JP2007228139A - Labeling method, labeling device, labeling program, and method for eliminating dust from digital image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a labeling method, labeling device, and a labeling program for extracting a region on a binary image in a short time. <P>SOLUTION: The labeling method comprises a first processing stage and a second processing stage. At the first processing stage, an effective pixel of a binary image composed of effective pixels and ineffective pixels is selected as a pixel of interest in order, and a label not overlapping with the one established at other effective pixels is established at the selected pixel of interest if the pixel of interest has no established label. The second processing stage is executed whenever the pixel of interest is selected at the first processing stage. At the second processing stage, setting processing or associating processing is performed. The setting processing is executed to establish the same label as the pixel of interest at an effective pixel located near the pixel of interest 8 if the effective pixel has no established label. The associating processing is executed to associate the label of the effective pixel located near the pixel of interest 8 with the label of the pixel of interest 8, if the effective pixel has an established label different from the pixel of interest 8. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a labeling method, a labeling device, a labeling program, and a method for removing dust from a digital image.

  2. Description of the Related Art Conventionally, there has been known an image reading apparatus that reads a document and outputs a digital image, which can remove dust captured during reading from the digital image (see, for example, Patent Document 1). . However, according to the image reading apparatus of Patent Document 1, there is a problem in that it is a burden on the operator because the operator must perform an operation of removing dust.

Conventionally, a labeling process for extracting a region of a digital image is known. Of the regions extracted using the labeling process, an area composed of a smaller number of pixels than a predetermined threshold is set as a dust area, and the predetermined dust removal process is performed on the dust area, thereby bothering the operator. It is possible to remove dust without any problems.
FIG. 15 is a schematic diagram for explaining a conventional labeling process. In the conventional labeling process, a valid pixel that has not been labeled is selected as a target pixel in order, and a first pixel that sets a label that does not overlap with the label of another valid pixel as the target pixel, and a target pixel in the first loop And a second loop that is executed each time a label is set, and sets a label that is the same as the pixel of interest to an effective pixel that has not been labeled and is in the vicinity of 8 of the pixel of interest. The However, in the conventional labeling process, after setting the same label as the target pixel to the effective pixel that has not been set, which is found first in the second loop, the second loop being executed is interrupted, and the first label is found. A new second loop is called by setting the effective pixel as the target pixel. In the example shown in FIG. 15, the label is set by the first loop only for the effective pixel in the upper left, and the labels are set for all other effective pixels by the chain of the second loop. That is, the conventional labeling process has a multiple loop structure in which another second loop is recursively called from the second loop. In general, since the processing time becomes longer as the number of loops is increased, the conventional labeling process has a problem that a long time is required for the process.

Japanese Patent Laid-Open No. 11-298660

The present invention was created in order to solve the above-mentioned problems, and it is a first object of the present invention to provide a labeling method, a labeling apparatus, and a labeling program for extracting a region on a binary image in a short time. And
A second object of the present invention is to provide a dust removing method for removing dust on a digital image in a short time.

(1) In the labeling method for achieving the first object, effective pixels of a binary image composed of effective pixels and non-effective pixels are sequentially selected as attention pixels, and the selected attention pixels are labeled. Is not set, a first processing step for setting a label that does not overlap with a label set for another effective pixel as the target pixel, and each time the target pixel is selected in the first processing step Setting process for setting the same label as the target pixel for the effective pixel for which the label is not set for the effective pixel in the vicinity of the target pixel in the second processing step And, for an effective pixel for which a label different from the label of the pixel of interest is set, an association process for associating the label of the pixel of interest with the label of the pixel of interest is performed. Comprising a second process stage, the.
If the process of selecting a target pixel in the first processing stage is referred to as a first loop, and the process of selecting an effective pixel in the vicinity of the target pixel in the second processing stage is referred to as a second loop, When the second loop is called from one loop, only the second loop is executed and the process returns to the first loop. That is, in the present invention, the second loop is not called from the second loop, and the loop is multiplexed only up to two layers. Therefore, the processing time can be reduced as compared with the case where the loop is multiplexed three or more times. Therefore, according to the present invention, a region on a binary image can be extracted in a short time.

