12, and the parts where NJ and rOJ stand are black pixels, respectively. Corresponds to a white pixel. In addition, as shown in FIG. 13, the four-panel window arranged horizontally at the top end has two frames on the left and right of the upper side, respectively, indicating the presence or absence of a forward-connected contour element at the X-locus LM X n-1 of pixel data, and the , shows the absence of the contour/element of the forward connection in , and the lower left and right two
A connection flag indicating the presence or absence of a backwardly connected contour element at the X coordinate X n -1 of pixel data of each frame and the presence or absence of a backwardly connected contour element at the X coordinate X11,
I'll and I'OJ mean connected and not connected, respectively. For example, the connection flag in the pixel area surrounded by the large frame in FIG. Only the top right frame, such as number 5 [-1, from the left in the connection flag column in Figure 1O, is represented by the rlJ window.

In FIGS. 10 and 11, the intersection of the pixel data and connection flag terms is blank, which means that such combinations of pixel data and connection flags are processed by the contour analysis unit 4J. This means that there is no need to do this. When a combination is marked with a circle, it is necessary to process it in the contour analysis section 43, and the connection/pattern code corresponding to that combination is sent to the contour extraction section 4. Occurs at. Combinations marked with an X indicate that such combinations do not exist.

For combinations marked with a △, a connection/pattern code corresponding to the combination is generated only when encoding the run range. Note that in Figures 6 and 7, the contour elements are drawn without considering the direction in which they occur (direction of the vector), but in reality, the contour elements are drawn at the outer edge of the object (or black image area). 1O and 1I are created corresponding to this way of generation.

The command string includes the connection/pattern hood obtained in this manner, a read ready code, a coordinate code, etc., and is expressed as shown in FIG. 14, for example. In the same figure, is the read ready code, and the combination of pixel data and connection flag is shown in Figures 1O and 11.
When it corresponds to X or X and Δ in the figure, the command string is invalidated. A is the X coordinate of the pixel data, A3°A is a code indicating the black and white of the two pixels lined up on the left and right at the bottom of the pixel data, and when performing run length conversion, A1. Δ is required. A indicates the end of one page (all lines) (EOI) (ENDOF PA
GE) code, A, is FOR(
END OF ROW) code, A7 is the connection/pattern code.

Next, the contour extraction section 4. A specific example of the configuration will be described with reference to FIG.

Pixel data latch! 01 receives pixel data manually from the signal line 5 and a line memory 105, which will be described later. This latch takes in and latches 2 pixels by 2 pixels as shown in FIG. 12, ie, I'i, and data d0 to d of four adjacent pixels from these manually input data.

Address generation circuit 108 generates X and Y coordinates in accordance with the progress of the raster scan shown in FIG. This circuit stores the generated coordinates in a line memory 105, which will be described later.
．． Peripheral determination circuit 109 and forward flag memory 106
and the contour analysis unit 4 through the signal line 8. as part of command sequence 1.

The line memory 105 stores the Y coordinate which is one smaller than the Y coordinate generated by the address generation circuit 108, that is, the Y coordinate of the pixel data d, d, latched by the pixel data latch 101.
Pixel data of coordinates is taken in from the output of the pixel data latch 101 and stored sequentially.

The forward flag memory 106 has the same number of addresses as the number of X coordinates (for example, when the number of is connected and the contour element is not connected to any other contour element, the logical rN data -) is written into the address corresponding to the X coordinate of the pixel.

On the other hand, the backward flag memory 107 also has the same number of addresses as the coordinates, but this memory contains information about cases where the rear end of a contour element is connected to a certain pixel and that contour element is not connected to any other contour element. At this time, logical "l" data is written into the address corresponding to the X coordinate of that pixel.

The connection flag latch 102 has a forward flag memory 106,
Output data from the rear flag memory 107 and a connection flag change circuit 104, which will be described later, is held as data indicating the connection state of contour pixels.

The periphery determination circuit 109 determines whether or not the four pixels d0 to d to be processed protrude from the boundaries of the screen, based on the X and Y coordinates generated by the address generation circuit 108. If the pixel to be processed protrudes from the periphery of the screen, a predetermined signal is output to the pixel data raffle 101 and the connection flag latch 102. When these latches accept this message, the pixels that protrude from the boundaries of the screen are forcibly set as white pixels (background pixels).

