JPH03224071A - Line thinning system for binary image - Google Patents

Abstract

PURPOSE:To prevent the shift of a line thinned image at a boundary part by performing the parallel line thinning processing only in an area where the boundary parts of plural divided screens overlap with each other and thinning the lines in sequence in other areas. CONSTITUTION:A screen dividing means 11 divides the data on an image data store part 14 storing the image data on an entire screen into the pieces of a prescribed size so that these divided data overlap each other at a boundary part. Then the image data included in an area of a divided screen to undergo the line thinning processing is loaded into a memory 16. A parallel line thinning processing means 12 applies the parallel line thinning processing to the data on the overlapping areas. Then a sequential line thinning processing means 13 is started to apply the sequential line thinning processing to the data on the parts except the overlapping area in a divided area. In this case, the line thinning result obtained in the overlapping area by the means 12 is stored in the memory 16. When the contour of a graphic obtained via the sequential line thinning processing is tracked, the continuous reference is given to the line thinning result of the overlapping area. Thus the shift of a thin line is prevented.

Description

[Detailed Description of the Invention] [Summary] A screen to be processed is divided into a plurality of regions, and each thinned divided image obtained by thinning the image of each divided screen is integrated to obtain an entire thinned image. Concerning a thinning method for binary images to obtain .

Line thinning of a binary image using sequential thinning processing when performing line thinning by splitting the screen so that there is no discrepancy between the thinning result in the overlapping area of the split screen and the thinning result of the adjacent split screen. For the purpose of providing a method, the image data of the entire screen is stored in the image data storage section, and the processing section divides the entire screen into predetermined sizes so as to have areas that overlap with adjacent divided images and stores them in a division table. a parallel thinning processing means that retrieves the image data of one divided screen into a memory and performs parallel AT line processing on the image of the overlapping area of each sub-screen; and an area excluding the overlapping area of each divided screen. The apparatus is configured to include a sequential thinning processing means for performing sequential thinning on the line and combining the processing results obtained by the parallel thinning.

[Industrial Application Field J The present invention divides the processing target screen into multiple areas.

The present invention relates to a binary image thinning method that obtains an entire thinned image by integrating thinned divided images obtained by thinning each split screen image.

In recent years, it has become popular to input drawings, documents, etc. written on paper into computers and utilize them. In particular, concave input devices have come into use that automatically perform polygonal line approximation after reading a drawing using a scanner or the like and provide input to a computer. In such devices, line thinning processing is important as one of the preprocessing techniques for performing polygonal line approximation.

Line thinning is a process that thins the image data obtained when reading it with a scanner etc. so that the line width becomes 1, but if you try to thin the entire image data at once, However, memory is required to load the image data. At this time 1 For example, A
Inputting 3 sizes of drawings at 400 dpi (dots per inch) requires approximately 4 M (mega) bytes, and inputting AO size drawings at 400 dpi requires 31 M bytes of memory. For this reason, in recent years, methods have been used to reduce the amount of memory required by dividing input image data into a plurality of regions and loading each region individually into memory.

However, after thinning the screen by dividing the screen, a problem arises when connecting the divided images to each other, and an improvement is desired.

[Prior Art] Fig. 7 is an example of dividing line thinning; Fig. 8 is an explanatory diagram of screen division with an overlapping area provided; Fig. 9 is an example of dividing line thinning with an overlapping area provided; Fig. 10. 11 is a schematic diagram of line thinning, and FIG. 11 is an explanatory diagram of problems in the sequential type line thinning method.

Conventionally, in order to thin an image on one screen, the entire drawing image data was loaded into one image memory, and line thinning was performed on this image data all at once. This method is the simplest, but as mentioned above, you can print large drawings at 400dp.
When input with i, a large capacity memory is required to store the image data.

This method requires a large amount of memory, which increases the cost of the device, so rather than loading the entire two screens into memory at once, the image data is divided into appropriate sizes and loaded into memory individually for processing. method is now in use. By dividing the image data, the amount of memory required for each piece of divided image data is sufficient, making it possible to significantly reduce the amount of memory required for the device compared to conventional systems.

In order to perform double-sided division, it is necessary to divide the image into two, perform processing, and then integrate the processing results again.

This division method will be explained with reference to FIG.

If the screen shown in Figure B is simply divided into upper and lower parts and line thinning is performed on each, the lines at the edges of the divided image may disappear due to the thinning process, or the lines may be thinned, as shown in Figure B. The position becomes unstable (thinning is difficult at the end of the line because there is no data beyond that point).