(2) In the first processing step, all the effective pixels in the upper row of the binary image are selected and then the effective pixels in the lower row are selected, and in each row, the effective pixels are sequentially from left to right. And in the second processing step, the setting process or the association process may be executed only for the effective pixels on the right, lower right, lower and lower left of the pixel of interest.
According to the present invention, since it is only necessary to perform processing not in the vicinity of 8 of the target pixel but in the vicinity of 4 in the second processing stage, the processing amount in the second processing stage can be reduced, and the area on the binary image can be shortened. Can be extracted in time.

  (3) When a magnitude relationship between different labels is defined, and a label is not set for the pixel of interest in the first processing stage, a label larger than a label set for another effective pixel Is set as the target pixel, and in the association process, when the label of the effective pixel is smaller than the label of the target pixel, the label of the effective pixel is associated as a parent label, and the label of the target pixel is associated as a child label. When the pixel label is larger than the target pixel label, the target pixel label may be associated as a parent label, and the effective pixel label as a child label.

(4) In the association process, when the parent label is associated with another label as a child label, the child label may be associated with the other label as a child label.
According to the present invention, since the number of parent-child relationships can be reduced, it is possible to reduce the time required to identify the associated label.

(5) In the association process, when another label is associated with the child label as a child label, the child label and the other child label may be associated with the parent label as a child label.
According to the present invention, since the number of parent-child relationships can be reduced, it is possible to reduce the time required to identify the associated label.

(6) In the association process, when the child label is associated as a child label with another label smaller than the parent label, the parent label may be associated with the other label as a child label.
According to the present invention, since the number of parent-child relationships can be reduced, it is possible to reduce the time required to identify the associated label.

(7) A labeling device for achieving the first object sequentially selects effective pixels on a binary image composed of effective pixels and ineffective pixels as target pixels, and selects the selected target pixels. When a label is not set, a first processing unit that sets a label that does not overlap with a label set for another effective pixel as the target pixel, and the target pixel is selected by the first processing unit. And, for the effective pixels in the vicinity of 8 of the target pixel, for the effective pixels for which no label is set, a setting process is performed for setting the same label as the target pixel for the effective pixel, and what is the label of the target pixel? For effective pixels set with different labels, there is provided a second processing means for executing an associating process for associating the label of the effective pixel with the label of the pixel of interest.
According to the present invention, a region on a binary image can be extracted in a short time.

(8) A labeling program for achieving the first object sequentially selects effective pixels on a binary image composed of effective pixels and ineffective pixels as target pixels, and selects the selected target pixels. When a label is not set, a first processing unit that sets a label that does not overlap with a label set for another effective pixel as the target pixel, and the target pixel is selected by the first processing unit. And, for the effective pixels in the vicinity of 8 of the target pixel, for the effective pixels for which no label is set, a setting process is performed for setting the same label as the target pixel for the effective pixel, and what is the label of the target pixel? For effective pixels set with different labels, a compiling process is performed as second processing means for executing an association process for associating the label of the effective pixel with the label of the pixel of interest. To work over data.
According to the present invention, a region on a binary image can be extracted in a short time.

(9) A method for removing dust from a digital image for achieving the second object includes: generating a binary image composed of effective pixels and ineffective pixels from the digital image; It is executed after applying the labeling method according to any one of claims 1 to 6 to set labels on effective pixels and setting labels on all effective pixels, and is not associated with other labels. For a label, an area composed of effective pixels to which the label is set is extracted as one area, and a label associated with another label is composed of effective pixels to which the associated label is set Extracting a region obtained by combining the regions to be processed as one region, comparing the number of effective pixels of each region with a predetermined threshold, identifying a dust region, and the digital image Of including the steps of performing a predetermined dust removal processing for a region corresponding to the dust region.
According to the present invention, since the area can be extracted in a short time, dust on the digital image can be removed in a short time.

(10) The dust removal process may be a process of applying a median filter.
According to the present invention, since the median filter is applied, dust can be made inconspicuous.

  Each function of the plurality of means provided in the present invention is realized by hardware resources whose functions are specified by the configuration itself, hardware resources whose functions are specified by a program, or a combination thereof. Further, the functions of the plurality of means are not limited to those realized by hardware resources that are physically independent of each other.

Hereinafter, embodiments of the present invention will be described based on examples.
FIG. 2 is a block diagram showing a hardware configuration of a personal computer (PC) 10 as a labeling apparatus according to an embodiment of the present invention. The PC 10 includes a CPU 11, a ROM 12, a RAM 13, a display controller 14, an interface unit 15, a removable memory controller (RMC) 16, an external storage unit 17, and an operation unit 18, which are connected to each other via a bus 19.

