JPS58134748A - Settling system for letter data compression block - Google Patents

Abstract

PURPOSE:To obtain the titled rettling system capable of raising the compression rate of letter data and also of reducing the number of sampling points to be set in entire letter outline by using a system in which the second dividing point is newly settled in the point where the setting density of sampling points exceeds a given threshold value and the block is divided into two sections. CONSTITUTION:Letter outline typically developed on X- and Y-coordinates are grasped as the aggregation of blocks B formed by singe-valued function with X as a variable. Letter outlines are approximated by curves passing through plural sampling points P1-1-Pi+3 set on each block. The case where the sum rhoi[i= 2-(n-3)]of the distance between the points Pi and Pi+1 and the distance between the points Pi+1 and Pi+2 becomes smaller than a threshold value rho' is extracted as a sampling point-densed section. Then, the farthest point Q in relation to the traight lines Pi-1 and Pi+3 is obtained, and when thetaq is smaller than a threshold value theta', the point Q is used as a new dividing point D and one block is diviced into two sections. Thus, the number of sampling points near the specific point is reduced and the compression rate of data is raised.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides data compression techniques for sampling points and blocks when performing data compression by approximating the contours of characters, images, etc. (hereinafter referred to as characters) with a set of function curves or straight lines. This is related to the setting method.

First, setting of sample points will be explained.

In a character data compression method that divides the outline of a character into sample points and approximates each divided outline segment with a function curve or a straight line, it is possible to It is important to set a small number of sampling points quickly.

Conventionally, the simplest method for setting such sample points is to obtain sample points by increasing a variable by a unit amount for a character outline developed on the X and Y coordinates.

However, since this method relies only on the variable axis to set the sampling points, a certain likelihood is not guaranteed between the approximated curve (or straight line) and the actual character contour, and therefore the optimal It was difficult to set sample points for

In addition, after calculating the likelihood of the difference between the approximated curve (or straight line) and the actual character outline with respect to candidates and sample points, test according to the likelihood. There is also a method that moves the candidate sample point and sets the candidate sample point as the set sample point when a predetermined likelihood is obtained, but this method requires a trial and error process to set one sample point. Therefore, the processing was redundant.

The sampling point setting method according to the present invention is a method for solving these same noodle points, and is a rational method that sequentially sets one sampling point for each likelihood comparison. It provides:

First, preprocessing of character contours, which is a prerequisite for setting sample points, will be explained.

Figure 1 shows the character outline in multiple blocks (B1.82,
B3...), and is a schematic explanatory diagram of a state where the character outline is understood as a set of each block B(7), and in the figure "
The "○" mark and the "•" mark are the end points of each block.

That is, as a preprocessing for setting sample points, dividing points are set as appropriate on the character outline developed in X and Y coordinates, and X is set as a variable.
Preliminary 1. S
Look for blocks.

For this block 11, the X coordinate value of the series of contours is
By setting an arbitrary interval that monotonically increases or decreases as a block, □ can be generally processed.

Furthermore, as shown in Fig. 1, a more appropriate block setting method is to find the position where the X coordinate value on a series of contours has an extreme value, and then set the position corresponding to the extreme value to the initial dividing point D. (marked with a "○" in the figure), and furthermore, as will be described in detail in connection with the block setting method described later, on each contour block divided by the dividing point, a second
By setting the division point D' (in the figure, the mark ')', the section divided at each division point (D or Di) is set as a block B, respectively.

The physical meaning of each block B set in this way is that each block B is
has a meaning i corresponding to a set of points that turn on or turn off a bright spot scanning in the Y direction.
□ The sampling point setting method according to the present invention is a sampling point setting method for further dividing the contour corresponding to the block into a large number of contour segments for any one block B set by the above-mentioned preprocessing. It is related to.

2A to 2C are schematic explanatory diagrams based on the method of the present invention, and FIG. 3 is a configuration diagram showing an embodiment of the sample point setting method of the present invention.