The command generation circuit 103 generates a predetermined command to the contour analysis unit 4 to perform contour tracking based on the four pixel data output from the pixel data controller 101 and the flag data output from the connection flag lunch 102. do. At this time, the command generation circuit +03 sends a read ready signal indicating whether or not processing is necessary through the signal line 6, and sends the connection/pattern code to the contour analysis unit 4 through the signal line 7. send to

The connection flag change circuit 104 changes the connection state of the pixels after the contour analysis unit 4 executes the process according to the command ``2'', so the connection flag after the process is output from the pixel data latch +01 and the connection flag latch +02. Based on the data, connect the front flag memory 106, the rear flag memory 107, and the flag latch! Output to 02.

Next, the operation of the circuit shown in FIG. 15 will be explained. The address generation circuit +08 generates and outputs X and Y addresses corresponding to the pixel data most recently sampled and manually entered from the signal line 5 as the raster scan progresses.

When pixel data sampled by raster scanning is manually input from the signal line 5, the pixel data latch 101 sequentially receives the pixel data along with the pixel data from the line memory 105, and divides the pixel data into 2 pixels x 2 as shown in FIG. Data d of pixels, that is, four pixels adjacent to each other. -Latch d3.

The line memory 105 takes in the pixel data d1 output from the pixel data line 101, thereby reading the pixel data of the previous line, that is, the pixel data d currently input from the signal line 5 from the Y address.

One row of pixel data that is one row smaller than the Y address of is stored. Then, pixel data d, d,
are sequentially input to the pixel data latch 101, the line memory +05 sequentially outputs pixel data d, , d, to the same latch based on the address data output from the address generation circuit 108. As a result, the pixel data latch 101 receives four adjacent pixel data d. -d3 can be latched.

The connection flag changing circuit +04 inputs the above pixel data d, -d
Based on 3 and the output data of connection flag launch +02,
Generates four connection flag data.

Of the four pieces of data, two are forward connection flag data corresponding to the L side two frames of the connection flag shown in FIG. 13, and the other two are backward connection flag data corresponding to the lower two frames.

Connection flag data and its generation will be explained in detail using FIG. 16. For example, there are two images A on screen P,
B is shown, and these images are each composed of pixels marked with a circle. In the figure, the four images surrounded by thick lines are the pixels currently targeted for contour tracking processing.

On the other hand, R shown on the screen P represents forward connection flag data of each X coordinate, and T represents backward connection flag data of each X coordinate. Each piece of data is drawn in the order of coordinates, and the right side has a larger X coordinate. A blank field indicates that the value of the flag data is rOJ, and a field in which 1 is written indicates that the value is "1". Specifically, since the rear end of the contour element Ca is connected to the pixel el, the rear connection flag data Tl of the X coordinate of this pixel is NJ. Conversely, since the front end of the contour element cb is connected to the pixel e2, the forward connection flag data R2 at its X coordinate is "1". Similarly, backward connection flag data T4 corresponding to pixel e3 is I
N, forward connection flag data R4 corresponding to pixel e4 is also 1
-1". Other flag data are all "0" because no contour element is connected to the corresponding pixel.
”.

These forward and backward connection flag data are written to predetermined addresses corresponding to the X coordinates of the forward flag memory 106 and backward flag memory +07, respectively, by the connection flag change circuit 104 as described later.

When the four pixels surrounded by thick lines in FIG. 16 are the targets of processing, the front flag memory 106 and the rear flag memory 107 respectively store connection flag data R2, Output T2. On the other hand, the connection flag change circuit 104 outputs connection flag data R3, T3 corresponding to the X coordinate of the processing target pixel on the right side.

Connection flag latch 102 latches these and outputs them to command generation circuit 103 and connection flag change circuit IO4.

The connection flag changing circuit 104 inputs the 4-bit connection flag data and the pixel data d from the pixel data latch 101.
0 to d are taken, and flag data after contour element tracking processing is obtained. That is, as shown in FIG. 17, the next contour element gCc is connected to the contour element cb in the pixel e2 through the tracking process, so the connection flag change circuit 104 changes the front connection flag data R2 to "0". ”, and the forward connection flag data [3] is set to “l”. The rear connection flag data is not changed because the rear end of the contour element continues to be in a state where it does not exist independently.

The connection flag change circuit 104 stores the flag data R2, T2 after the above change processing in the predetermined addresses of the front flag memory 106 and the rear flag memory 107, and stores the flag data +<a,'r3 after the change processing into the connection flag latch. 1
It is output to 02 and latched in preparation for the next process.