In this case, when merging two screens that have been split into one, a thin line may disappear at the position where the two screens connect.

A problem arose in that even if there were thin lines, the positions would not match.

One of the methods to solve this problem is the title of the invention: ``Line figure broken line approximation method by screen division''
18887) has been proposed.

This method will be explained with reference to FIGS. 8 and 9.

When the screen is divided into multiple parts as shown in A of Fig. 8, the width d of the border line of each divided screen is equal to or greater than the maximum number of times the image is thinned (corresponding to the line width used to express the image). An overlapping area with . In example A, the entire 5 screens are
This is the case where the image data is divided into three parts, and each divided screen thins the image data of the area including the part wider by width d than the border line (therefore, the image data is divided into three parts with a common width of 2d around the border line). (will have an overlapping area).

That is, each adjacent split screen has a width d.
Line thinning processing is performed on image data in a wide area. FIG. 8B shows an example of a split screen when there are adjacent screens on the left side, right side, and bottom side, and overlapping areas are provided for each adjacent screen.

An example of division when such an overlapping area is provided is explained with reference to FIG. 9. When the entire screen 90 shown on the left side of FIG. The screen is divided into two screens 91 and 92 having overlapping areas. Next this split screen 91.92
When thinning processing is performed on each of the images, thinning screens 91' and 92 are obtained as shown on the right side of FIG.

The thinned images match on the boundary as shown in this figure,
A better result can be obtained than in the case of not providing the overlapping region shown in FIG. 7 above.

On the other hand, there are two conventional methods for thinning image data: sequential thinning and parallel thinning.
A schematic diagram of line thinning is shown in Figure 0.

Sequential thinning is shown at A in FIG. 1O.

This method performs mask calculations while tracing the outline of a figure to remove outline pixels. The arrows in the figure indicate contour tracking. Furthermore, what is mask operation?

3x3 (center dot up against that center dot.

Using mask patterns (down, left, right, and four diagonal directions), surrounding pixels (dots) are used for the pixel of interest in one image.
The shape and width of the line are detected by comparing and calculating the distribution of the lines.

Then, each dot is examined one by one according to the line locus of one figure, and if the line width consists of a plurality of dots, the dots are removed one by one.

A feature of this sequential thinning process is that the amount of calculation is relatively small, so it is easy to implement on a general-purpose processor.

Next, parallel thinning is performed by scanning the top of the screen (image) containing nine figures in the horizontal direction from the left end to the right end using a mask pattern, as schematically shown in Figure 10B. In this method, the calculation is performed, and when the calculation is completed, the same scanning and calculation is performed in the horizontal direction below the L dot, and the same process is sequentially repeated until the bottom of the dot, and the line width is detected and the line thinning process is performed. In this method, the results of line thinning are stored in, for example, another image memory, so as not to affect the processing. An example of the thinning result is shown in FIG. 10 C1. However, this figure is a scaled-down figure.

This parallel thinning is suitable for hardware implementation because the processing is simple and the amount of calculation is large. Also.

Since the image that is being thinned does not affect the processing, the thinned image is always the same. (
This is because if an intermediate result of line thinning is written into the same image memory, the figure at the previous stage will disappear, and the result of line thinning will not necessarily be the same. ) [Problems to be Solved by the Invention] In the above-described conventional method of dividing a screen into multiple areas and thinning the lines, it is difficult to determine the position of lines on adjacent screens when the divided screens are integrated after the thinning process. In order to always match them, it is necessary to use a parallel thinning method that always yields a constant thinning result, but when implemented on a general-purpose processor, there is a problem in that it takes a long processing time.

On the other hand, in the case of sequential thinning method 2, the amount of processing is limited compared to parallel thinning (the processing target is only the graphic part on the screen), so it is not suitable for processing by general-purpose processors. However, there is a problem in that the thinned image obtained differs depending on the tracking starting point.

To explain this problem using Fig. 11, A in Fig. 11,
An example (partial) of a figure with a dot configuration is shown in . This figure is thinned by mask operation from the upper left to the lower right as shown in ■.

When performing line thinning on the dot (double circle) indicated by a in A, there is also a dot to the right of it, so if you delete a and perform line thinning, B in Figure 11, It will look like this. On the other hand, when processing is performed in the direction indicated by ■ in A, that is, from the bottom of the figure to the right side and upward, the dots indicated by b are erased by thinning and the line is thinned (the line width in the diagonal direction is reduced to one dot). ) is performed and becomes like C0. As a result, B and C have different shapes.

