JPH077455B2 - Partial image area separation device - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for automatically separating / extracting a part of a figure or an image for recognition or the like.

2. Description of the Related Art In recent years, the practical use of OA, FA, etc. has advanced, and the need for recognizing figures and images has increased. ), Only the image of the partial area to be recognized, such as a character area, is separated, and the information of the partial image area such as the coordinates / pixel values of the constituent pixel groups of the partial area and the circumscribed rectangular frame information is extracted. The importance of equipment is also increasing.

Hereinafter, two examples of the conventional partial image area separation device will be described with reference to the drawings.

FIG. 5 is a block diagram showing the configuration of a first example of a conventional partial image area separation device. In FIG. 5, 1 is an edge extraction unit, 7 is an edge labeling unit, 8 is a closed region extraction unit, 5
Is an image storage unit, and 6 is an image area information storage unit.

The operation will be described below. First, the edge extraction unit extracts the contour portion of the region to be separated, that is, the edge portion, using the density difference of pixels, color information, or information such as a higher resolution image. Next, a label (pixel value) indicating the edge is given to each pixel forming the edge as shown in FIG. In FIG. 6, “◯” is a label that represents an edge constituent pixel, and “•” is a label that represents a pixel other than an edge in the separation target image area. However, a pixel other than an edge can have a label other than “◯”. Better, it doesn't have to have the same label. Further, the pixels outside the separation target image area may have the labels other than “◯”, and may be the same as the labels of the pixels in the area.

The image labeled in this way is stored in the image storage unit, and when it is desired to obtain the target partial image region and its information, the closed region extraction unit extracts the information of the edge and the pixels inside it, and Stored in the area information storage unit.

Further, FIG. 8 is a diagram showing a configuration of a second example of the conventional partial image area separating device. In FIG. 8, 9 is a region extraction unit, 10 is a region labeling unit, 11 is a labeling region extraction unit, 5 is an image storage unit, and 6 is an image region information storage unit.

The operation will be described below. First, the pixels belonging to the region to be separated are determined based on various kinds of information used in the edge extraction unit of the first example. Next, a label peculiar to the area is given to all the pixels forming the area. FIG. 9 shows a state where two regions to be separated separately are labeled when they are adjacent to each other. Pixels having the label "" belong to one area, and pixels having the label "" belong to the other area.

The image labeled in this way is stored in the image storage unit, and when it is desired to obtain the target partial image region and its information, the labeling region extraction unit extracts the information of the pixel having the label of the region, It is stored in the image area information storage unit.

However, the above-mentioned configuration has a problem that a large-scale and complicated image which has been handled in recent years cannot be processed at high speed and economically.

That is, in the first example of the conventional example, when two or more regions to be separated / extracted are adjacent to each other as shown in FIG. 7, either the contour tracking type or the raster scan type is used. It is difficult to correctly extract the edge of each region. In order to avoid this problem, if each edge of each partial image area to be separated / extracted is given a unique label, the number of labels to be added increases when the number of areas to be separated increases. Since the amount of information required for labeling also increases, it is necessary to increase the storage capacity required for each pixel and also increase the capacity of the entire image storage section. Alternatively, the seventh
After labeling / separating / extracting the area A in the figure, restore the edge label and then label / separate / separate the area B.
A method of extraction is also conceivable, but this is again the area A
When the image and its information are needed, the first edge extraction (unless the images and information of all partial image areas are stored in the image area information storage unit using a huge storage capacity) I have to start over.

In the case of the second example of the conventional example, the problem of the adjacent area and the problem of the reprocessing do not occur, but like the problem of the first example described above,
If the number of regions to be separated increases, a large-capacity image storage unit will be required, and not only edge pixels but all pixels in the region will be labeled, so that the region labeling unit will take a lot of time. There is a problem that it will end up.

In view of the above-mentioned problems, the present invention can correctly separate and extract each area even if a plurality of image areas to be separated are adjacent to each other. The present invention provides a partial image area separation device capable of performing high-speed processing without requiring large capacity in the image storage unit.