  The CPU 11 controls the entire PC 10 by executing programs stored in the ROM 12 and the external storage unit 17. In this embodiment, a dust removal program for removing dust from a digital image is implemented as a part of the scanner driver, and a labeling program is implemented as a part of the dust removal program. The CPU 11 also functions as a first processing unit and a second processing unit by executing a scanner driver. The ROM 12 is a memory that stores various programs and data in advance, and the RAM 13 is a memory that temporarily stores various programs and data. These various programs and data may be downloaded and input from a predetermined server via a communication network, or may be read from a removable memory and input.

The display controller 14 is controlled by the CPU 11 to display a digital image on a display 20 such as a CRT (Cathode Ray Tube), an LCD (Liquid Crystal Display), or an FPD (Flat Panel Display).
The interface unit 15 includes a USB controller and a USB connector for communicating with an external system such as the image scanner 22. The communication standard is not limited to USB, but may be any standard such as IEEE 1394, infrared, Ethernet (registered trademark).

The RMC 16 is connected to the removable memory via a connector (not shown), and controls data transfer between the removable memory and the RAM 13. The removable memory may be a card-type flash memory (so-called memory card), or any other non-volatile storage medium that can be repeatedly written.
The external storage unit 17 includes a hard disk, a hard disk controller, and the like, and stores an operating system (OS), a scanner driver, and other various programs and data.
The operation unit 18 includes a keyboard controller to which a keyboard 21 is connected and a mouse controller to which a mouse (not shown) is connected.

Next, the labeling process will be described.
FIG. 1 is a schematic diagram for explaining the labeling process. In the binary image, each pixel value of the edge image generated based on the luminance component of the digital image is compared with a predetermined threshold value, so that an effective pixel whose pixel value is greater than or equal to the predetermined threshold value and a pixel value less than the predetermined threshold value. It is an image binarized into a certain ineffective pixel. For example, 1 is set as the pixel value for the effective pixel, and 0 is set as the pixel value for the ineffective pixel.

  In the labeling process, first, a first process is executed. In the first process, effective pixels of a binary image are selected as target pixels in order, and if no label is set for the selected target pixel, a label that does not overlap with a label set for another effective pixel is selected as the target pixel. It is a process to set. Here, the label means an integer selected in ascending order from 1 in this embodiment. For example, when 1 to 5 are already set to other effective pixels, 6 is selected as the next label. . In the first process of this embodiment, all the effective pixels in the upper row are selected, and then the effective pixels in the lower row are selected. In each row, the effective pixels are selected in order from left to right.

  As shown in (1) of FIG. 1, the effective pixel (0, 0) is initially selected as the target pixel. The selected pixel of interest (0, 0) is set to “1”, which is a label that does not overlap with the labels of other effective pixels, and the label “1” is displayed in the label management table as shown in FIG. Is registered. At this time, 1 which is the number of pixels for which the label “1” is set is registered in the “number of labels” column of the label management table.

  When the target pixel is selected in the first process, the second process is executed each time. The second process executes a setting process for setting the same pixel as the target pixel for the effective pixel for which the label is not set for the effective pixel in the vicinity of the target pixel, and the label of the target pixel Is a process of executing an associating process for associating the label of the effective pixel and the label of the target pixel with respect to the effective pixel set with a different label. In this embodiment, the association process is referred to as a label parent-child relationship registration process.

  However, in the present embodiment, the setting process or the label parent-child relationship registration process is executed only for the effective pixels that are not in the vicinity of 8 but on the right, lower right, lower and lower left of the target pixel. This is because the first process described above selects all the effective pixels in the upper row and then selects the effective pixels in the lower row, and in each row, the effective pixels are selected in order from left to right. When effective pixels are selected in this order, setting processing or label parent-child relationship registration is performed only by performing setting processing or label parent-child relationship registration processing on only the effective pixels on the right, lower right, lower and lower left. The setting process or the label parent-child relationship registration process can be executed without fail for the effective pixels to be processed. If the process is executed only on the four neighboring effective pixels on the right, lower right, lower and lower left, instead of the eight neighboring neighborhoods, whether or not the pixel is an effective pixel for only the four neighboring pixels of the target pixel. Therefore, the number of pixels to be determined can be halved compared to the case of determining whether or not each pixel in the vicinity of 8 is an effective pixel. Thereby, the processing speed is greatly improved as compared with the conventional labeling processing.

