JPH08272980A - Closed area extracting device - Google Patents

Abstract

PURPOSE: To extract a closed area and perform closed area extraction which is faster than before by detecting 1st-4th edge directions. CONSTITUTION: An edge detection part 20 performs specific operation for pixel data and detects whether or not each pixel is an edge. When it is judged that the pixel is an edge, which of the 1st-4th edge directions the pixel has is detected. An extraction part 30 extracts a pixel which is in the edge direction from the pixel detected as the edge by the edge detection part 20 and has the 1st-4th edge directions crossing each other as a pixel in the closed area. Consequently, the operation by the edge detection part 20 which detects the pixel as the edge can be made small. Further, since the crossing of the 1st-4th edge directions is only necessary to be detected, the process of the extraction part 30 is made easy.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for extracting a closed region surrounded by edge portions from input image data. In particular, the present invention relates to a device for extracting a closed target shape having a size within a predetermined range.

[0002]

2. Description of the Related Art First, a conventional shape extraction technique will be described. When extracting a target from image data, it is generally assumed that the target is a bright image in the image data. The edge refers to a portion of the image where there is a rapid change in brightness. Further, the edge direction means a direction from a dark portion to a bright portion in the image. A filter that performs a process of extracting the edge and the direction of the edge is called an edge extraction filter.

FIG. 38 is a diagram for explaining an edge. In the figure, 90 is an edge which is assumed to be the target contour. FIG. 39 is a diagram for explaining the edge direction. FIG. 39A is an enlarged view of the portion 91 shown in FIG. Further, FIG. 39 (b) is an enlarged view of a portion 92 in FIG. In the figure, the shaded area is the dark area in the image, and the plain area is the bright area in the image. The boundary line between the two is an edge, and FIG.
The direction indicated by the arrow in 9 is the edge direction. Figure 40
FIG. 6 is a diagram illustrating a conventional edge extraction process. In the figure, one box in a rectangle corresponds to one pixel. In order to detect whether or not the target pixel e has an edge, an area of 9 pixels from pixel a to pixel i is used. Each pixel is, for example, 0
It has a pixel value of any of .about.255.
FIG. 41 is a diagram showing a conventional X-direction edge extraction filter. FIG. 42 is a diagram showing a conventional Y-direction edge extraction filter. It is detected whether or not there is an edge by multiplying the filter shown in FIGS. 41 and 42 with the pixel value of the region of the target pixel shown in FIG. The calculation formula of the Sobel operator is shown in FIG. After the values of X and Y are obtained by the calculation formula shown in FIG. 43, if the sum of the absolute value of X and the absolute value of Y is larger than the threshold value T, the pixel e is determined to have an edge. . FIG. 44 shows the criteria for determining the edge direction. As shown in FIG. 44, the edge direction is classified into eight directions and determined according to the magnitude of each value of X and Y. FIG. 45 is a diagram showing values for quantizing the edge direction.

In the conventional shape extraction method, it is particularly assumed that the target is a circle or an ellipse. In circles and ellipses, all edge directions are toward the center. That is, a method is adopted in which the edge direction detected by using the edge extraction filter is used to extract a portion where the eight directions intersect. As shown by the dotted line in FIG. 46, the edge directions intersect at the center of the circle. FIG. 47 is a diagram in which line elements “spokes” of length L pixels are extended in the quantized direction 1. In this conventional example, the length of the spoke is set to 7 pixels. That is, the spokes are extended by 7 pixels in the direction indicated by the arrow a from the pixel P11 where the edge is detected toward the quantized direction 1. At this time, each pixel P11 to P17 in the spoke has a value of 1, which is the quantized direction. FIG. 48 shows a spoke when the quantized direction is 5. The direction from the pixel P23 having the edge direction 5 to the arrow P, that is,
Extend the spoke in direction 5 to P17.
As a result, as shown in FIG. 49, the pixel P17 has both a value of 1 and a value of 5, and thus has a value of 6. Similarly, all the spokes extended from other directions also overlap in the pixel P17 which is the center. As shown in FIG. 50, the pixel P17 has 1 + 2 + 3 + 4 + 5 + 6 + 7 +
It has a value of 8 or 36.