In this way, after performing sequential thinning on the divided screens, the thinning images differ at the tracking start point.

There is a problem that when merging screens, adjacent images and thin lines no longer match.

The present invention provides a binary image in which there is no discrepancy between the thinning results in the superimposed area of a split screen and the thinning results of an adjacent split screen while using sequential thinning processing when performing line thinning by splitting the screen. The purpose is to provide a method for thinning wires.

[Means to solve the problem] FIG. 1 is a diagram showing the principle configuration of the present invention.

In FIG. 1, 10 is a processing section, 11 is a screen dividing means,
Reference numeral 12 represents a parallel thinning processing means, 13 a sequential thinning processing means, 14 an image data storage section, 15 a partitioning table, and 16 a memory.

The present invention performs parallel line thinning processing only on areas that overlap at the boundaries of multiple split screens, and performs sequential line thinning on other areas, thereby speeding up line thinning processing and reducing thin lines at the border areas. This prevents misalignment of the converted image.

[Operation] In the processing unit 10, the data in the image data storage unit 14 in which the image data of the entire screen is initially stored is divided into predetermined sizes by the two-screen dividing means 11 so that they overlap at the boundary.

The width of the overlapping part corresponds to the width of the line to be thinned.

When the line width is wide, the width of the overlapping part becomes wide. The data of each split screen area is stored in the split table 15.

Next, line thinning processing is performed, and the image data included in one split screen area to be thinned is loaded from the image data storage unit 14 to the memory 16, and first, the data in the superimposed area is Parallel thinning processing means 12
Parallel thinning processing is executed. This parallel type thinning processing means 12 performs line thinning processing by scanning the data of the five overlapping areas using a mask pattern in the same manner as in the conventional method. This thinning result is stored in the memory 16. In this case, either method may be used, such as rewriting the original data of one overlapping area with the thinning process result or storing the thinning process result in another area within the memory 16.

When the thinning process of the superimposed area is completed, the sequential line thinning processing means 13 is activated as a primary step, and performs the sequential line thinning process on the data of the portion of the two-divided area excluding the superimposed area using the same method as before. It will be done.

At this time, the thinning result of the superimposed area by the parallel thinning processing means 12 is stored in the memory 16, and when tracing the outline of the figure by this sequential thinning process, the thinning process result of the superimposed area is continuously used. and refer to it. In this case, since the portion of the double overlapping area has already been thinned, the data subject to the one-sequential thinning process and the data for which the parallel thinning process has been executed are processed consecutively, and the line in the one double area is continuity is maintained.

The results of the processing by the sequential thinning processing means 13 are stored in the memory 1.
It is stored in 6.

[Example] Fig. 2 is a block diagram of the broken line approximation device of the embodiment, Fig. 3 is a diagram showing a specific example of division, Fig. 4 is a diagram showing an example of a screen division table, and Fig. 5 is an example. Processing flow diagram, Figure 6(a)
A specific example of parallel thinning processing FIG. 6(b) is a specific example of sequential thinning processing in which the processing of overlapping areas is changed.

In FIG. 2, 920 is a magnetic disk that stores original image data, 21 is a control unit that controls the flow of processing, and 22 is a screen division table creator that creates a screen division table from information about the size of the original image and the line width in the original image. 23 is an image loading section 524 that loads image data into the image memory 27 according to the division table; 25 is a sequential thinning section; and 26 is a broken line that performs near-line processing from the thinned image. 27 is an image memory.

To explain the operation, input image data is temporarily stored on the magnetic disk 20. This can be loaded into the image memory at once with a normal wire thinning mechanism, but since the present invention uses an image memory with a small capacity, it is necessary for temporary storage on the magnetic disk 20.

Next, the size of the image data manually created in the one-screen split table creation unit 22, the size of the necessary superimposed area between each split screen (proportional to the line width of the superimposed area), and the implementation size of the image memory 27. From this, a screen division table, which is information for dividing input image data, is created.

An example of screen division is shown in FIG.

A in FIG. 3 shows the entire screen, and when this is divided into six parts, each divided screen includes a portion expanded by a width d with respect to the boundary line of each divided screen. An example of a split screen is shown in B in the figure and represents the center/upper part of A. As shown in this figure, a band-shaped area with a width d inside and outside the boundary line overlaps with the adjacent split screen, and in this example, it is parallel to the parallel thinning areas 1 to 3 as shown in the figure. Type thinning processing is performed.

