JPS61145689A - Area labeling circuit - Google Patents

Abstract

PURPOSE:To label efficiently many separating picture areas extracted out of a picture by detecting the label, which is already unnecessary, in accordance with the scanning of the binary picture, and reusing the label which is decided to be unnecessary. CONSTITUTION:A label number table 13, when an unnecessary label deciding part 15 detects the unnecessary label in accordance with the scanning of the binary picture in a continuous mode deciding part 11, stores a label number, which is decided to be unnecessary, to an address designated by a table address counter 12. In the counter 12, receiving the control of the deciding part 11, each time one label is read from the table 13, the count value is +1, and each time the deciding part 15 detects the unnecessary label, the count value is -1. By such a counter 12, the address of the table 3 is designated, and the label number stored in said address is read at an intermediate label generating part 14. Consequently, the label number, which is decided to be unnecessary and registered in the table 13, can be read and reused for labeling.

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention is capable of extracting desired separated image regions from a binary image and effectively attaching mutually distinguishable labels to each separated image region. Regarding area labeling circuits.

[Technical background of the invention and its problems]

In recent years, image processing technology has come to be applied to various diagnostic and measurement devices. For example, in cytodiagnosis and particle measurement, etc.
A method is to extract only a partial image of a specific shape and size from an input image as a separate image region, and measure this. In this case, a label is attached to each separated image area in order to mutually identify the plurality of separated image areas extracted from the input image.

For example, from a binary image 1 as shown in FIG. 6(a, ),
A collection of pixels whose pixel data is "1" is extracted as a separate image area, and a label number is sequentially assigned to each separated image area (pixel group forming the separated image area) as shown in FIG. The labeling is done accordingly.

Extraction of this separated image area is carried out, for example, by scanning the binary image and sequentially inputting its pixel data, using a mask 2 as shown in FIG. This is done by determining gender. Specifically, since pixel data is input sequentially by scanning the binary image, pixel data GO of a scanning point C, pixel data GA of a pixel A adjacent to that pixel C and one scanning line before, and From the pixel data GB of the previous scanning point B, the isolation of the pixel C when the pixel C is indicated by pixel data "1" is determined, and the binary values are determined while mutually grouping pixels that are connected to adjacent pixels. This is done by detecting isolated clusters of pixel data "1" in the image.Then, by sequentially assigning label numbers to the separated image regions extracted in this way, the image shown in FIG. It becomes possible to obtain a labeling image 3 as shown in b).

However, such labeling processing has conventionally been performed by a circuit configured as shown in FIG. In FIG. 8, the connection mode determination unit 4 is a block that determines the connectivity of a scanning point pixel with adjacent pixels using the above-mentioned mask 2, and the intermediate label generation unit 5 Detected separated image I! This block sequentially attaches label numbers indicated by the label number counter 8 to the areas. The label number counter 6 is incremented each time a label number is attached to one separated image area, and is controlled so that new label numbers are attached one after another to the separated image areas that are sequentially extracted.

However, in the conventional area labeling circuit configured as described above, due to hardware constraints of the label number counter 6, for example, the available label numbers are
Limits are set, such as from 0) to (511). However, there are many cases in which a number of separated image regions far exceeding the above-mentioned constraint number are extracted among measurement objects, and in such cases, the above-described labeling process is hindered.

Conventionally, the input image 1 is divided into a plurality of processing target regions in advance, and the above-mentioned labeling processing is performed for each processing target region. However, this process requires a large amount of time, etc. There was a problem.

It has also been considered to increase the capacity of the label number counter 6, but it cannot be denied that the hardware configuration of the circuit would become considerably complicated.

(Object of the Invention) The present invention has been made in consideration of the above circumstances, and its purpose is to provide a simple configuration that can efficiently label a large number of separated image regions extracted from an image. The object of the present invention is to provide a region labeling circuit.

[Summary of the invention]

The present invention scans a binary image, sequentially inputs pixel data, detects separated image areas in the binary image from the connection relationship of pixel data with adjacent pixels, and mutually identifies each separated image area. In a region labeling circuit that attaches an identifiable label,
For example, a label that is no longer needed due to reasons such as the processing of a previously labeled separated image area having been completed and the need to use the label attached to that separated image area, etc.
Depending on the scanning of the binary image, for example, 1 of the binary image
The present invention is characterized in that when scanning of a line is completed, labels that have appeared on the current line and the previous line are determined and detected, and the labels that have been determined to be unnecessary are reused.

For example, to assign a label to the separated image area, a label table storing a plurality of labels is addressed by a table address counter and the labels are selectively read out, and one label is read from the label table. Each time the label is output, the table address counter is incremented (+1) to control reading of the label. When the above-mentioned unnecessary label is detected, the above-mentioned table address counter is decremented (-1), and the above-mentioned unnecessary label is registered at the address of the above-mentioned label table specified by this decremented counter. By reading out the registered labels again from the label table, the unnecessary labels can be reused.