In the above description, in order to show the quantized direction in an easy-to-understand manner, it has been expressed by using the numerical values from 1 to 8. However, in reality, a register having a bit corresponding to each edge direction is used. Is used to express the edge direction. FIG. 51 shows the correspondence between the quantized edge direction and the 8-bit register. As shown in the figure, in the register, the 0th bit to the 7th bit are arranged from right to left.
When the edge direction is 1, the 0th bit is 1, and when the edge direction is 2, the 1st bit is 1. Similarly, when the edge direction is 8, the 7th bit becomes 1.
As shown in FIG. 51, when the quantized edge direction is expressed using 8 bits and the spoke is extended, the register of the pixel P17 shown in FIG. 50 has all eight directions. As shown at 52, all bits are turned on, that is, 1 is absent. From this, it is determined that this shape is a circle. Further, FIG. 52 shows only the register of the pixel corresponding to the center when the target is a circle, but the register of the same 8-bit expression is provided corresponding to all the pixels. That is, the number of pixels in the image data of the pixel is the same as the number of registers for each pixel indicating the edge direction.

Next, a case where the direction of the detected edge is a rectangle will be described with reference to FIGS. 53 and 54. FIG. 53 (a) shows an example of the spokes extended from the detected rectangular edge. Moreover, FIG. 53 (b)
Indicates the direction of the detected edge. In such a case, the register value indicates only the edge directions 1, 3, 5, and 7, and the register contents are as shown in FIG. When the rectangle has a slope, the register may have a value as shown in FIG.

56 and 57 show the case where the detected edges form parallel lines. As shown in FIG. 56, when an edge having edge direction 1 and edge direction 5 is detected, the register has a value as shown in FIG. Similar to the case of the rectangle described above, when the parallel lines have an inclination, the register may take a value as shown in FIG. 58 depending on the inclination.

[0008]

Since the conventional target extraction apparatus is constructed as described above, it has the following problems. The use of the matrix-shaped edge extraction filter requires the calculation of 3 × 3 pixels, that is, 9 pixels for the edge detection of one pixel and the detection of the edge direction, resulting in a problem that the calculation time is long. Also, a large storage capacity is required. Also,
There is a problem that only special shapes such as circles and ellipses can be extracted, and elongated shapes and complicated shapes cannot be extracted. Examples of shapes that cannot be extracted are shown in FIGS. 59 and 60.
FIG. 59 is an example in which a long and narrow area is not extracted. Also,
FIG. 60 is an example in which a complicated area is not extracted. Also,
There is a problem that only a part of a circle or an ellipse may be extracted. An example of this case is shown in FIGS. 61 and 62. FIG. 61 is an example in which only a part of one region is extracted, and the actual shape is shown in FIG.
The extracted shape is as shown in FIG. 61 (b). Further, FIG. 62 is an example in which one continuous area is divided into a plurality of areas and extracted. The actual area is shown in FIG.
The shape shown in FIG. 62A is shown in FIG. 62B while the shape shown in FIG. These problems are based on the idea that this conventional method is that all edge directions are toward the center.

The present invention has been made to solve the above problems, and has the following objects.
(EN) A closed region extraction device that does not erroneously extract noise of image data as a target. (EN) A closed region extraction device that does not miss a target object or misses the extraction of a target object. (EN) A closed region extraction device that does not mistakenly extract a part of a larger object than a target object as a target object.

[0010]

The closed area extraction device of the present invention has the following elements. (A) a storage unit that stores an image composed of a plurality of pixels based on pixel data; (b) a predetermined calculation is performed on the pixel data stored in the storage unit, and each pixel is stored in the image. In addition to detecting whether the edge is a contour, if it is an edge, in the first to fourth edge directions,
An edge detection unit that detects which edge direction the edge has, (c) a pixel existing in the edge direction from each pixel detected as an edge by the edge detection unit, and the first to fourth edges An extraction unit that extracts pixels whose directions intersect each other as pixels in a closed region.

The extraction unit detects, for each pixel of the image, a register unit having first to fourth registers corresponding to the first to fourth edge directions and an edge detected by the edge detection unit. A register setting unit that turns on a register corresponding to an edge direction of a pixel and turns on a register corresponding to an edge direction of a predetermined number of pixels existing in the edge direction from the pixel, and the register setting unit, ~ It is characterized by comprising a selection unit for selecting a pixel for which the fourth register is set to ON as a pixel belonging to the closed region.

[0012] The edge detecting section is + in the X direction.
An X direction edge extraction filter that detects edges in the X direction and the -X direction, and a Y direction edge extraction filter that detects edges in the + Y direction and the -Y direction in the Y direction orthogonal to the X direction are provided. It is characterized in that an edge and an edge direction are detected by calculating pixel data.