  In the case of the example shown in (1) of FIG. 1, since there is an unlabeled effective pixel (0, 1) under the target pixel (0, 0), the effective pixel (0, 1) is shown in FIG. 0,1) is set with the same label “1” as the pixel of interest (0,0), and as shown in FIG. 3B, the “label number” of the row in which the label “1” of the label management table is registered 1 is added to the column.

In the example shown in (1) of FIG. 1, a label different from the label “1” set for the target pixel (0, 0) is displayed on the right, lower right, lower and lower left of the target pixel (0, 0). Since there is no set effective pixel, the label parent-child relationship registration process is not executed.
Next, returning to the first process, the effective pixel (2, 0) is selected as the next pixel of interest as shown in (3) of FIG. Since no label is set for the selected pixel of interest (2, 0), “2”, which is a label that does not overlap with any other effective pixel label, is set, as shown in FIG. As described above, the label “2” is registered in the label management table.

  Next, a second process is performed on the pixel of interest (2, 0). In the case of the example shown in (3) of FIG. 1, since there is an effective pixel (2, 1) for which no label is set under the target pixel (2, 0), it is effective as shown in (4) of FIG. The same label “2” as the target pixel (2, 0) is set to the pixel (2, 1), and the “label” of the row in which the label “2” of the label management table is registered as shown in FIG. 1 is added to the “Number” column.

In the example shown in (3) of FIG. 1, a label smaller than the label “2” set for the target pixel (2, 0) is displayed on the right, lower right, lower, and lower left of the target pixel (2, 0). Since no pixel is set, the label parent-child relationship registration process is not executed.
If the effective pixel (8, 0) is selected as the target pixel as shown in (5) of FIG. “5”, which is a label that does not overlap with the labels of other effective pixels, is set, and the label “5” is registered in the label management table as shown in FIG.

  Next, a second process is performed on the pixel of interest (8, 0). In the case of (5) in FIG. 1, since there is an unlabeled effective pixel (8, 1) under the target pixel (8, 0), the target pixel (8, 0) is included in the effective pixel (8, 1). The same label “5” is set, and as shown in FIG. 3F, 1 is added to the “label number” column of the row in which the label “5” of the label management table is registered.

  In the case of the example shown in (5) of FIG. 1, the effective pixel (4) smaller than the label “5” of the target pixel (8, 0) is set to the lower left of the target pixel (8, 0) ( 7, 1), a label parent-child relationship registration process for registering label "5" as a child in label "4" is executed. Specifically, as shown in FIG. 3F, 1 is added to the “number of child label numbers” column of the row in which the label “4” of the label management table is registered, and the “child label of the row” The label “5” is registered in the “number” column.

If the effective pixel (2, 1) with the label “2” set as shown in (6) of FIG. 2, 2) is executed.
In the example shown in (6) of FIG. 1, an unlabeled effective pixel (2, 2) below the target pixel (2, 1) and an unlabeled effective pixel (3, 2) located at the lower right. ), The same label “2” as that of the target pixel (2, 1) is set, and as shown in FIG. 3G, the label management table label “2” is registered in the “number of labels” column. 2 is added.

  In the case of the example shown in (6) of FIG. 1, an effective pixel (“1”) smaller than the label “2” of the target pixel (2, 1) is set at the lower left of the target pixel (2, 1). 1, 2), a label parent-child relationship registration process for registering label “2” as a child in label “1” is executed. Specifically, as shown in FIG. 3G, 1 is added to the “number of child label numbers” column of the row in which the label “1” of the label management table is registered, and the “child label” of the row Label “2” is registered in the “No.” column.

Next, returning to the first process, as shown in (7) of FIG. 1, when the effective pixel (4, 1) with the label “3” is selected as the target pixel, the target pixel (4, 4) is selected. The second process for 1) is executed.
As shown in (6) of FIG. 1, since there is an unlabeled effective pixel (4, 2) under the target pixel (4, 1), as shown in (7) of FIG. 2) is set with the same label “3” as the pixel of interest (4, 1), and in the “number of labels” column of the row in which the label “3” of the label management table is registered as shown in FIG. 1 is added.

