JPH03214354A - smoothing device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application] The present invention relates to a signal smoothing method and apparatus for inputting a series of digital signals and smoothing these signals.

[Conventional technology]

Conventionally, in signal processing devices such as image processing,
When smoothing continuous digital data, the moving average method, which is an acyclic linear filter given by the following equation, has been used.

In equation {1}, x(1) is the input data, y(1
) is smoothed data, aj is a coefficient applied to each input data, and satisfies equation (2).

In this way, conventionally, even if the input data contained clearly significant signals other than normal noise, the same processing was performed uniformly.

[Problem to be solved by the invention]

FIG. 8 shows an example of processing using a conventional moving average.

In FIG. 8, the solid line is the input data string, and the broken line is the smoothed data. Note that FIG. 8 shows only the area where the moving average is actually effective (that is, excluding the n1 data at the leading edge and the n2 data at the trailing edge).

In the input data shown in FIG. 8, there are subtle changes in the signal that are clearly different from the gradual changes in signals such as A, B, and O. When smoothing using equation (1) is applied to such input data, it is affected by large input fluctuations as shown by the broken line in Figure 4, and the smoothed data tends to move in the direction of input fluctuations, as shown by the broken line in Figure 4. is pulled.

As described above, according to the conventional method, depending on the condition of the prominent data in the input data, the average value is There was a problem that the 'iI results fluctuated considerably.

The purpose of this invention is to realize good smoothing without being affected by the presence of peculiar data during input.

[Means to solve problems]

The signal smoothing method according to the present invention performs nonlinear variation when calculating a moving average.

Further, the signal smoothing device according to the present invention is equipped with a converter as a nonlinear converter that outputs a linear output within a set range and outputs O for inputs outside the range.

Also, a signal smoothing device according to another invention of the present invention! is non,
It is equipped with a transformer that outputs linearly within the set range and holds the boundary data of the set range in response to input outside the range.

Further, a signal smoothing device t according to another aspect of the present invention is provided with a converter having a characteristic of a hyperbolic function in the relationship between input and output as a nonlinear conversion section.

[Effect]

Since the present invention includes nonlinear elements, even if there is large protrusion data in the input, the original low frequency component can be extracted as smooth data without being affected by the data.

〔Example〕

Embodiments for implementing the signal smoothing method according to the present invention will be described below.

FIG. 1 is a block diagram of a first embodiment of the present invention. 1st
In the figure, (1) is an input memory that stores input data in the number necessary for smoothing. Also, (2) is a register that temporarily stores the earliest smoothed data, and (3)
is input memory! A subtractor that subtracts the contents of register (2) from each data stored in +1, (4) is a subtraction IE
The output of the device {3} is input, a nonlinear operation is performed, and the output is output. After weighting each output of the nonlinear converter (4) corresponding to each data in the input memo (January 1), This is a multiplier that adds the contents of register (2) and the output of the signal generator 151.

Next, the operation will be explained.

It is now assumed that the first smoothed data is obtained, and the input data required for smoothing are (N1 + N2 + 1) consecutive data from the 11N1th to the 10N2th data. At this time, the input memory (11 contains X(1-N1) ~
Data of X(i+N2) is stored. Here, X represents input data, and x(1) represents the first input data. Also in register (2)! ! #L analysis smoothed data 7(i-1) is stored.

First, data is read out sequentially from the input memory (1) and the I
For E +31, u (jl= x fjl - 7 (i-1)(
3) j=i-N1, iN1+I..., i+N2 are calculated. Next, in the nonlinear converter 14), u (j)
is input, performs nonlinear transformation, and outputs F(u(jl)).

However, it is assumed that F(u(jl)) has the following characteristics.

δ>IF(uij))]4) Here, δ is a constant set in advance. Furthermore, arithmetic unit (5)
Then, weighted addition of the outputs of the nonlinear converter 14) as shown in the following equation is performed.