Then, in the command generation circuit 103, the pixel data line, 7
A connection/pattern code is generated based on pixel data d0 to d from the chip 101 and 4-bit flag data from the connection flag latch 102, and the lower two data of the 2 pixels x 2 pixels pixel data are generated. d, , d are also output, and these codes are sent via the signal line 7 to the contour analysis section 4. given to.

Furthermore, the command generation circuit 103 outputs a read ready code to the signal line 6. For example, if run length conversion is not performed, the connection/pattern code is set to ○ in Figures 1O and 11.
When the read ready code corresponds to a mark, the read ready code is set to rN, and when it corresponds to a Δ mark or an X mark, the read ready code is set to rOJ. Further, the peripheral determination circuit 109 receives an E01) signal indicating the end of one page through a signal line 9, and an E01) signal indicating the end of one page.
A FOR signal indicating the end of the line is input, and as a result, an EOP food and l': OR code is output from there to the signal line 6. In this way, the EOP code and the EOR code are given to the signal line 6 along with the coordinate code indicating the X coordinate of the pixel data from the address generation circuit 108, and these hoods are given to the contour analysis section 43.

In the above, as in the embodiment described above, FIG.

Generating read ready codes according to the combinations shown in FIG. 11 has the advantage that isolated pixels can be removed as noise. In other words, there are four types of pixel data patterns generated for an isolated pixel, as shown in Figure 18 (a) to (d), but in this case, there is no connection of contour elements to this pixel, so reading The ready code is [-〇],
Therefore, the contour analysis section 4. This is because no processing is being performed.

g However, when implementing run-length conversion, the first thing is important! In the case of the patterns shown in FIGS. (a) and (b), the read ready code is "l", but this has nothing to do with the contour tracking process.

Here, the processing performed by the contour analysis section 43 is shown in FIGS. 19 to 2.
This will be explained with reference to FIG. First, as shown in Figure 19, after setting the Y address to 0, a command string is fetched, and if the EOP instruction is included in this command string, the process is terminated, and if there is no instruction, the FOR Determine whether or not the instructions have been given. If FOR is instructed, add one Y address and take in the command string, if not instructed, internal memory/lJ
Each table (see FIG. 3 to FIG. 5 and FIG. 26) stored in the server is updated.

The flow of updating each table is as shown in FIG. 20, and steps ST to ST in the figure are as follows.

ST1: It is determined whether or not a single contour element has newly occurred.

ST: When a contour element is generated, it is determined whether or not it is connected to an already existing contour element.

S'"3... It is determined whether the ends of the two contour elements that have already occurred exist in the 2 pixel x 2 pixel image data.

Next, I will explain the details of processes 1 to 3 in Figure 120. In process 1, as shown in Figure 21, step S'F
, ~s'r, are executed. Each step sT1 to ST,
is as follows.

s'r, . . . new contour element number Ci and contour number Si are secured in the contour element table shown in FIG. 3 and the contour management table shown in FIG. 4.

ST, ... Enter the relevant information in the direction and coordinate column of Ci number [1] of the extension element table, and enter N0NEζ (a code indicating that there is no applicable item) in both the forward connection and backward connection of the connection element number. , S: is called into the field of belonging contour number, respectively.

ST: Enter 3 Ci into both the contour element number 1 at the Si-th tip and the contour element number at the end of the Si-th contour management table.

S'r, . . . Input C1 +'F into the corresponding columns of the front connection and the rear connection of the contour connection table shown in FIG. 5, respectively.

In process 2, as shown in FIG. 22, step ST is performed.

~S'r6 is executed. Each step s'r, ~S
Ts is as follows.

ST1: Newly secure the contour element quantity C1 in the contour element table.

S'r... Find the contour element number Cj of the connection destination from the corresponding column of the contour connection table, and change the column to N0NH.

S'F, . . . Ci is written in the corresponding Cj-th forward connection and backward connection columns of the contour element table, and the contour number Sk to which Cj belongs is determined from the belonging contour number column.

ST4...Ci number 1 [direction of 1,
Write the corresponding item in the coordinate column, Cj in the forward connection and backward connection columns, N0NE in the other column, and Sk in the belonging contour number column.

s'r, 909 Contour Contour Connection Table The corresponding one of the columns of the k-th tip contour element number and the end contour element number is changed to Ci by 7+.

S T s...Ci in the corresponding column of the contour connection table
Write.

In process 3, as shown in FIG.
T is executed. Each step S'"1 to ST" is as follows.