In FIG. 3, reference numeral 31 denotes a contour information storage unit that stores the coordinate information of the contour corresponding to any one block B shown by the solid curve in FIG. 2, and outputs this as necessary.
2m, a sample point coordinate storage unit 33 which sequentially stores the coordinate information of the sample points set by the method of the present invention and outputs it as necessary; Regarding the existing sample points stored in the sample point coordinate storage unit 32, a distance calculation unit calculates the distance between adjacent sample points, 34 is an approximate curve calculation unit to be described later, 35 is,
- an area calculating unit that calculates an area S surrounded by a contour segment cut out by adjacent sample points and an approximate curve connecting each sample point already set on the contour; 3';
6 is an evaluation amount calculation unit that calculates an evaluation amount ξ of S/L for each segment based on the distance calculated by the distance calculation unit 33 and the area S calculated by the area calculation unit 35; 37 is the maximum The evaluation amount detection section 38 includes a comparison section 40 that compares the maximum value ξlaX of the evaluation amount calculated by the evaluation amount calculation section 36 with a predetermined tolerance value ξ' supplied from the tolerance value setting section 39. , the comparison result in the comparison unit 38 is ξlaX >ξ
', the newly installed sample point calculation unit calculates a new sample point based on a predetermined relationship, for example, the deviation calculation unit 41,
It is composed of a maximum deviation detection section 42.

The operation of the configuration shown in FIG. 3 will be described in conjunction with FIG.

First, both end points P1 and P2 of block B in FIG.
(These end points correspond to the dividing point or D' in FIG. 1 obtained by the above preprocessing) coordinate information is combined with the coordinate information of the existing sample point and 1. and store it in the sample point coordinate storage unit 32,
Set initial conditions for operation.

When the setting of the above initial conditions is completed, the distance L between the two main points (second equal point) is calculated based on the coordinate information of the existing sample points P1 and P2 read from the sample point coordinate storage unit 32. The calculation unit 33 outputs this to the evaluation amount calculation unit 36.

On the other hand, in parallel with this, the approximate curve calculation unit 34 also calculates the existing sample points PI, P2 read from the sample point coordinate storage unit 32.
Based on the coordinate information of , an approximate curve connecting both of the two markers is calculated, and this is output to the area calculation section 35 .

However, in the present case where the operation is started from the initial conditions, there are two existing sample points, so the approximate curve becomes a straight line C1 as shown in FIG. 2A.

Next, the second! ! Area shown in IA 3 a + 3 ##
=3 is calculated by the area calculation unit 35, and this is calculated by the evaluation amount calculation unit 3.
Output to 6.

The evaluation amount calculation unit 361 calculates an evaluation amount ξ of S/[1 based on the distance calculated by the distance calculation unit 33 and the area S calculated by the area and product calculation unit 35.

In this case, since the number of evaluation amounts to be calculated is one, this value is passed through the maximum evaluation amount detection unit 37, and the maximum evaluation amount ξ−
The signal is supplied as ax to the comparator 3β.

The comparison unit 38 is given a predetermined tolerance value ξ− as the other input from the tolerance value setting unit 39, and ξ−a
When X >ξ', the newly installed sample point calculation unit 40 is activated.

The deviation calculation section 41 configuring the newly installed sample point calculation section 40 calculates the curve obtained by the approximate curve calculation section 34 (especially the straight line C1 in this case) and the contour coordinate information stored in the contour information storage section 31. Based on the deviation Δ shown at the second tie point, point 81
The calculation is performed over the entire interval from point P2 to point P2, and the maximum deviation detection unit 42 determines the positional information of point P3 on block B where the value is the maximum. Then, the point P3 is stored and set in the sample/point coordinate storage section 32 as a newly established sample point.

When the new sample point P3 is thus set on the contour of block B, the configuration of FIG. 3 again performs the operations described above. This operation process will be explained below with reference to FIG. 2B. At this time, the newly installed sample point P3 is updated to the existing sample point.