〔Effect of the invention〕

Thus, according to the present invention, since so-called used labels are reused, it is possible to attach distinguishable labels to many separated image areas without increasing the absolute number of labels. Moreover, labeling can be performed by simple control using a label table and by a relatively small-scale circuit configuration. Therefore, even when a large number of separated image regions are extracted from an input image, there is no need to take measures such as dividing the processing target region in advance as in the conventional method, so the processing time does not increase. , and there is no problem of complicating the processing, and a great practical effect can be achieved.

[Embodiments of the invention]

Hereinafter, one embodiment of the present invention will be described with reference to the drawings.

FIG. 1 shows a schematic configuration of an embodiment circuit, which basically scans an input binary image, sequentially inputs the image data, and uses, for example, the mask 2 as described above, between pixels. The connection mode determination unit 11 determines the connectivity of the labels selectively read from the label number table 13 in response to the address designation of the table address counter 12.
An intermediate label generation unit 14 attaches to scanning point pixels according to the connection mode between pixels determined and detected, and an unnecessary label determination unit 15 determines and detects labels that are already unnecessary in synchronization with scanning of the binary image. configured.

The connection mode determination unit 11 and the intermediate label generation unit 14 are configured in the same manner as in the conventional 0 path, and when the pixel data of a scanning point pixel of a binary image is “1”, the pixel data of the adjacent pixel is It is determined whether or not is "1". If it is detected that the pixel is not connected to the adjacent pixel, it is determined that the pixel has appeared for the first time as a separated image area, and the intermediate label generation unit 14
A label read out from the label number table 13 is attached to the label. In addition, if it is determined that the above pixel is connected to its adjacent pixel, since a label has already been attached to the adjacent pixel before scanning, the label is referred to and the same label is attached. are doing.

In this case, since the connectivity in four directions is determined using the mask 2, the pixels A and B, which were given different labels, are connected by the appearance of the pixel C, and the In some cases, it may be determined that there are two separate image areas. In such a case, the intermediate label generation unit 14 performs label correction, such as unifying the labels to those with smaller label numbers, for example.

Specifically, for example, as shown in the processing algorithm in FIG. 2, when the number of pixels in one scanning line is (512), first, the pixel control value N is initialized to zero (0) (step 21), It is determined whether the pixel data of the processing target pixel (pixel at the scanning point position) C of the mask 2 is "1" (step 22). And if pixel C is not “1” (
If the pixel is "0"), the pixel control value N is incremented as not the pixel to be labeled (step 23), and it is determined whether the value N is the final pixel of one scanning line (step 24). , confirm that it is not the final pixel, and perform the processing for the next scanning point pixel in step 2 above.
Repeat from 2.

If the pixel C is "1", the next step 25 is to set the pixel data of the pixel A one scanning line before to "1".
”, and further, depending on the determination result, in step 26 or step 21, pixel B one scanning point before is determined.
It is determined whether the pixel data of is "1". When the pixel C is “1” by these determination processes,
Connectivity with adjacent pixels is classified into the following four modes.

One of the cases is when pixels A and B are both °゛0''. In this case, it is determined that pixel C forms a newly appeared separated image area, and the process shown in step 28 The label number stored in the specified address TA is read from the label number table 13 and labeled.Then, in step 29, the specified address TA is incremented and connected to step 23, and the next step is performed. Focus on processing pixels.

If only pixel B is "1", it is determined that pixel C is connected to pixel B to form the same separated image area, and in step 30, pixel C is assigned the same label number as pixel B. After adding , the next pixel is processed. Similarly, if only pixel A is 111 II, pixel A
The label number assigned to pixel C is assigned to pixel C.

Furthermore, if the pixels A and B are both "1'', the pixel C
is determined to be connected to both of them. In this case, different label numbers may be attached to the pixels A and B before the processing, so first, in step 32, the label numbers are mutually determined, and, for example, the label with the lower number is to be unified. Then step 3
In step 3, the above-mentioned unified label number is attached to the pixel C, and processing for the next pixel is reserved. In this way, while the previously attached labels are corrected, the same labels are sequentially attached to a plurality of interconnected pixels.

In this way, the intermediate label generation unit 14 attaches a new label read from the label number table 13 only when a pixel that has no connectivity with the previously scanned pixel appears, and adds a new label to the previously scanned pixel. When a pixel that has connectivity with a pixel that has been identified appears, the label attached to that pixel is referred to and the same label is assigned.

On the other hand, the label number table 13 has, for example, 256 addresses, and is configured by storing label numbers (0) to (255) in advance in the address area. The label number table 13 stores the label number determined to be unnecessary at an address designated by the table address counter 12 when the unnecessary label determination section 15 detects an unnecessary label as described later. It becomes.