  In the example shown in (7) of FIG. 1, a label “2” smaller than the label “3” of the target pixel (4, 1) is set to the lower left of the target pixel (4, 1). Since there is a pixel (3, 2), a label parent-child relationship registration process for registering label “3” as a child in label “2” is executed. However, as described above, the label “2” is registered as a child of the label “1”. In this case, the label “3” is registered not as a child of the label “2” but as a child of the label “1”. Specifically, as shown in FIG. 4H, 1 is added to the “number of child label numbers” column of the row in which the label “1” of the label management table is registered, and the “child label” of the row Label “3” is registered in the “No.” column. As a result, the number of parent-child relationships can be reduced, so that the time required to identify the associated label can be shortened.

FIG. 1 (8) and FIG. 4 (J) show the results of omitting halfway and selecting up to the effective pixel (4, 2) as the target pixel.
Next, when the effective pixel (6, 2) with the label “4” set as shown in (9) of FIG. 1 is selected as the target pixel in the first process, the target pixel (6, 2) A second process is executed for.

  As shown in (8) of FIG. 1, there are effective pixels (6, 3) that have not been labeled under the target pixel (6, 2). Therefore, as shown in (9) of FIG. , 3) is set with the same label “4” as the pixel of interest (6, 2), and the “number of labels” column of the row in which the label “4” of the label management table is registered as shown in FIG. 1 is added to.

  In the example shown in (9) of FIG. 1, a label “3” smaller than the label “4” of the target pixel (6, 2) is set at the lower left of the target pixel (6, 2). Since there is a pixel (5, 3), label parent-child relationship registration processing for registering label “4” as a child in label “3” is executed. However, since the label “3” is registered as a child of the label “1” as described above, the label “4” is registered as a child of the label “1” instead of the label “3”. Specifically, 1 is added to the “child label number” column of the row in which the label “1” of the label management table is registered, and the label “4” is added to the “child label number” column of the row. be registered. As a result, the number of parent-child relationships can be reduced, so that the time required to identify the associated label can be shortened.

  Since the label “5” is registered as a child of the label “4” as described above, the label “5” is re-registered as a child of the label “1”, and the child label is registered as the label “4”. It is returned to the state that has not been done. Specifically, the value registered in the “number of child label numbers” column of the row in which the label “4” of the label management table is registered, and the number registered in the “child label number” column are shown in FIG. As shown in 4 (K), the label “1” is moved to each registered line. As a result, the number of parent-child relationships can be further reduced, so that the time required to identify the associated label can be further shortened.

Next, returning to the first process, as shown in (10) of FIG. 1, when the effective pixel (5, 3) with the label “3” is selected as the target pixel, the target pixel (5, 5) is selected. The second process for 3) is executed.
As shown in (9) of FIG. 1, since there is an unlabeled effective pixel (6, 4) at the lower right of the target pixel (5, 3), an effective pixel (6) as shown in (10) of FIG. , 4) is set with the same label “3” as the target pixel (5, 3), and the “number of labels” column of the row in which the label “3” of the label management table is registered as shown in FIG. 1 is added to.

In the example shown in (10) of FIG. 1, there is an effective pixel (6, 3) to which the label “4” is set to the right of the target pixel (5, 3), but the label “4” has already been labeled “ Since it is registered as a child of the label “1”, which is the parent of “3”, nothing is performed in the label parent-child relationship registration process, and the process ends.

Next, returning to the first process, when the effective pixel (6, 3) for which the label “4” is set is selected as the target pixel, the second process for the target pixel (6, 3) is executed. The
In the case of the example shown in (10) of FIG. 1, since there is no valid pixel that has not been labeled in the vicinity of the target pixel (6, 3), the process for setting the label in the neighboring pixel is not executed.

Further, there is an effective pixel (6, 4) for which the label “3” is set below the effective pixel (6, 3) for which the label “4” is set, but the label “4” has already been labeled “3”. ”Is registered as a child of the label“ 1 ”, which is the parent of“ ”, the process ends without performing anything in the label parent-child relationship registration process.
FIG. 1 (11) and FIG. 4 (M) show the result of the labeling process for the binary image that is omitted halfway.

Next, the dust removal of the digital image using the labeling method according to an embodiment of the present invention will be described.
5 and 6 are flowcharts showing the flow of the dust removal processing of the digital image executed by the PC 10.
In S <b> 105, the CPU 11 acquires a digital image from the image scanner 22.