That is, in this case, the distance calculating section 33 first calculates the distances L1 and L2 shown in FIG. 2B. Further, in parallel with this, an approximate curve C2 (indicated by a dotted line in FIG. 2) passing through the existing sample points P 1 sP2 and P3 on block B is determined by the approximate curve calculation unit 34.

The area calculation unit 35 separately calculates areas 81 and 82 (indicated by diagonal lines in FIG. 2) surrounded by each contour segment cut out by the sample points and the approximate curve C2.

Next, the corresponding distances L+, L2 and areas Sl,
82, the evaluation amounts ξ1 and ξ2 where S/L=ξ are calculated by the evaluation amount calculation unit 36.

Now, in the case of Figure 2 B, ξ1 x ξ2, so
The value of ξ2 is supplied from the maximum evaluation amount detection section 37 to the comparison section 38 as the maximum evaluation amount ξ-ax.

Then, when ξg+ax (=ξ2)>ξ′, the comparison unit 3
In response to the command from No. 8, the cutting mark Mokugyo nose protrusion 40 is activated again, and in the same manner as described above for the case shown in Fig. 2A.
Point P4, which is ΔIaX, is set in the sample point coordinate storage section 32 as a newly established sample point.

The processing state after the newly established sample point P4 is obtained in this way is as shown in Figure 2C, but the operation of the configuration in Figure 3 at that time itself is a repetition of the operation that has already been described. Therefore, the explanation will be omitted below.

Then, in the process of sequentially finding new sample points PL, when the comparing section 38 detects ξlaX≦ξ', the operation of the configuration shown in FIG. 3 regarding any one block B is completed.

Of course, at this time, the sample points P1 to PL stored in the sample point coordinate storage section 32 are set as desired sample points regarding the outline of the block B.

Note that the distance value in the distance calculation unit 33 is L−J
It can be obtained very easily by calculating rτx)T+(Ay)T. Further, the area calculation unit 35 can calculate the desired area S relatively easily by integrating the difference in the y direction between the two curves in the X direction. Therefore, as mentioned above (3
The calculation of the value of the evaluation quantity ξ given as S/L is also...
It's relatively easy. Since this evaluation quantity ξ is a quantity that has the dimension of length by dividing the area S by the length L, the two curves (the contour and the approximate curve) forming the area S have a unique unevenness. Unless otherwise specified, these two curves serve as an indicator of the dissociation relationship between the two curves.

FIG. 4 is a flow diagram showing the operation process of the configuration shown in FIG. 3.

Next, in connection with the case where the character contour has the above-mentioned unique unevenness, a block setting method will be explained to make the sample point setting method using the configuration in FIG. 3 or the process in FIG. 4 even more effective. .

That is, when implementing the sampling point setting method according to the configuration in Figure 3 or the process in Figure 4, as shown schematically in Figure 5, there is a tendency for the sampling points P to be concentrated in parts where the radius of curvature of the character contour is small. This may prevent improvement of the data compression rate depending on the number of sample points.

Therefore, the improved block setting method described below focuses on the above-mentioned -1.
Sample=,,In the part where the first set density exceeds a predetermined threshold,
A second dividing point (corresponding to dividing point D' in Fig. 1) is newly set to divide the block into two.
This is intended to contribute to reducing the number of sample points set in the entire character contour by carrying out the sample point setting according to the configuration shown in FIG. 3 again for each of these 10 divided blocks.

An embodiment of this improved block setting method will be described below with reference to FIG.

In FIG. 6, Pl and Pn are end points of any one block, and PL-1' to PL+3 are sample points set by applying the sample point setting method to block B.

In implementing the improved block setting method, for example, first, the density of sample points is calculated for all cases of .about.(n-3) and compared with the threshold value ρ'.

Then, the case where ρ is ρ' is extracted as a sample point dense interval.