Therefore, the table address counter 12 that controls the address of the label number table 13 is configured to count each time one label is read out from the label number table 13 under the control of the connection mode determining section 11, for example. The value is incremented (+1), and each time the unnecessary label determining section 15 detects an unnecessary label, the count value is decremented (-1). Writing (registration) of the unnecessary label mentioned above to the label number table 13
is performed in synchronization with this decrement control, and the label determined to be unnecessary is registered at the decremented address. A table like this
A label number table 13 in which labels determined as unnecessary by the address counter 12 are registered as described above.
is addressed, and the label number stored at the corresponding address is read out to the intermediate label generation section 14. Therefore, even label numbers that are determined to be unnecessary and registered in the label number table 13 are read out under the address control and reused for labeling.

By the way, the unnecessary label determining unit 15 determines that a label is unnecessary, for example, as follows. FIG. 3 shows an example of the processing algorithm.

Now, while performing labeling processing by scanning the binary image, image processing (
(measurement, etc.) are performed simultaneously based on the label number, it is generally possible to complete processing for several extracted separated image regions each time one line scan is completed.

In this case, since the label number assigned to the separated image area for which processing has been completed is no longer necessary for subsequent processing, the label number can be determined to be used and unnecessary.

Therefore, in the unnecessary label determination unit 15, each time the scanning of one line of the binary image is completed, first, in step 41, the unnecessary label determination unit 15 initializes the label determination units 11JIl (III) such that, for example, the label determination unit 11JIl (III) is determined to be unnecessary in order from the label with the largest label number. .

Then, in step 42, it is checked whether the label number indicated by the control value I has appeared in the previous line, and then in step 43, it is similarly checked whether it has appeared in the scanning of the current line. Then, the label number appears on the previous line,
If it does not appear in the current line, it is determined that this is attached to the separated image area extracted up to the previous line and is no longer needed at this point. When an unnecessary label is determined and detected, in step 44, the address of the label number table 13 pointed to by the table address counter 12 is decremented (-1) t, and the label number determined to be unnecessary is set at that address. is registered (step 45). Thereafter, the label determination control 111i1■ is decremented (step 46), and it is confirmed that the control value I does not reach zero (step 41), and the processing from step 42 described above is repeated.

Therefore, if multiple unnecessary labels are detected during this unnecessary label determination, each time an unnecessary label is detected, the above address is decremented and the label number determined to be unnecessary is refined to that address. become.

As described above, in the circuit of this embodiment, when scanning an input binary image, sequentially extracting separated image regions based on pixel connectivity, and assigning label numbers, unnecessary labels are determined for each line scan. While registering label numbers that are determined to be unnecessary in the label number table 13, the addresses of the label number table 13 are controlled and label numbers are attached to newly extracted separated image areas one after another. This makes it possible to effectively reuse label numbers that have been determined to be unnecessary.

Next, a specific example thereof will be explained with reference to FIGS. 4 and 5.

FIGS. 4(a) and 4(b) show a comparison between the input binary image 16 and the labeling image 11 labeled by this 21-image circuit, and FIGS. 7 is a diagram showing the transition state of the label number table 13 during labeling processing and the relationship between its designated addresses. FIG. Note that "1" in the binary image 16 indicates a pixel whose pixel data is "1", and the pixel data of other pixels is "O". Also, (1) in labeling image 11
, +2).

... indicates a label number attached to a pixel.

When starting scanning the binary image 16, the label number table 13 is first initialized as shown in FIG. 5(a), and its first address is designated as shown by the arrow (→). . When the first line PO of the binary image 16 is scanned in this state, no pixel with pixel data "1" is detected in this scanning line, so no label is applied as described above. Therefore, there is no change in the state of the table 13.

Thereafter, when the second line P1 is scanned, three "1" pixels are detected in succession.

However, in this case, first, the label number (1) specified by the address in the label number table 13 is assigned to the first pixel. By assigning this label number, the designated address for the label number table 13 is incremented. When trying to attach a label to the next pixel, since that pixel is connected to the previous pixel, the next label number is not read out from the label number table 13, and one The same label number @(1) as the previous pixel is assigned. Therefore, at this time,
The state of the label number table 13 is maintained as it is.

Labeling is then performed on the next pixel in the same manner.

Thereafter, a pixel of 1 is detected again after several pixels. In this case, since there is no connectivity with the previous pixel and the pixel on one line, it is determined that the pixel has newly appeared, and the next label number @ from the label number table 13 is determined.
('2 is read out and given to the pixel. Similarly,
A label number (3) is read out from the label number table 13 and assigned to a pixel "1" detected after a few more pixels. In this line P1, since the scanning ends after extracting three pixel groups, reading of the label number table 13 is temporarily stopped when the fourth address is specified.