The closed area extraction device further includes a contour confirmation section for confirming that the pixel detected as an edge by the edge detection section is the contour of the closed area based on the edge and the edge direction detected by the edge detection section. It is characterized by having.

The contour confirmation unit traces and confirms that adjacent pixels are edges based on the edge detected by the edge detection unit and the edge direction.

The contour confirmation unit is characterized by checking the existence of an edge having an edge direction opposite to the edge direction of the edge detected by the edge detection unit.

[0016]

In the closed region extracting apparatus of the present invention, the edge detecting section performs a predetermined calculation on the pixel data to detect whether each pixel is an edge. When the pixel is detected as an edge, it is detected which of the first to fourth edge directions the pixel has. The extraction unit is a pixel in the edge direction from the pixel detected as an edge by the edge detection unit, and
Pixels in which the first to fourth edge directions intersect are extracted as pixels in the closed region. According to the present invention, the number of calculations in the edge detection unit that detects pixels as edges is small. Moreover, since it is sufficient to detect that the first to fourth edge directions intersect, the processing in the extraction unit becomes easy.

The extraction section has first to fourth registers for each pixel corresponding to the first to fourth edge directions. The register setting unit turns on a pixel detected as an edge and a register corresponding to the edge direction of a predetermined number of pixels existing in the edge direction of the pixel from the pixel detected as the edge. By performing such a register setting operation for each pixel, the pixels existing in the closed region are present in the first to fourth edge directions, so that all the first to fourth registers are turned on. . The selection unit selects pixels for which the first to fourth registers are set to ON,
Select as a pixel in the closed area.

The edge detection section includes an X-direction edge extraction filter and a Y-direction edge extraction filter. X
The direction edge extraction filter extracts edges in the + X direction and the -X direction. Further, the Y direction edge extraction filter extracts the edges in the + Y direction and the −Y direction.

Further, the closed area extraction device of the present invention includes a contour confirmation section. The contour confirmation unit confirms whether the edge detected by the edge detection unit is the contour of the closed region. The contour confirmation unit can recognize whether the area is a closed area or an open area.

The contour confirmation unit traces and confirms that the adjacent pixel is an edge by tracing the edge detected by the edge detection unit and the edge direction. As a result of the tracking, when the pixel returns to the pixel where the confirmation of the contour is started, it can be confirmed that it is a closed region.

The contour portion is confirmed to be a closed region by confirming the existence of an edge waiting for an edge direction opposite to the edge direction of the edge detected by the edge detecting portion. For example, when a certain pixel has an edge in the + X direction, it is confirmed that a pixel having an edge in the -X direction exists corresponding to the pixel. Also, if a pixel is +
If it is an edge in the Y direction, it corresponds to the pixel −
Make sure that there are pixels waiting for an edge in the Y direction.
If there is an edge having a facing direction with respect to all the edges, it can be confirmed that it is a closed region.

[0022]

【Example】

Example 1. FIG. 1 is a flowchart showing a target finding algorithm according to the embodiment of the present invention. Also in this embodiment, it is premised on the idea that a bright portion in the input image as compared with surrounding pixels is the target. That is, the process of extracting a bright portion (a portion that seems to be the target) in the input image as compared with the surrounding pixels, binarizing the target based on the extracted shape, and outputting it as an image is performed.

This will be described with reference to FIG. First, in S10, an image of a target from which a target is extracted is input. Then S
At 11, edges are extracted using an edge extraction filter. At the same time, the edge direction is also detected. Next, in S12, the target candidate is detected by the filter. Subsequently, in S13, labeling, that is, labeling operation is performed on the detected target candidates. next,
In S14, the target is detected from the labeled target candidates by the moment feature amount. Finally, in S15, the image is output. The binarized image is output.

FIG. 2 is a diagram showing a distribution arrangement of bright portions detected as target candidates in S12 of the above-mentioned flowchart. FIG. 3 shows a state after the labeling is performed in S13. Labeling is a naming used to identify the detected target candidates. A unique identifier is assigned to each target candidate. In FIG. 3, labeling is performed using a combination of numbers and alphabets, but other combinations of characters may be used as long as the target candidates are unique. Further, in the target detection of S14, the target is detected by using the area of the detected target candidate, the moment feature amount (the length of the target, the aspect ratio), and the like.