Next, this sample point dense area, that is, block B (F)
(Kfll [PL-1; Pt-1] Koiku 1.
,5-0,\P,. Find the point Q that is farthest from . ＼Angle formed (P L-1, Q, Pt or 3
)−θq is compared with the threshold θ−, and θq
〈ρ', the point Q is the new dividing point D', and 1
Divide one block into two.

The improved block setting method has been described above. Then, by applying the sample point setting method shown in Figure 3 again to each of these two divided blocks, the sample point setting method shown in Fig. This reduces the number of points, thereby contributing to the data compression effect.An example of processing actual character contours based on the sample point setting method and block setting method described above in detail is as follows. It is shown in Figure 7 for reference.

In FIG. 7, the "○" mark is the initial dividing point explained in relation to FIG. 1, and the "・" mark is the second dividing point determined by the improved block setting method, '' marks are sample points obtained after setting the second division points, and in the case shown, the numbers of these points are 4019 and 99.

As described in detail above, the present invention grasps the character outline schematically developed on the X and Y coordinates, with X as a variable, and further,
Regarding the character data reduction method in which the character contour is approximated by a curve passing through multiple sample points set on each block, this method focuses on the property that sample points are concentrated in the part of the character contour with a small radius of curvature. It uses
More specifically, the density of multiple sample points set on any one block is calculated for each keyaki main point, and then when a sample point dense area where the sample point density exceeds a predetermined threshold is detected, the corresponding Set a new division point in the dense area, and use the division point to
The feature is that one block is divided into two blocks. By doing so, the present invention provides a block setting method that makes the above character data compression method even more effective by setting sample points on the entire character contour.

[Brief explanation of the drawing]

Fig. 1 is a schematic explanatory diagram of a state in which a character outline is understood as a set of multiple blocks, and Fig. 2 is a schematic explanatory diagram of the process of setting a sample point on any one block B based on the method of the present invention. 3 is a block diagram showing an embodiment of the sampling point setting method of the present invention, FIG. Figure 6 is a diagram schematically illustrating an example of implementing the sample point setting method based on the process, Figure 6 is a diagram schematically illustrating an example of the block setting method improved by the present invention, and Figure 7 is a diagram schematically illustrating an example of the block setting method improved by the present invention. FIG. 7 is a reference diagram showing an example of processing an actual character outline using both the sample point setting method and the block setting method according to the invention. B...Block C...Approximate curve D1D'
, Q... Division point P... Sample point 31... Contour information storage section, , 32... Sample point coordinate storage section: 1 33... Distance calculation section ' 34... Approximate curved line Drawing part 35...Area calculation part 36...Evaluation amount calculation part 37...Maximum evaluation amount detection part 38...Comparison part 39.
...Tolerance value setting section 40...New sample point calculation section 41
... Deflection calculating section 42... Maximum deviation position detecting section "
○”・・・Initial dividing point “・”・・・Second dividing point “Δ”・・・Sample point Patent applicant Shaken Co., Ltd. Figure 3 Figure 5 1゛ Figure 6

Claims (2)

[Claims] (1) The character outline developed schematically on the X and Y coordinates is
A unique character data compression method in which the character contour is approximated by a curve passing through a plurality of sample points set on each block. The density of the sample points set on the block is determined for each sample point, and when a part of the sample point band where the density exceeds a predetermined threshold is detected, a new division point is set in the dense area, and the division is performed. A block setting method for compressing character data, in which the arbitrary one block is divided into two blocks according to points. (2) The sample point set on any one block is P
L (where i-1 to n, and Ps and Pn are the start and end points of the block), P L , P L
12 +P L+l s P L◆25ρL (However,
i-2~(n-3)) is smaller than a predetermined threshold ρ1, the straight line P L-
1. Find the farthest point Q from PL and 3, and find the angle (
PL-1, Q N , PL+3 )-θq is a predetermined value θ
2. The block setting method for character data compression according to claim 1, wherein when the point Q is smaller than 1, the point Q is set as a newly set desired dividing point.