ST,... Find two contour elements Ci and Cj to be connected from the relevant columns of the contour connection table, '! 1 Set the column to N0NE! Change F.

S' . . . Insert C3 into the appropriate one of the forward connection and backward connection of Ci number 11 in the contour element table, and find the contour number Sk to which Ci belongs from the column of belonging contour number.

S' ``3...Write Ci in the applicable one of the Cj-th forward connection and backward connection in the contour element table, and write the contour number S to which Cj belongs from the belonging contour number column.
Find Q.

ST... Determines whether Sk and SQ are the same number.

S'",... Find the contour element number Cm of the tip contour element number and the end contour element number of the Sk number in the contour management table, whichever is not the target of connection, and s ka
[Rewrite each column of 1 to N0NI.

ST, . . . Of the tip contour element number and the end contour element number of SQ number L1 in the contour management table, the column on the end side to be connected is rewritten to Cm.

FIG. 24 is a diagram showing the relationship between the above processes 1 to 3 and the extraction position of 2 pixels x 2 pixels pixel data. In the figure, the numbers are the numbers of the contour elements, the dotted rectangles are the pixel data, and the black points are indicate black pixels, respectively. The processing of the command sequence generated for the pixel data D, corresponds to processing l, and the processing of the command string generated for the pixel data D,
The processing of the command string generated for this corresponds to processing 2. Further, the command strings generated for the pixel data D, , D correspond to the cases where Sk and Se are equal and different in process 3, respectively.

Next, a step for removing short vectors using the contour vector table shown in FIG. 25 will be described.

FIG. 26 is a flowchart illustrating short vector removal means. In step ST, a circumscribed rectangle (the 27th
) (X, .l Yarn), (X-ax, Y
man) is calculated. Next step ST.

d X = Xaan Xa+to +d Y = y
Compare max-yain and threshold values DX, h, DY, h. Step ST is a comparison/judgment section of step ST, and in step ST3, (d X > D
l,) Determine U (d y > D Y lh),
If YES, in step ST, information on a circumscribed rectangle is created as a contour vector. An example of this information is shown in FIG. 28 as a circumscribed rectangle information table. Said step S
If NO in TJ, then in step ST, vectors of the same series are deleted from the contour vector table shown in FIG. 25 to remove noise.

FIG. 27 is an explanatory diagram showing the circumscribed rectangle of the contour vector,
In FIG. 27, X ma*+ X sl Ill
Y mjz+y, In is as follows.

X aax= m a X (”・+ X ++
”')X m r +, = min (・, x
++ −・−)y, 4, =max(・・+yl+ ・
・・) y+m1n= min (・・・, y6.・
・)dX=Xwaax---X+m1ndy
-y,,1,-yln (dX<I)X,1. ) n (d y<I)Y,,
, ) (11shi, I) X ,,,, D Y ++,
There are two thresholds.

11. Effects of the Invention According to the present invention, the arrangement state of the extension elements can be determined by knowing the connection relationship between pixels and extension elements related to pixel data of 2 pixels x 2 pixels, and the black and white arrangement pattern of the pixel data. Since the connection relationship can be created based on the pixel data taken out l line before, image memory can be saved significantly compared to the conventional method of storing image data for one screen. be able to. Here, the line memory has a general resolution of 16 pixels/yzm and is approximately 19,000 bits in the longitudinal direction even in AO size.
A few kilobytes at most can cover sizes from A4 to AO, and having that much memory makes it possible to create a hardware circuit that is not affected by image size or resolution.

In addition, since processing is performed in units of l-line scanning time, it takes approximately the same amount of time as the human labor required for one screen (maximum 19
Since the contour can be extracted within a delay time of 47 minutes (delay time of 47 minutes), processing speed is significantly faster than that of conventional methods.