FIG. 4 is a block diagram showing the configuration of the closed region extracting apparatus of this embodiment. In the figure, 1 is a closed region extraction device, which comprises a storage unit 10, an edge detection unit 20, an extraction unit 30, and a contour confirmation unit 40. The edge detector 20
An X-direction edge extraction filter 22 and a Y-direction edge extraction filter are provided. The X direction edge extraction filter
The edges in the + X direction and the −X direction are detected. The Y direction edge extraction filter detects edges in the + Y direction and the −Y direction. The extraction unit 30 also includes a register unit 32, a register setting unit 34, and a selection unit 36.

First, the edge detector 20 will be described. FIG. 5 is a diagram showing an example of a target formed of a closed region according to this embodiment. In FIG. 5, the rectangle indicated by the dotted line corresponds to one pixel. Further, FIG. 6 is a diagram in which the target edge direction shown in FIG. 5 is represented by arrows. FIG. 7 is a diagram showing an example of the X-direction edge extraction filter used in the embodiment of the present invention. As shown in the figure, it is a linear filter having a length r in each of the + direction and the-direction. FIG. 8 shows a Y-direction edge extraction filter according to the embodiment of the present invention. Similar to the X-direction edge extraction filter shown in FIG. 7, it is a linear filter having a length r in each of the + Y direction and the −Y direction. As shown in FIGS. 7 and 8, the edge extraction filter used in this embodiment has a length of 2r + 1 for one pixel in width. Pixels a, b, c, d, e
Are arranged in the X direction and the pixel of interest is pixel c, X = −a−b + d + e is calculated. Also, the pixels a, b, c, d, and e are Y
When the pixel of interest is arranged in the direction and the pixel of interest is pixel c, Y = −a−b + d + e is calculated. Then, by comparing the values of X and Y with a predetermined threshold value T given in advance by the target to be extracted, if X> T, the pixel of interest c is +
It is detected that the edge has the X direction, that is, the edge direction facing right. When X <-T, it is detected that the edge has a -X direction, that is, an edge direction that is leftward. Similarly, when Y> T, the target pixel c is detected to be in the + Y direction, that is, an edge having an upward edge direction, and Y <
When -T, the target pixel c is detected as an edge having a -Y direction, that is, a downward edge direction. The value of this edge extraction filter is variable. in this way,
The use of the linear edge extraction filter having a width of 1 pixel has an advantage that the calculation time is shorter than that of the conventionally used matrix-shaped (planar) edge extraction filter.

FIG. 9 is a flow chart showing the flow of processing from edge detection to shape extraction by the extraction unit 30. When an edge is detected, in S20, the edge direction is quantized into four vertical and horizontal directions. FIG. 10 is a diagram showing the values in the edge direction quantized in four directions in this embodiment. 1 is + X
Direction, 2 is defined as -Y direction, 3 is defined as -X direction, and 4 is defined as + Y direction. FIG. 11 is a diagram showing a 4-bit register provided for each pixel in this embodiment.
In this embodiment, since the quantization in the edge direction is performed in four directions, the register also has 4 bits for each pixel. The 0th bit corresponds to the edge direction 1, the 1st bit corresponds to the edge direction 2, the 2nd bit corresponds to the edge direction 3, and the 3rd bit corresponds to the edge direction 4. For example, when an image having a resolution of 200 pixels in the vertical direction and 200 pixels in the horizontal direction is input, the number of pixels is 200 × 200 (40000) pixels, and the number of 4-bit registers is 40,000. The register unit 32 is a memory that holds these registers.