Means for Solving the Problems In order to solve the above problems, a partial image area separation device of the present invention is an edge extraction unit that detects an edge (contour) of a separation target image area, and an edge constituent pixel is an edge. From the pixel values of the adjacent edge labeling unit that gives the pixel value indicating, the edge outer edge labeling unit that gives the pixel value to the pixel adjacent to the outside of the edge constituent pixel, and the edge constituent pixel and the pixel adjacent to the outside. An adjacent closed area extraction unit that separates / extracts the separation target image area, an image storage unit that stores pixel value information of the image, and an image area information storage unit that stores information about the separated / extracted partial image areas, The adjacent edge labeling unit and the edge outer edge labeling unit are known at the time of assigning pixel values when a plurality of separation target image regions are adjacent to each other. It is characterized in that pixel values indicating edges and outer edges are provided according to the number of adjacent separation target image areas.

Effect The present invention, by the above-described configuration, while maintaining the same high-speed processing as conventional, performs labeling in consideration of the adjacent state between the plurality of separation target regions in the adjacent edge labeling section and the edge outer edge labeling section, By cutting out the internal region of the image labeled as described above by the adjacent closed region extraction unit, each separated pair region does not have a unique label, so that the image storage unit does not need a large capacity to perform multiple times. Separation and extraction of processable partial image areas are performed.

Embodiment A partial image area separation device according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows the configuration of an embodiment of the present invention. In FIG. 1, 1 is an edge extraction unit, 2 is an adjacent edge labeling unit, 3 is an edge outer edge labeling unit, 4 is an adjacent closed region extraction unit, 5 is an image storage unit, and 6 is an image region information storage unit.

The operation will be described below. The digital image is assumed to have a square lattice shape, but a hexagonal lattice shape or the like can be similarly operated with a slight modification.

The edge extraction unit is the same as in the first example of the related art, and extracts the contour of the separation target image area using density information, high resolution image information and the like. Next, the manner of labeling by the adjacent edge labeling portion and the edge outer edge labeling portion will be described with reference to FIG. First, a label "" representing the edge is given to the edge of the area A. Pixels outside the edge of region A (pixels that are adjacent to the outside of the edge by connecting 8)
Is given a label “◯” representing the outer edge. If the area A is alone (without other adjacent separation target areas)
If it exists, the form is kept as it is, and the separation / extraction may be performed as in the first example of the prior art. As shown in FIG. 2, when another separation target area B is adjacent to the area A, the following is performed. The adjacent edge labeling unit is not actually just mechanically labeling the edge pixels with "". The type of label to be given is determined by looking at the pixel value (label) before labeling edge pixels.
In FIG. 2, some of the edge pixels in the region B (region A
The portion adjacent to the edge of is labeled with a label "o" that represents the outer edge (when labeling area A). In this case, it means that the label is adjacent to the edge of another separation target area (here, the area A), so the label to be given is the adjacent edge label “”. At this time, in the edge outer edge labeling section, the label “” representing the outer edge of the adjacent edge is given to the pixels connected to the outside 8 of the area of the adjacent edge label. However, the edge outer edge labeling unit does not change the label even if the pixel is an edge outer edge, but is an edge pixel in another area. Therefore, the second
In the figure, the edge of the area A is not cut. If the edge is diagonal and there is an edge of another separation target area adjacent to it, the inside of the area of interest may be labeled as the outer edge when labeling the adjacent area. . The adjacent edge labeling unit detects such a pixel and changes the label to a pixel value that does not represent an edge or an edge outer edge. "X" in area B in Fig. 2
Pixels marked with correspond to this. Conversely, a pixel inside an already labeled area may become the outer edge of another adjacent area. When the edge outer edge labeling unit detects such a pixel, it does not label the outer edge. Pixels marked with “Δ” in the area A in FIG. 2 correspond to this. As described above, the adjacent edge labeling portion and the edge outer edge labeling portion may be labeled with reference to the labels of the surrounding pixels. In the case of a square grid digital image, up to four areas may be adjacent to one point as shown in FIG. 3, so that labels for edges and outer edges of triple and quadruple are prepared. When the original label of the edge pixel is the outer edge "" of the adjacent edge, a triple adjacent edge label is given, and the pixel of the outer edge is given a label representing the outer edge of the triple adjacent edge. The same applies to the case of quadruple. 3 if the digital image is hexagonal grid
Double labels are all right. In the actual labeling, the edge pixel and the pixel at the outer edge thereof can be labeled at the same time (in parallel), so that high speed performance is not impaired.

