JPS6016671B2 - Preprocessing method for shapes with closed loops - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention makes it possible to easily label internal regions of figures that feature closed loops, especially closed loops with depressions.

Concerning preprocessing methods for figures with closed loops. Figures represented by open loops (hereinafter referred to as closed figures) include convex figures with no depressions, such as triangular figures shown in Figure 1a, and arbitrary open curved figures, as shown in Figure 1B. There are concave shapes with depressions.

In shape recognition, etc., the internal Ai and external Ao of these shapes are
It may be necessary to perform a different labeling or data assignment, and to perform logical judgments or operations based on the labels. By the way, in the case of a convex figure as shown in Figure 1a, the internal area Ai can be labeled by a simple mask operation, but in the case of a concave figure as shown in Figure 1b, the concave part is also labeled. To prevent this, the mask logic becomes quite complex. The present invention aims to solve this problem and provide a method that can easily label internal regions of convex shapes as well as complex concave shapes by simple mask operations. .

In a figure preprocessing method that labels the internal area of a figure expressed in an open loop, the present invention creates a frame that surrounds the figure, and then performs a mask operation to create a frame that is sandwiched between the frame and the figure. The method is characterized in that the area is labeled with the same data as the data representing the data frame, and then the data is inverted to make the internal area of the figure the same data as the frame. This will be explained in detail below with reference to the drawings. In the present invention, as shown in FIGS. 2a and 2b, first, a frame F surrounding the closed figure is created in the outer white area Ao of the closed figure to be labeled.

Any suitable method may be used for this, but for example, a video signal is obtained by photoelectrically scanning the limbal line f3 with black "1" and white "0" outside and inside, and this is stored in the memory. ,
Next, the memory is retrieved, the cutting edge coordinate values of the closed figure in the row and column (x, Y axes) directions are determined, a margin is added to the coordinate values, the four sides of the frame F in the X and Y directions are determined, and this The memory storage value of the address corresponding to the four sides may be set to another color, for example, gray "2", instead of closed loop black "1" and white "0". Of course, the frame F does not necessarily have to be quadrilateral as shown in figure a, but may have any shape as long as it surrounds the closed figure f, as shown in figure b. Once the frame surrounding the closed figure has been created as shown in FIG. 2a, a mask operation is then performed within the frame.

The mask used at this time is 3 lines × as shown in Figure 3 a to d.
There are 4 types of masks M, ~M in 3 columns, and in these masks, the white part W is white or the same gray part as the frame, and the part B shown by mesh is the black "1" corresponding to the limbal line of the closed loop figure.
'', the hatched portion L is the gray "2" portion corresponding to frame F, and the portion marked with an "x" is a region that should not be labeled. Such a mask M・～
A mask operation within frame F is performed using the ground.

That is, mask scanning is performed in such a manner that the memory is sequentially shifted by 3 rows x 3 columns and 1 column at a time, and when the end of the row is reached, it returns to the beginning of the row and shifts by 1 row. If scanning is performed by horizontally scanning from the upper left of Figure 2a, for example around point P, to the right and vertically scanning downward, at first everything in the mask is white, or "0", but frame F has not been detected yet. Therefore, labeling (for example, changing the memory contents) is not performed. When frame F eventually reaches the center of the mask, the data "2" at the center of the mask and the memory stored value "2" corresponding to frame F match, and thus frame F is detected. After detecting frame F, a labeling operation is entered. That is, in the figure shown in Fig. 2a, in the next horizontal scan after frame detection, the upper edge of frame F will be in the first row of the mask, and in this part, the data will be in frame ``2'' or around it. Since it is white “0”, Figure 3a
Neither the mask deviation nor the mask deviation shown in d. to d applies, and in such a case, the data "0" in the memory at the position corresponding to the mask center position is changed to "2". In other words, label it. When horizontal scanning is repeated while performing the above operations, the white portion in frame F is gradually switched to the color of the frame. In other words, frame F begins to expand. Eventually, the mask M comes to the position shown in FIG. 2a. At this time, the outline of the closed figure f3 appears in a part of the mask, and the area o within the frame and outside the closed figure near the outline is labeled, so the figure area of the mask M part is shown in FIG. It is in the same state as the pattern of mask M shown in a. In this way, if the graphic data matches the mask, labeling of the x portion designated by the mask, which in this example corresponds to the third row and third column, is prohibited. Below, the same operation is performed for each scan, and the mask M
, ~M, and repeating the same operation several times by changing the presentation direction as necessary, the area Ao between the frame F and the open figure 3 is all labeled as shown in Fig. 4. Only the area within the closed figure f3 is left as white "0". To further explain this point, as is clear from FIGS. 3a to 3d, the black part of the mask M is in the fourth quadrant of the circular curve, the mask M2 is in the third quadrant, and the mask M3 is in the second quadrant of the circular curve. M4 can be seen as representing the same first quadrant, any part of the curve can be matched with the black part of any of these masks, and the central part of the mask is labeled The white areas within the frame that correspond to the parts that have been labeled are sequentially labeled as described above.
In the end, the fact that figure data equal to one of these masks was obtained means that closed figure 3 has been detected, and the mask × part corresponding to the inside of this open figure 3 must be labeled. Then, all areas outside the closed loop and within frame F can be labeled, leaving only the inside of the closed loop.

After labeling as shown in Figure 4, inversion processing is performed to make the labeled part white "0" and the unlabeled part gray "2".
As shown in the figure, the area N within the closed figure can be labeled.

FIG. 6 is a block diagram showing the procedure of the above labeling operation.

As shown in this figure, for the input shape, we first create a frame and detect the white area, perform a mask operation to label the area between the frame and the closed shape, and then create an inverted pattern. Generate the desired filler pattern. As described in detail above, according to the present invention, closed-loop figures, especially closed-loop figures with depressions, can be labeled with a simple mask operation, and this is extremely effective for figure recognition.

[Brief explanation of drawings]

Figure 1 a, b is an explanatory diagram showing an example of a closed figure, Figure 2 a, b
is an explanatory diagram showing how to create a frame surrounding a closed figure, the third
Figures a to d are diagrams showing examples of four types of masks used in the present invention,
Fig. 4 shows a state in which the area between the frame and the closed figure is labeled, Fig. 5 shows a fill-in pattern obtained by reversing the ribs, and Fig. 6 shows the figure of the present invention. FIG. 2 is a block diagram showing the outline of a preprocessing method. A figure represented by an open loop f, ~ f3' in the drawing, Ai
is its internal area, Ao is its external area, and F is its frame. Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6

Claims (1)

[Claims] 1 In a figure preprocessing method that labels the internal area of a figure expressed in a closed loop, a frame surrounding the figure is created, and then the area sandwiched between the frame and the figure is labeled using a mask operation. A figure with a closed loop characterized by labeling the figure to be the same data as the data representing the frame, and then performing a data inversion operation to make the internal area of the figure the same data as the frame. pretreatment method.