Next, the operation of the closed region extracting apparatus of this embodiment will be described in detail with reference to the drawings. FIG. 12 is an enlarged view of the diagram shown in FIG. 6 in which the edge directions are indicated by arrows.
Further, FIG. 13 is a diagram expressed using the quantized value in the edge direction, which has the same size as FIG. Also, FIG.
FIG. 4 is a diagram in which the edge direction is expressed using the register for each pixel shown in FIG. It should be noted that, from this point onward, in the figure, the register of the pixel having no edge direction is displayed as blank. The values in the registers are all 0s, although they are shown as blanks for the sake of clarity. As shown in FIGS. 11 to 14, in the flow chart shown in FIG. 9, in S21, all pixels have four quantized registers. Next, in S22,
The register setting unit 34 sets 1 to the corresponding bit of the register in the edge direction for the length L pixels from the detected edge. FIG. 15 is a diagram for explaining the area extraction of the area L in the range of the length L surrounded by four directions. If the target length to be extracted is less than L pixels, the register setting unit 34
Thus, the target can be detected by inserting the value in the edge direction detected by the length L pixels from the edge portion and extracting the portion where the four vertical and horizontal directions intersect. This will be described in detail with reference to FIGS. 16 to 23. FIG. 16 is a diagram in which a linear filter having a length of L pixels (6 pixels in this embodiment) is extended from a pixel having an edge direction of + X direction. There is a 1 in the 0th bit register, which is the bit corresponding to the quantized direction 1 of each pixel. FIG. 17
[FIG. 6] is a diagram in which a linear filter is extended in the −Y direction from a pixel having a (−Y) edge direction in the quantized direction.
Bit corresponding to −Y direction 1 is set in the second bit of the register of each pixel. FIG. 18 shows that the quantized direction 3 (−
It is the figure which extended the filter for 6 pixels from the pixel which has the edge direction of (X) to the -X direction. There is a 1 in the second bit of the register for each pixel corresponding to the quantized direction 3. FIG. 19 is a diagram in which a linear filter for 6 pixels is extended from a pixel having a quantized direction 4 (+ Y) edge direction. There is a 1 in the 3rd bit of the register for each pixel corresponding to the quantized direction 4. The target shape extraction is performed by ORing each bit of each register of each filter shown in FIGS. 16 to 19. 20
The register indicating the detected edge direction shown in FIG. 14 is obtained by ORing the value of the filter in the + X direction shown in FIG. 21 is similar to the diagram shown in FIG.
FIG. 9 is a diagram showing a register for each pixel obtained by performing an OR operation on the value of the filter in the −Y direction shown in FIG. 7. FIG. 22 shows the register shown in FIG. 21 obtained by ORing the values of the −X direction filter shown in FIG. In addition, FIG.
FIG. 20 is a diagram showing a register for each pixel obtained by ORing the value of the filter in the + Y direction shown in FIG. 19 to the value of the register shown in FIG. 2. With the above procedure, the process in S22 of the flowchart shown in FIG. 9 is completed.

Subsequently, in S23, the selection unit 36 of the extraction unit 30 extracts a pixel whose four bits are all 1. FIG. 24 is a diagram in which only pixels in which the value of each bit of the register for each pixel shown in FIG. 23 is 1 are extracted and binarized. In the drawing, a pixel in which all four bits of the register are 1 is represented by 1 and is indicated by a hatched portion. Pixels having other register values are set to 0 and are represented by blanks. In this way, the set of pixels indicated by the shaded portions has the extracted shape.

As described above, in the closed region extracting apparatus of the present invention, the edge detecting section 20 and the extracting section 30 are used to detect whether or not each pixel in the image is an edge. In the case of an edge, it is detected which of the first to fourth edge directions the edge has. Further, the register setting unit 34 sets a value in the register unit 32 provided for each pixel, so that the selection unit 36 selects the pixel whose register is set to ON as a pixel belonging to the closed region.

Next, the operation of the contour confirmation section 40 will be described. FIG. 25 is a diagram showing detected edges when the entire shape is extracted. FIG. 26 is a diagram showing edges when only a part of the shape is extracted.
As shown in FIGS. 25 and 26, if the edge direction is displayed so that a person can visually check it, it is possible to immediately judge whether the entire shape is extracted or only a part thereof is extracted. It can be carried out. On the other hand, in the case of a device that stores the presence of an edge or the edge direction of an existing edge as data, a certain algorithm is required to determine whether the detected edge surrounds the entire shape. . FIG. 27 is a diagram for explaining the operation of the contour confirmation unit 40 of this embodiment. When the contour confirmation unit traces the contour, the tracking is started with the edge direction and the tracking direction being the same. Furthermore, the tracking is done clockwise. Details thereof will be described with reference to FIGS. 28 to 34.
FIG. 27A shows the tracking when the edge direction of the pixel to be tracked is the quantized direction 1. That is, it is tracked whether or not there is an edge in the clockwise direction from the same direction as the edge direction. Make a chase. FIG. 27B shows a case where the edge direction of the pixel that starts tracking is the quantized direction 2. FIG. 27C shows a case where the edge direction of the pixel that starts tracking is the quantized direction 3. FIG. 27D shows a case where the edge direction of the pixel that starts tracking has the quantized direction 4. In this way, when the existence of an edge is found, tracking is started from that edge, and from the next edge detected during tracking, tracking is continued in the direction determined by the direction of that edge, and finally The tracking is completed when the first pixel (the pixel a at the starting point) where the tracking is started is returned. As described above, it is possible to confirm that the extracted shape is a closed region (closed region) by tracing continuously and returning to the starting point. Further, as shown in FIG. 26, when the existence of edges is not continuous, the contour confirmation unit determines the edge 10 of FIG.
1 or the edge 102 cannot detect the existence of the edge in the corresponding pixel to be tracked even if the edge is tracked, so it can be determined that the extracted shape is not a closed region.