In S110, the CPU 11 generates an edge image of the digital image acquired from the image scanner 22.
In S115, the CPU 11 compares each pixel value of the edge image with a predetermined threshold value to generate a binary image.
In S120, the CPU 11 selects one pixel from the binary image.

In S125, the CPU 11 determines whether the selected pixel is a valid pixel. If the selected pixel is an effective pixel, the CPU 11 sets that pixel as a target pixel and proceeds to S130. If not, the CPU 11 proceeds to S145.
In S130, the CPU 11 determines whether or not a label is set for the target pixel. If the label is not set, the process proceeds to S135, and if it is set, the process proceeds to S140.

In S135, the CPU 11 sets a label that does not overlap with any other effective pixel as the target pixel.
In S140, the CPU 11 executes a second process for the target pixel. Hereinafter, the second process will be described.

7 and 8 are flowcharts showing the flow of the second process.
In S <b> 205, the CPU 11 selects one pixel on the right, lower left, lower and lower right of the target pixel.
In S210, the CPU 11 determines whether the selected pixel is an effective pixel. If the selected pixel is an effective pixel, the process proceeds to S215, and if not, the process proceeds to S245.
In S215, the CPU 11 determines whether a label is set for the selected effective pixel. If the label is not set, the process proceeds to S220, and if the label is already set, the process proceeds to S225.

In S220, the CPU 11 executes setting processing for the selected effective pixel. Specifically, the CPU 11 sets the same label as the target pixel in the selected effective pixel, and sets 1 in the “label number” column of the row in which the label set for the target pixel is registered in the label management table. to add.
In S225, the CPU 11 determines whether the label set for the selected effective pixel is smaller than the label set for the target pixel. If the label is smaller, the process proceeds to S230, and otherwise the process proceeds to S235.

  In S230, the CPU executes a label parent-child relationship registration process using the effective pixel label as a parent label and the target pixel label as a child label. Hereinafter, the label parent-child relationship registration process will be described.

9 and 10 are flowcharts showing the flow of label parent-child relationship registration processing.
In S305, the CPU 11 determines whether a parent-child relationship between the parent label and the child label has already been registered. Specifically, the CPU 11 determines that the child label has been registered if the child label has already been registered in the “child label number” column of the row in which the parent label is registered in the label management table. Further, the CPU 11 determines that the parent label is registered as a child of another label, and the child label is also registered as a child of the other label. If it is not registered, the CPU 11 proceeds to S310, and if registered, the label parent-child relationship registration processing is terminated and returns.

In S310, the CPU 11 determines whether the parent label is registered as a child of another label. If the parent label is registered as a child of another label, the CPU 11 proceeds to S315, and if the parent label is not registered as a child of another label, the process proceeds to S330.
In S315, the CPU 11 registers the child label in the label management table as a child of the other label. Specifically, the CPU 11 adds 1 to the “number of child label numbers” column of the row in which the other label is registered, and registers the number of the child label in the “child label number” column of the row.

  In S320, the CPU 11 determines whether another label is registered as a child in the child label. Specifically, the CPU 11 determines that another label is registered as a child if the value in the “number of child label numbers” column of the row in which the child label is registered in the label management table is 1 or more. If another label is registered as a child in the child label, the CPU 11 proceeds to S325, and if another label is not registered as a child in the child label, the CPU 11 ends the label parent-child relationship registration process and returns.

  In S325, the CPU 11 re-registers another child label registered in the child label as a child of the parent label. Specifically, the CPU 11 uses the value registered in the “number of child label numbers” column of the row in which the child label is registered in the label management table and the number registered in the “child label number” column as the parent. Move to the line where the label is registered.

In S330, the CPU 11 determines whether the child label is registered as a child of another label smaller than the parent label. The CPU 11 proceeds to S335 if the child label is registered as a child of another label smaller than the parent label, and proceeds to S340 if not registered.
In S335, the CPU registers the parent label as a child of the other label. FIG. 11 shows a case where S335 is executed. When the effective pixel (1, 2) is selected as the target pixel in FIG. 11, the label parent-child relationship registration process is executed with the label “2” as the parent and the label “3” as the child. In this case, since the child label “3” should already be registered as a child of another label “1” smaller than the parent label “2”, the parent label “2” is a child of the other label “1” in S335. Will be registered as As a result, the number of parent-child relationships can be further reduced, so that the time required to identify the associated label can be further shortened.
In S340, the CPU registers the child label as a child of the parent label.