An example of the screen division table is shown in FIG.

The division table 40 is provided with division tables 1 (41) to N (4N) corresponding to each of the plurality of division screens, and each division table stores area information of each division screen. FIG. 4 shows the contents of the information stored in the partition table 1 (41).

That is, first, there is a flag 4111 indicating whether this split screen has been processed or not, upper left coordinates and lower right coordinates 412 of the split screen as information representing the entire area of the split screen, and then an area that does not include the margin (boundary line of the split screen). 4131 Upper left and lower right coordinates of the area that is to be thinned sequentially (inner area excluding overlapping areas within the divided area) 414. Furthermore, the upper left coordinates and lower right coordinates 415 to 418 of each of the four regions 1 to 4 (each rectangular) on the top, bottom, left, and right are stored as regions for parallel thinning (regions that overlap with adjacent regions). be done.

Here, as mentioned above, overlapping parts with other split screens may occur in four locations, top, bottom, left, and right, so areas for these four locations are reserved in this split table; When the screen is divided into six as shown in A in the figure, the division table corresponding to the divided screen shown in B has overlapped portions with adjacent divided screens only in three places: left, bottom, and right. This split table is created by the split screen number frame two screen split table creation unit 22 (second time).

After the screen division table is created, thinning is performed for each area registered in this table.

An example of this processing flow is shown in FIG.

This flow is a diagram expressing a well-known flow called a YAC chart. 1For example, in the case of rlF, the contents of the condition are shown on the right side, and if one condition is satisfied, the direction is shown at the bottom right of the letters rlFJ as rYESJ. If NO, proceed to the next rI from the tip of the letter rlFJ downwards.
We are now proceeding to FJ and DOJ processing.

The contents of the processing flow shown in FIG. 5 will be explained below with reference to FIGS. 2 to 4.

First, the control unit 21 determines whether or not an unprocessed area remains in the screen division table by referring to each flag 411 of division tables 1 to N shown in FIG. 4 (51) in FIG. The screen doesn't exist.

Proceed to [THINN-END) to complete thinning.
2) If it exists, the portion indicated by the unprocessed partitioned table is loaded from the magnetic disk 20 into the image memory 27 by the image loading unit 23 (53).

The flow of thinning processing for the image loaded into the image memory 27 is executed in a thinning loop starting from 54 and ending at 68 in FIG. ~6l) Second, sequential thinning processing is executed (62 to 66 in FIG. 5).

The parallel thinning process is executed in the parallel thinning unit 24 shown in FIG. 2, and first it is determined whether there is an overlapping area in the upper part of the split screen (54 in FIG. 5).

If there is, perform parallel thinning on the overlapping area.

A flag is attached to the deleted pixel (55). This flag is used because if the target pixel is actually deleted, the boundary between the area to be subjected to parallel thinning and the area to be subjected to sequential thinning will not be a correct thinned image.

When the thinning of the overlapping area at the top of the split screen is completed (or if there is no overlapping area at the top), it is then determined whether there is an overlapping area at the bottom of the split screen (see 56), and if there is, the same process is performed. Performs parallel thinning processing (57). Thereafter, parallel line thinning processing is similarly performed on the right and left parts of one split screen (steps 58 to 61).

Thereafter, the sequential thinning process is started in the sequential thinning section 25 shown in FIG. This process performs sequential line thinning processing while performing tracing so that all the figures in one screen go around the circle of two figures (62.63). If this tracking enters the overlapping area (ibid. 64),
Only tracking is performed, but if a flagged pixel is detected in the above parallel thinning process, that pixel is deleted (see 65). If it is not a single overlapping area, sequential thinning is performed (see 66).
). Then, it is determined whether or not there are still points that can be thinned for all of one screen, and if there are, the process returns to 54 and restarts the thinning loop (67.68).

Finally, when there are no more points that can be thinned, the broken line reply process is executed in the broken line approximation unit 26 in FIG.
0).

Next, the details of the thinning process are shown in Figure 6 (a) and Figure 6 (1)).
This will be explained using a specific example shown in .

FIG. 6(a) shows a specific example of parallel thinning processing.

FIG. 6(b) shows a specific example of sequential thinning processing.

In FIG. 6(a), A indicates the image data of the original picture before thinning, and the symbol ``■'' indicating one image indicates a black pixel with a pixel value of 1, and other white pixels are not shown.