If the image labeled in this way is stored in the image storage unit, the adjacent closed region extraction unit can separate and extract the image and information any number of times. The adjacent closed region extraction unit traces the edge in the clockwise direction, and while advancing the left edge (that is, while the vertical component of the advancing direction is facing upward), from that pixel to the boundary pixel in the horizontal direction to the right. Extract a row of pixels. The boundary pixel is a pixel one pixel to the left when the edge pixel at the start point and the edge pixel corresponding to a different number of adjacent image areas are hit, and also hits a pixel representing any kind of outer edge. Sometimes it is the pixel one pixel to the left. If this process is performed while tracing around the edge of the region (until returning to the starting point), all pixels in the region can be separated / extracted. Edge tracking is performed if there is an edge pixel of the same type in eight pixels in the eight connected directions of the target edge pixel in a direction other than the edge pixel at the time point one pixel before (the pixel in the original direction that has advanced). Direction, or if there is an edge pixel corresponding to a different number of adjacent image areas, go to that direction,
Otherwise it is possible to go back in the direction of the previous pixel. However, when there is a protruding pixel (this is a pixel in the area C) as shown in FIG. 4, it is not necessary to proceed in the direction of the edge pixel in the area D. When labeling, the label "" is attached in advance (because it can be easily detected), and when the pixel having this label is reached, it is unconditionally returned to the original direction. Adjacent edge pixels and triple
Similar endpoint labels are prepared for quadruple adjacent edge pixels. It is considered that these end point labels and the edge pixels corresponding to the number of the adjacent image areas, the adjacent edge pixels, and the like are of the same kind in terms of edge tracking (otherwise, the end point pixels are left out).

As described above, by simply adding the above-mentioned 12 types (9 types in the case of a hexagonal grid image) to some pixels, even if there are adjacent separation target image regions, each region is correctly separated / extracted. be able to. Although the pixel value is rewritten when the label is given to each pixel here, a method such as adding each label as an offset to the pixel value when retaining the information of the original pixel value is used. You can take it.

EFFECTS OF THE INVENTION As described above, the present invention provides the adjacent edge labeling portion and the edge outer edge labeling portion, and the adjacent edge labeling portion and the edge outer edge labeling portion have a plurality of separation target image areas adjacent to each other. , The separation target image areas are adjacent to each other by giving pixel values indicating edges and outer edges according to the number of the adjacent separation target image areas that are known at the time of providing the pixel values. It is possible to perform multiple times of processing, does not require a large capacity in the image storage unit, and enables high-speed partial image area separation. It provides flexible pretreatment.

[Brief description of drawings]

FIG. 1 is a block diagram of a partial image area separating apparatus according to an embodiment of the present invention, FIG. 2 is a data diagram showing an example of an image labeling result by the above apparatus, and FIG. 3 is an adjacent area on a digital image. FIG. 4 is a data diagram for explaining the number, FIG. 4 is a data diagram showing an example of a labeling result of an image having a protruding portion by the above device, and FIG. 5 is a configuration diagram of a partial image area separating device of the first example of the conventional example, FIG. 6 is a data diagram showing an example of an image labeling result by the device, FIG. 7 is a data diagram showing a problem of labeling by the device, and FIG. 8 is a partial image area separating device of the second conventional example. FIG. 9 is a data diagram showing a result of labeling images by the same apparatus. 1 ... edge extraction part, 2 ... adjacent edge labeling part,
3 ... Edge outer edge labeling unit, 4 ... Adjacent closed region extraction unit, 5 ... Image storage unit, 6 ... Image region information storage unit, 7
...... Edge labeling section, 8 ...... Closed area extraction section, 9 ......
Region extraction unit, 10 ... Region labeling unit, 11 ... Labeling region extraction unit.

Claims (1)

[Claims] 1. An edge extraction unit for extracting an edge of an image to be separated, an adjacent edge labeling unit for giving a pixel value indicating an edge to an edge constituent pixel, and a pixel adjacent to the outside of the edge constituent pixel. A separation target image area is separated from the edge outer edge labeling section that gives a pixel value and the pixel values of the edge constituent pixels and the pixels adjacent to the outside.
An adjacent closed area extracting unit for extracting, an image storage unit for storing pixel value information of an image, and an image area information storage unit for storing information about the separated / extracted partial image area are provided, and the adjacent edge labeling unit and the above The edge outer edge labeling unit is configured to separate the plurality of separation target image areas when they are adjacent to each other.
A partial image area separation device, which applies pixel values indicating edges and outer edges according to the number of adjacent separation target image areas that are known at the time of applying pixel values.