Returning to the second process shown in FIG. 8, in S235, the CPU 11 determines whether the label set in the selected effective pixel is larger than the label set in the target pixel. If larger, the process proceeds to S240. Otherwise, the process proceeds to S245.
In S240, the CPU executes a label parent-child relationship registration process using the target pixel label as a parent and the effective pixel label as a child.
In S245, the CPU 11 determines whether selection of the pixels on the right, lower left, lower, and lower right is finished. If the selection is finished, the process is finished. If the selection is not finished, the process returns to S205 to select. Repeat the process until is finished.

  Returning to the first process shown in FIG. 5, in S145, the CPU 11 determines whether selection of all the pixels of the binary image is completed. If selection is completed, the process proceeds to S150 shown in FIG. If not completed, the process returns to S120 and the process is repeated until the selection is completed. S120 to S145 correspond to the “first processing stage” recited in the claims.

In S150, the CPU 11 executes label parent-child relationship reconstruction processing for reconstructing the label management table. Hereinafter, the label parent-child relationship reconstruction process will be described.
FIG. 12 is a flowchart showing the flow of the label parent-child relationship reconstruction process, and FIG. 13 is a schematic diagram showing the label management table after the label parent-child relationship reconstruction process.

In S405, the CPU 11 selects one row in which the child label number is registered in the “child label number” column from the label management table.
In S410, the CPU 11 selects the number of labels registered in the “number of labels” column of the selected row and the “#” of the row in which each child label number registered in the “child label number” column of the selected row is registered. Add the number of labels registered in the “Number of labels” column.

In S415, the CPU 11 adds the “label number” column of the selected row and the “label number” column of the row in which each child label number registered in the “child label number” column of the selected row is registered to S410. Register the totals obtained in step 1 above.
In S420, the CPU 11 determines from the label management table whether selection of all rows in which the child label number is registered in the “child label number” column has been completed. If not, the process is repeated until the selection is completed. If the selection is completed, the label parent-child relationship reconstruction process is terminated. If the label parent-child relationship reconstruction process is performed, the number of labels does not have to be added every time an effective area is determined.

  Returning to the first process shown in FIG. 6, in S155, the CPU 11 executes the third process. Specifically, the CPU 11 extracts, as a region, an area composed of effective pixels in which the label is set for a label that is not associated with another label, and for a label that is associated with another label. Extracts a region obtained by combining regions composed of effective pixels to which an associated label is set as one region. For example, in the case of the example shown in (11) of FIG. 1, three areas 31, 32, and 33 are extracted as shown in FIG.

  In S160, the CPU 11 identifies the dust area by comparing the number of effective pixels in each area with a predetermined threshold. Specifically, for example, the CPU 11 sequentially selects a label number in the reconstructed label management table, and if the value registered in the “label number” column of the selected label number is less than a predetermined threshold, the label An area composed of effective pixels for which is set is specified as a dust area.

  In S165, the CPU 11 performs a predetermined dust removal process on the area corresponding to the dust area in the digital image. Specifically, for example, the CPU 11 applies a known median filter to an area corresponding to the dust area. The median filter is an excellent filter for removing noise caused by a large pixel value protruding from surrounding pixel values or a small pixel value protruding.

  According to the labeling method according to the embodiment of the present invention described above, when S120 to S145 are referred to as a first loop and S205 to S245 are referred to as a second loop, when the second loop is called from the first loop. Only that second loop is executed and returns to the first loop. That is, in the present embodiment, the second loop is not called from the second loop, and the loop is multiplexed only up to a maximum of two. Therefore, the processing time can be reduced as compared with the case where the loop is multiplexed three or more times. Therefore, according to the present invention, a region on a binary image can be extracted in a short time.

  In addition, this invention is not limited to the said embodiment, In the range which does not deviate from the summary, it is applicable to various embodiment.