When the parallel line thinning process shown in FIG. 4
(Figure), and by referring to it, the area for parallel thinning is set. In parallel type thinning, pixels are not actually deleted, but only the pixel values of the pixels to be deleted are changed (flags are attached) as shown in B of FIG. 6(a). In the example shown in the figure, the pixel value is changed to ■.

In order to change 3 pixel values in this way, each pixel in the area where parallel thinning processing is performed should be assigned at least 2 focus points (the 2nd focus corresponds to a flag), whereas each pixel in the area where parallel thinning processing is performed is usually represented by 1 bit. good.

Next, in the sequential thinning process, the internal image data other than the nine overlapping areas are thinned. In this case, the pixel to be deleted is actually deleted, and if a flagged pixel is detected by thinning while tracing the overlapping area, that pixel is deleted while tracking.

This situation is shown at C0 and C0 in FIG. 6(a).

By repeating this process for all the figures in one screen until there are no more lines to be thinned, it is possible to thin lines without discrepancies at the boundaries.

In the configuration of the embodiment shown in FIG. 2 above, the parallel thinning section 24
Although the and sequential thinning unit 25 are shown as separate processing units,
It is possible to configure it as one processing unit by using a general-purpose processor and switching the firmware, but it is more advantageous to have one processing unit in terms of the cost of the device.

Furthermore, in the above embodiment, a flag was attached to the deletion pixel in the parallel thinning process (55, 57, 59, 61 in FIG. Since it is necessary to allocate at least two focal points to each pixel in order to distinguish between the two, the capacity of the image memory increases. In order to avoid this, a method of not attaching a flag to one pixel will be explained using FIG. 6 et al.

When parallel thinning is performed on the image data of the original image shown in A of FIG. 6(b), the pixels targeted for deletion by thinning are deleted in the overlapping areas other than the border areas, but the same As shown in B in the figure, for the boundary part (the part separated by the dotted line in the figure), only the thinning results (consisting of data to be deleted and data not to be deleted) are stored in different working memories (1 not shown). evacuate to.

After this, sequential thinning processing is executed, but for the two overlapping regions, only tracking is performed as shown in C9 in Fig. 6 (G) (in 65 in Fig. 5, flag processing is not performed). After the sequential thinning is completed, the pixels in the boundary area saved in the working memory are returned to the original boundary area to obtain the final state shown in the same figure. The results obtained by this method are clear from the word diagram, which is the same as the method shown in FIG. 6(a).

[Effects of the Invention] According to the present invention, line thinning processing that requires inputting a large drawing into a computer can be realized without using a large capacity image memory, and at the same time thinning of a split screen can be performed at the boundary between adjacent screens. It is possible to prevent the thin lines from shifting.

Furthermore, line thinning by screen division can be performed by sequential line thinning suitable for processing by a general-purpose processor, which has the effect of improving processing speed and eliminating the need for hardware.

[Brief explanation of drawings]

Fig. 1 is a diagram showing the principle configuration of the present invention, Fig. 2 is a diagram showing the configuration of a broken line approximation device according to an embodiment, Fig. 3 is a diagram showing a specific example of division, and Fig. 4 is a diagram showing an example of a screen division table. , FIG. 5 is a processing flow diagram of the embodiment, FIG. 6(a) is a specific example of parallel thinning processing, FIG. 6(b) is a specific example of sequential thinning processing, and FIG.
The figure is an example of dividing thinning, FIG. 8 is an explanatory diagram of screen division with an overlapping area, FIG. 9 is an example of thinning of a divided screen with an overlapping area, and FIG. 10 is a schematic diagram of line thinning. FIG. 11 is an explanatory diagram of problems in the sequential thinning method. In Figure 1. ] 0: Processing section 11: Image dividing means 12: Parallel thinning processing means I3: Sequential thinning processing means 14: Image data storage section 15: Division table 16: Memory

Claims (1)

[Claims] A binary method that divides the processing target screen into a plurality of regions, performs thinning on the image of each divided screen, and integrates each thinned divided image to obtain an entire thinned image. In the image thinning method, the image data of the entire screen is stored in the image data storage section (14), and the processing section (10) divides the entire screen into predetermined sizes so as to have regions that overlap with adjacent divided images. partitioned table (
15); and parallel thinning processing means (12) that retrieves the image data of the split screen into the memory (16) and performs parallel thinning on the image in the superimposed area of each split screen. and sequential line thinning processing means (13) that performs sequential line thinning on areas other than the overlapping area of each split screen and combines the processing results obtained by the parallel line thinning. Features a thinning method for binary images.