A detailed description will be given with reference to FIGS. 28 to 34. The contour confirmation unit scans the image data having the edge direction, and starts the contour tracking from the edge found first. FIG. 28 is a diagram showing the tracking direction of the first pixel.
In the figure, it is assumed that the edge surrounded by the dotted line is the pixel having the edge found first. Edge direction is 2
Therefore, when the tracking is performed in the direction of FIG. 27B, the next edge is detected (lower left in the drawing). As shown in FIG. 29, the edge direction of the detected pixel is 1, so tracking is performed in the direction shown in FIG. When a pixel having a next edge is detected during tracking, the pixel is focused on for the next tracking. As shown in FIG. 30, the edge direction of the detected edge is 1, so tracking is performed in the direction of FIG. 27A, as in the case shown in FIG. Next, since the detected edge has an edge direction of 2 as shown in FIG. 31, tracking is continued in the direction shown in FIG. 27 (b). Similarly, in FIG. 32, tracking is performed according to the edge direction 2. As shown in FIG. 33, since the detected edge direction is 1, the tracking is continued in the direction shown in FIG. Thereafter, the tracking is continuously performed in the same manner, and finally, the pixel next to the pixel which has started is reached as shown in FIG. Since the detected edge direction is 2 in FIG. 34,
Tracking is performed in the direction shown in FIG. 7B, and the edge existing on the left side in the drawing is detected. The contour checking unit constantly monitors whether or not the pixel having the edge detected during the edge tracking is the pixel having the first detected edge. At the time of tracking shown in FIG. 34, it is determined that the pixel having the detected edge coincides with the first pixel and has returned to the starting point, and the tracking is ended. At this point, the contour checker
Make sure that the extracted shape you are tracking is a closed region. As described above, in this embodiment, the contour confirmation unit that traces and confirms that the adjacent pixel is an edge based on the edge detected by the edge detection unit and the edge direction has been described. In this example, the case where the tracking is started in the same direction as the edge direction is shown, but the case where the tracking is started in the opposite edge direction may be used. Even when tracing in the direction opposite to the edge direction, the edge can be searched by tracing in the clockwise direction.

Next, another operation example of the contour confirmation section will be described. FIG. 35 is a diagram showing edges when the same entire shape as that shown in FIG. 25 is extracted. As shown in the figure, when the entire closed shape is extracted, there is always an edge having an edge direction opposite to the edge direction of the edge. For example, as shown in FIG. 35, the edge 1a is opposed to the edge 1b. Similarly, 2b faces edge 2a and 3b faces edge 3a. On the other hand, as shown in FIG. 36, when only part of the shape is extracted, the edges 1b and 2b shown in FIG. 35 do not exist. The contour confirming unit may judge that the shape is not a closed shape by utilizing this fact and not being able to confirm the existence of an edge having an edge direction opposite to the edge direction of a certain edge.

Embodiment 2 FIG. Although the length r of the edge extraction filter is composed of 2 pixels in the above-described embodiment, the length r may be composed of 3 pixels as shown in FIG. Alternatively, an edge extraction filter that performs extraction with different characteristics can be obtained by changing the value set in the filter. As shown in FIG. 37 (a),
When the number of pixels is set to be larger than that of the edge extraction filter used in the above-described embodiment, it is possible to realize an edge extraction filter in which a difference in luminance is likely to occur even if the change in luminance of image data is faint. Further, as shown in FIG. 37B, when the number of pixels having a value of 0 is set to be large, that is, wide, it is possible to obtain an extraction filter that is easy to detect even if the difference in brightness of image data is small. That is, the edge can be extracted from the image data whose outline is blurred. In FIG. 37, an example of the X-direction edge extraction filter for extracting the horizontal edge is shown, but the same extraction filter may be used as the Y-direction edge extraction filter. Although not shown, the width may be composed of two or more pixels instead of one pixel and used as an edge extraction filter. However, in that case, there is a drawback that the memory used increases and the calculation time also takes a long time. However, even in that case, since calculation in only the X direction or the Y direction is sufficient, the amount of calculation is smaller and the calculation can be performed faster than in the case of using a conventional 3 × 3 matrix (planar) filter. There is.