Based on the command string from the contour extraction unit, a table that describes the coordinates of the contour element fIj, connection relationships with other contour elements, etc., and a table that describes the contour elements located at the front and rear ends of the contour. By creating a data string that has meaning as a contour element as a line or figure, extract the contour in series units (into a closed line hj that forms the contour)
It can be done easily. In addition, since short contour vector sequences can be removed from among the contour vector sequences, noise components are reduced, and the performance of the apparatus itself can be improved -L. Further, since the circumscribed rectangle information is obtained as a secondary information, it can be effectively utilized when character segmentation processing is performed.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the hardware configuration of the present invention,
FIG. 2 is an explanatory diagram showing the state of raster scanning, FIG. 3 is an explanatory diagram showing a contour element table, FIG. 4 is an explanatory diagram showing a contour management table, FIG. 5 is an explanatory diagram showing a contour connection table, Figure 6 (a). (b) is an explanatory diagram showing the relationship between pixel data and contour elements, FIG. 7 is an explanatory diagram showing the connection state between contour elements, FIG. 8 is an explanatory diagram showing the direction of contour elements, FIG. 9 is a configuration diagram showing the hardware configuration of the present invention, FIGS. Fig. 13 is an explanatory diagram showing connection flags, Fig. 14 is an explanatory diagram showing command strings,
Figure 15 is a circuit diagram showing the contour extraction section, Figure 16 and Figure 1.
Figure 7 is an explanatory diagram of each connection flag data, Figure 18 is an illustration)
A pattern diagram showing a pattern of one pixel, FIG. 19 is a flowchart showing the overall processing of the contour analysis section, FIG. 24 is an explanatory diagram showing the relationship between table updating processing and the position of pixel data, and FIG. 25 is a contour 26 is an explanatory diagram showing a vector series table, FIG. 26 is a flowchart of short vector removal means, FIG. 27 is an explanatory diagram of a circumscribed rectangle, FIG. 28 is an explanatory diagram showing a circumscribed rectangle information table, and FIG.
The figure is an explanatory diagram showing a conventional outline pixel extraction mode. 1.4. ... Contour extraction section, 2, 43 ... Contour analysis section, 3 ... Table storage section, 4 ... Contour detection device, 4
4...Internal memory. 31... Macroscopic processing unit, 32.
. . . Microscopic processing unit, 33 . . . Contour vector series table, 34 . . . Contour vector removal unit. Figure 1: Configuration diagram of the present invention Figure 3: Explanation of contour element table Explanation of contour management table Figure 5: Contour connection table Figure 6: Relationship diagram between element data and contour elements (a) (b) Xn
-I Xn Xn-1Xn
An explanatory diagram of the connection state An explanatory diagram of the direction of contour elements FIG. 9 A configuration diagram of the embodiment - White part memory 45 - - Bus interface 46 - - Main bus 10th drawing Genealogical diagram between establishment of element data and connection relationship IO + Tagawa 8i Hita 11th drawing Genealogical diagram of offensive relationship between element data and connection relationship Figure 12 13. Explanatory diagram of element data. 15. Diagram of connection flag. Explanatory diagram of command string. Pixel pattern explanatory diagram (a) (b) (c)
(d) Figure 19: Flowchart of overall processing of the contour analysis section Figure 2o: Flowchart of update processing Figure 21 Figure 22: Flowchart of process 1 Figure 23: Flowchart of process 2 Figure 23: Flowchart of process 3 Figure 24: Update processing and pixel data Fig. 25: Contour vector series table Fig. 26: Flowchart of note vector removal means Fig. 28: Circumscribed symbol gingata information table Fig. 29: Conventional outline pixel extraction method

Claims (1)

[Claims] (1) In a device that detects the contour corresponding to the boundary between black and white based on binary pixel data corresponding to black and white obtained by raster scanning the subject, pixel data of 2 pixels x 2 pixels horizontally and vertically is converted into a scan line. The connection relationship between the contour element corresponding to the vector connecting two adjacent black pixels and the pixel related to the pixel data is expressed as 1
a contour extraction unit that generates a command string based on the pixel data extracted before the line, and generates and outputs a command string including information on the connection relationship and information on the black and white arrangement pattern of the pixel data; A contour element table for recording the code and the codes of other contour elements connected before and after the contour element, and a contour management table for recording the codes of the contour elements located at the tip and end of each contour. and an import for writing the X coordinates aligned in the raster scan direction, the code of an unconnected contour element whose front end or rear end is located at that coordinate, and the distinction between the front and rear ends of the unconnected end of the contour element. a table storage unit for storing a contour connection table; and a table storage unit for storing a contour connection table; and updating the description of each table in the table storage unit based on a command string from the contour extraction unit, and updating the description of each table in the contour connection table, and The code of is updated by referring to the contour connection table, and the point that is the maximum when viewed from the starting point is extracted from the contour element table and contour management table among the points constituting one contour pixel series. , a contour analysis section is provided which stores contour vectors extracted from points having the maximum distance between two consecutive extracted points in a storage section, and removes short contour vectors from among the stored contour vectors. A contour detection device characterized by a contour detection device.