When one line of scanning is completed in this manner, unnecessary label detection processing is performed by the unnecessary label determining section 15. However, in this case, since labeling is not performed on the previous line PO, unnecessary labels are not detected. Therefore, scanning of the next line P2 is started.

However, when scanning line P2, the pixel data is “
Since each pixel detected as 1" has connectivity with the pixel one line before, new labeling is not performed, and the pixels are associated with the pixel one line before and given the same label number. Thereafter, when the unnecessary label is determined upon completion of the scanning of line P2, label number (2) is detected as an unnecessary label. As a result of the detection of this unnecessary label, as shown in FIG. As shown, the designated address in the label number table 13 is returned by one (decremented), and the label number (a) that was determined to be unnecessary is registered at that address.Then, this label number (2) is read out next. controlled.

However, when the next line P3 is scanned, no pixel detection is performed on this line. As a result, labeling is naturally not performed, and the process proceeds to the unnecessary label determination process described above. In this unnecessary label determination process, first, the label number (a) is detected as an unnecessary label, the address of the label double 13 is decremented, the label number (3) is registered to that address, and then the label number (a) is detected as an unnecessary label.
1) is detected as an unnecessary label. Also for this label number (1), the address is decremented and the label number (1) is registered at that address. This state is shown in FIG. 5(C).

After that, when scanning the next line P4, a new “1” is generated.
Since the pixels are detected, they are read out in order starting from the label number of the designated address shown in FIG. 5(C), and labeled with label number (1) i31.

In other words, the label number (1) i3 that was used first and then became unnecessary will be reused. Then, at the end of scanning line P4, label number table 13
will wait with the third address specified, as shown in FIG. 5(d).

After that, when scanning the next line, the label number (2)
, (4) are read in order, and after the state of the label number table 13 changes to the state shown in FIG. 5(e), unnecessary label determination is performed again. And at this time, the label number (
1) is again detected as an unnecessary label and is re-registered in the label number table 13 as shown in FIG. 5(f).

From now on, such labeling control and unnecessary label determination,
This and the re-registration are sequentially repeated, and a labeling image 17 as shown in FIG. 4 (1) is obtained in which the label numbers determined to be unnecessary are reused.

As described above, according to the present circuit, it is possible to effectively control label numbers with a relatively small-scale circuit configuration, detect unnecessary labels, and reuse them.

Therefore, even if more separated image regions are extracted than the number of labels prepared in the label number table 13, these separated image regions can be effectively labeled.

Further, it is possible to achieve great practical effects, such as not causing problems such as the hardware configuration becoming a disaster or the processing time becoming longer than in the past.

It should be noted that the present invention is not limited to the embodiments described above, and can be implemented with various modifications without departing from the gist thereof. For example, labels do not need to be given as numbers, as long as they are mutually distinguishable. Further, the number of labels etc. may be determined according to specifications such as the number of pixels of both images to be processed. Furthermore, it is of course possible to use a mask that determines the connection of pixels in eight directions.

[Brief explanation of the drawing]

FIG. 1 is a schematic diagram of a circuit according to an embodiment of the present invention, FIG. 2 is a diagram showing an example of a processing algorithm for extracting separated image regions and labeling thereof, and FIG. 3 is a processing algorithm for detecting unnecessary overlaps. Figure 6 is a diagram showing an example of the transition state of the bell number table, Figure 6 is a diagram showing conventional labeling processing,
FIG. 7 is a diagram showing an example of a mask for detecting pixel connectivity, and FIG. 8 is a diagram showing an example of the configuration of a conventional area labeling circuit. DESCRIPTION OF SYMBOLS 11... Concatenation mode determination unit, 12... Table address counter, 13... Label number table, 1
4... Intermediate label generation unit, 15... Unnecessary label determination unit.

Claims (3)

[Claims] (1) A binary image is scanned, pixel data is input sequentially, separated image areas are detected from the connection relationship of pixel data with adjacent pixels, and a mutually distinguishable label is attached to each separated image area. An area labeling circuit characterized in that the labeling circuit determines and detects labels that are already unnecessary in response to scanning of the binary image, and reuses the labels that have been determined to be unnecessary. (2) Adding a label to a separated image area involves selectively reading out a label table storing a plurality of labels by addressing the label table using a table address counter, and then reading out one label from the label table. Each time reading is performed, the table address counter is incremented to control reading of the label,
To reuse a label that has been determined to be unnecessary, the table address counter is decremented when an unnecessary label is detected, and the unnecessary label is registered at the address in the label table specified by this decremented counter. Claim 1 which is carried out
Area labeling circuit as described in Section. (3) Area labeling according to claim 1, wherein the unnecessary determination of a label is performed by determining the labels appearing on the current line and the previous line when one line scanning of the binary image is completed. circuit.