[0035]

As described above, according to the present invention, the first
~ By detecting the fourth edge direction, the closed region can be extracted, and the closed region can be extracted at a higher speed than ever before.

Further, according to the present invention, the first pixel is provided for each pixel.
Since the fourth register is provided, the closed area can be extracted with a smaller storage capacity than the conventional one.

Further, according to the present invention, since the edge and the edge direction can be detected by using the X-direction edge extraction filter and the Y-direction edge extraction filter which are orthogonal to each other, the edge can be detected with a small number of calculations.

Further, according to the present invention, since the contour confirmation section is provided, it is possible to remove noise and clutter contained in the image.

The contour checking section can surely check that the area is a closed area by sequentially checking the adjacent pixels.

Further, by confirming the existence of opposing edges, it is possible to confirm that the contour portion is a closed region at a higher speed than in the case of tracing the pixel.

[Brief description of drawings]

FIG. 1 is a flow chart of a target finding algorithm of the present invention.

FIG. 2 is a diagram for explaining labeling in the embodiment of the present invention.

FIG. 3 is a diagram for explaining labeling in the embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration of a closed region extraction device of the present invention.

FIG. 5 is a diagram showing a closed region extracted in the embodiment of the present invention.

FIG. 6 is a diagram showing a closed region extracted by an embodiment of the present invention as an edge having an edge direction.

FIG. 7 is a diagram showing an X-direction edge extraction filter according to an embodiment of the present invention.

FIG. 8 is a diagram showing a Y-direction edge extraction filter according to an embodiment of the present invention.

FIG. 9 is a flowchart of edge detection to shape extraction according to the embodiment of this invention.

FIG. 10 is a diagram showing values in the edge direction according to the embodiment of this invention.

FIG. 11 is a diagram showing a register for each pixel according to the embodiment of the present invention.

FIG. 12 is an enlarged view of FIG. 6 of the embodiment of the present invention.

FIG. 13 is a diagram in which the edge direction of the embodiment of the present invention is represented by the numbers 1 to 4 indicating the quantized direction.

FIG. 14 is a diagram showing the edge direction of an embodiment of the present invention using a register for each pixel.

FIG. 15 is a conceptual diagram of region extraction within a range of length L surrounded in four directions according to the embodiment of the present invention.

FIG. 16 is a diagram of a linear filter in the + X direction according to the embodiment of the present invention.

FIG. 17 is a diagram of a linear filter in the −Y direction according to the embodiment of the present invention.

FIG. 18 is a diagram of a linear filter in the −X direction according to the embodiment of the present invention.

FIG. 19 is a diagram of a + Y direction linear filter according to an embodiment of the present invention.

FIG. 20 is a diagram showing a register obtained by ORing a linear filter in the + X direction according to the embodiment of the present invention.

FIG. 21 is a diagram showing a register obtained by ORing a linear filter in the −Y direction according to the embodiment of the present invention.

FIG. 22 is a diagram showing a register obtained by ORing a linear filter in the −X direction according to the embodiment of this invention.

FIG. 23 is a diagram showing a register obtained by ORing a linear filter in the + Y direction according to the embodiment of the present invention.

FIG. 24 is a diagram showing an extracted shape according to the embodiment of this invention.

FIG. 25 is a diagram showing a case where the entire shape of the embodiment of the present invention is extracted.

FIG. 26 is a diagram showing a case where only a part of the shape of the embodiment of the present invention is extracted.

FIG. 27 is a diagram for explaining the operation of the contour tracking unit according to the embodiment of the present invention.

FIG. 28 is a diagram showing tracking directions of pixels according to the embodiment of the present invention.

FIG. 29 is a diagram showing tracking directions of pixels according to the embodiment of the present invention.

FIG. 30 is a diagram showing tracking directions of pixels according to the embodiment of the present invention.

FIG. 31 is a diagram showing tracking directions of pixels according to the embodiment of the present invention.

FIG. 32 is a diagram showing tracking directions of pixels according to the embodiment of the present invention.

FIG. 33 is a diagram showing tracking directions of pixels according to the embodiment of the present invention.

FIG. 34 is a diagram showing tracking directions of pixels according to the embodiment of the present invention.