The schematic diagram for demonstrating the labeling process which concerns on one Example of this invention. 1 is a block diagram of a labeling device according to an embodiment of the present invention. The schematic diagram which shows the label management table which concerns on one Example of this invention. The schematic diagram which shows the label management table which concerns on one Example of this invention. The flowchart which concerns on one Example of this invention. The flowchart which concerns on one Example of this invention. The flowchart which concerns on one Example of this invention. The flowchart which concerns on one Example of this invention. The flowchart which concerns on one Example of this invention. The flowchart which concerns on one Example of this invention. The schematic diagram for demonstrating the labeling process which concerns on one Example of this invention. The flowchart which concerns on one Example of this invention. The schematic diagram which shows the label management table which concerns on one Example of this invention. The schematic diagram of the binary image which concerns on one Example of this invention. The schematic diagram for demonstrating the conventional labeling process.

Explanation of symbols

11 CPU (first processing means, second processing means), 10 personal computer (labeling device)

Claims (10)

When a valid pixel of a binary image composed of valid pixels and non-valid pixels is selected as a target pixel in order, and a label is not set for the selected target pixel, a label set for another valid pixel A first processing step of setting a label that does not overlap with the pixel of interest;
In a second processing step that is executed each time the target pixel is selected in the first processing step, the effective pixels that are not set for the effective pixels in the vicinity of 8 of the target pixel. A setting process for setting the same label as the target pixel for the effective pixel is executed, and for the effective pixel in which a label different from the label of the target pixel is set, the label of the effective pixel and the label of the target pixel are set. A second processing stage for performing an associating association process;
Including labeling method. In the first processing step, all effective pixels in the upper row of the binary image are selected and then effective pixels in the lower row are selected, and in each row, effective pixels are selected in order from left to right. ,
2. The labeling method according to claim 1, wherein in the second processing stage, the setting process or the association process is executed only for effective pixels on the right, lower right, lower and lower left of the target pixel. The size relationship between different labels is defined,
In the first processing step, when a label is not set for the target pixel, a label larger than a label set for another effective pixel is set for the target pixel.
In the association process, when the effective pixel label is smaller than the target pixel label, the effective pixel label is associated as a parent label and the target pixel label as a child label, and the effective pixel label is the target pixel label. The labeling method according to claim 1, wherein, when the label is larger than the label, the label of the pixel of interest is associated as a parent label, and the label of the effective pixel is associated as a child label.   The labeling method according to claim 3, wherein, in the association process, when the parent label is associated with another label as a child label, the child label is associated with the other label as a child label.   The labeling according to claim 3 or 4, wherein, in the association processing, when another label is associated with the child label as a child label, the child label and the other child label are associated with the parent label as a child label. Method.   In the association process, when the child label is associated as a child label with another label smaller than the parent label, the parent label is associated with the other label as a child label. The labeling method described. Effective pixels on the binary image composed of effective pixels and non-effective pixels are selected as target pixels in order, and if no label is set for the selected target pixel, it is set to another effective pixel. First processing means for setting a label that does not overlap with a label as the pixel of interest;
When the target pixel is selected by the first processing means, for the effective pixel in the vicinity of the target pixel, for the effective pixel for which no label is set, the same label as the target pixel is set for the effective pixel. Second processing means for executing an association process for associating a label of the effective pixel and a label of the target pixel with respect to an effective pixel in which a label different from the label of the target pixel is set. ,
A labeling device comprising: Effective pixels on the binary image composed of effective pixels and non-effective pixels are selected as target pixels in order, and if no label is set for the selected target pixel, it is set to another effective pixel. First processing means for setting a label that does not overlap with a label as the pixel of interest;
When the target pixel is selected by the first processing means, for the effective pixel in the vicinity of the target pixel, for the effective pixel for which no label is set, the same label as the target pixel is set for the effective pixel. As a second processing means for executing an association process for associating the effective pixel label and the label of the target pixel with respect to the effective pixel for which a label different from the label of the target pixel is set. A labeling program that makes a computer work. Generating a binary image composed of effective pixels and non-effective pixels from the digital image;
Applying a labeling method according to any one of claims 1 to 6 to the binary image to set a label on effective pixels;
It is executed after setting labels for all effective pixels, and for labels that are not associated with other labels, an area composed of effective pixels for which the label is set is extracted as one area, For the associated label, extracting a region obtained by combining regions composed of effective pixels in which the associated label is set as one region;
Comparing the number of effective pixels in each of the regions with a predetermined threshold to identify a dust region;
Performing a predetermined dust removal process on an area corresponding to the dust area of the digital image;
A method for removing dust from a digital image. The digital image dust removing method according to claim 9, wherein the dust removing process is a process of applying a median filter.

Images