FIG. 35 is a diagram showing edges when the entire shape of the embodiment of the present invention is extracted.

FIG. 36 is a diagram showing edges when only part of the shape of the embodiment of the present invention is extracted.

FIG. 37 is a diagram showing another example of the edge extraction filter according to the embodiment of the present invention.

FIG. 38 is a diagram illustrating a conventional edge.

FIG. 39 is a diagram illustrating a conventional edge direction.

FIG. 40 is a diagram illustrating conventional edge extraction.

FIG. 41 is a diagram showing a conventional X-direction edge extraction filter.

FIG. 42 is a diagram showing a conventional Y-direction edge extraction filter.

FIG. 43 is a diagram showing a conventional edge detection calculation formula.

FIG. 44 is a diagram showing a conventional criterion for determining an edge direction.

FIG. 45 is a diagram showing a conventional value for quantizing an edge direction.

FIG. 46 is a diagram showing that the conventional edge directions intersect at the center of a circular shape.

FIG. 47 is a diagram showing an example in which a conventional spoke is extended.

FIG. 48 is a diagram showing a conventional spoke when the quantized edge direction is 5.

FIG. 49 is a diagram showing a pixel in which conventional spokes are overlapped.

FIG. 50 is a diagram showing a pixel in which conventional spokes are overlapped.

FIG. 51 is a diagram showing a correspondence between a conventional quantized edge direction and 8 bits.

FIG. 52 is a diagram showing a conventional register for each pixel.

FIG. 53 is a diagram showing an example of a conventional rectangular shape.

FIG. 54 is a diagram showing an example in which the conventional shape is a rectangle.

FIG. 55 is a diagram showing an example of a conventional rectangular shape.

FIG. 56 is a diagram showing an example of conventional parallel lines.

FIG. 57 is a diagram showing an example of conventional parallel lines.

FIG. 58 is a diagram showing an example of conventional parallel lines.

FIG. 59 is a diagram showing a conventional shape that cannot be extracted.

FIG. 60 is a diagram showing a conventional shape that cannot be extracted.

FIG. 61 is a diagram showing a conventional unextractable shape.

FIG. 62 is a diagram showing a conventional shape that cannot be extracted.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 closed area extraction device, 10 storage unit, 20 edge detection unit, 22 X direction edge extraction filter, 24 Y direction edge extraction filter, 30 extraction unit, 32 register unit,
34 register setting section, 36 selection section, 40 contour confirmation section.

Claims (6)

[Claims] 1. A closed region extraction device having the following elements: (a) a storage unit for storing an image composed of a plurality of pixels based on pixel data; and (b) for the pixel data stored in the storage unit. Performs predetermined calculation to detect whether or not each pixel is an edge that is a contour in the image, and when it is an edge, in the first to fourth edge directions,
An edge detection unit that detects which edge direction the edge has, (c) a pixel existing in the edge direction from each pixel detected as an edge by the edge detection unit, and the first to fourth edges An extraction unit that extracts pixels whose directions intersect each other as pixels in a closed region. 2. The extraction unit detects, as an edge by the edge detection unit, a register unit having first to fourth registers corresponding to first to fourth edge directions for each pixel of the image. The register setting unit that turns on the register corresponding to the edge direction of the selected pixel and turns on the register corresponding to the edge direction of a predetermined number of pixels existing in the edge direction from the pixel, and the register setting unit, The closed area extraction device according to claim 1, further comprising a selection unit that selects a pixel whose first to fourth registers are set to ON as a pixel belonging to the closed area. 3. The edge detection unit in the X direction,
An X-direction edge extraction filter that detects edges in the + X direction and the -X direction, and + Y in the Y direction orthogonal to the X direction.
3. A closed region extracting apparatus according to claim 1, further comprising a Y-direction edge extraction filter for detecting edges in the Y-direction and the Y-direction, and detecting the edge and the edge direction by calculating each filter and pixel data. . 4. The contour area checking device further confirms, based on the edge and edge direction detected by the edge detector, that the pixel detected by the edge detector as an edge is the contour of the closed area. The closed region extraction device according to claim 1, further comprising a confirmation unit. 5. The closed region extraction according to claim 4, wherein the contour confirmation unit traces and confirms that adjacent pixels are edges based on the edge detected by the edge detection unit and the edge direction. apparatus. 6. The closed region extracting apparatus according to claim 4, wherein the contour checking unit checks the presence of an edge having an edge direction opposite to the edge direction of the edge detected by the edge detecting unit.