However, Ilj is the weight applied to each data, and is . In subtraction 4f6) after hanging, add the contents of register +2) and the output of operation *515> as shown in the following formula,
Output as smoothed data y(1).

ylil=y (i - I) + ω (1) As described above, according to the smoothing device according to the present invention, F(x(j
Since l - y(i- + )) is kept below a certain value, even if there is a large variation in the input Does not affect data.

Next, a second practical example of the smoothing device according to the present invention is to convert a ROM of a conversion table as shown in FIG. 2 into a nonlinear converter 14).
It is provided inside the. If Figure 2 is expressed as a formula, it becomes , and when this is substituted into formula (7), it becomes . As shown in equation 19), when the reality of the input data exceeds the set range, the Heiwaka Delirion 1 according to the present invention excludes it from the smoothed data calculation, and instead uses the previous smoothed data. It adds weight.

FIG. 5 shows a processing example of a second practical example of the present invention.

Next, a third embodiment of the smoothing device according to the present invention is one in which a ROM containing a conversion table as shown in FIG. 3 is provided inside the nonlinear converter 14). Expressed as follows, and by substituting this into equation (F), we get j=1 N1.As shown in equation 111, the smoothing device according to this embodiment changes the smoothed data when the input data exceeds the set range. It is excluded from the calculation and calculated by adding or subtracting the width of the range from the previous smoothed data.

In this way, in this example of the 30th life fear by Genmei, the influence of the input data becomes thicker due to the addition of δ in the smoothed data calculation compared to the actual example by Genmei mentioned above, but the influence of the input data becomes thicker, but the signal decreases. Even if the frequency component changes discontinuously on the way, it can be tracked.

FIG. 6 shows a processing example of this third embodiment.

Further, a fourth implementation line of the smoothing device according to the present invention is provided with a ROM of a variable table as shown in FIG. 4 inside the nonlinear converter 14). , F Ijl= tan h ( u (j) ) 0z, and if the fluctuation of input data is small, the rate of data change is also small, and if the fluctuation of input data is large, the profit of data change becomes large, and the future As shown in the embodiment, there are no discontinuous points in data changes.

The smoothing device according to the fourth practical example of the present invention is the above-mentioned αυ
The equation shows that the smoothing is close to b! ! ! Shows j-ness.

FIG. 1 shows a processing example of the fourth life cycle example of the present invention.

As shown above, depending on the characteristics of input data and the eight variations of input fluctuations, by using any of the smoothing devices, smoothed data that is not affected by input fluctuations can be obtained.

〔Effect of the invention〕

As described above, according to the present invention, even if a large field of view signal different from the original low river wave component is superimposed during input, the influence on smooth data can be removed by the non-forming effect.

[Brief explanation of drawings]

Fig. 1 is a block diagram of one embodiment of this invention, Figs. 2, 3, and 4 are diagrams of other embodiments of this invention, Fig. 5, Fig. 6, and so on. FIG. 7 is a diagram showing an example of processing according to another embodiment of the present invention, and FIG. 8 is a diagram showing an example of processing according to a conventional method. In the figure, (1) is an input memo 'J, f21 is a register, (3) is a command q unit, } is a nonlinear transformer, }5) is an arithmetic unit, and (6) is a DoW unit. In addition, the same guardian symbols in the figures indicate the same or corresponding parts.

Claims (4)

[Claims] (1) Input the results of the previous smoothing on a series of digitized signals and a certain number of input data used for smoothing, and output within a certain range as shown in the following formula. A smoothing method that performs sequential smoothing using a nonlinear function that is suppressed by . ▲There are mathematical formulas, chemical formulas, tables, etc.▼ However, F(y(i-1), x(j)) is δ≧|F(y(i-1), x(j)) for the set value δ. | Let x(i) be the input data to be smoothed, y(i) be the smoothed data, ■_1, ■_2
is a variable that gives the number of data used for smoothing. (2) a register that temporarily stores the results of the previous smoothing on a series of digitized signals; an input memory that stores a certain number of input data used for smoothing; A subtracter that subtracts the contents of the above register from each data stored in the input memory and the output of the above subtracter are input, and if the input value is within a set range, it is output linearly, and when it exceeds the set range, it is 0. a nonlinear converter having characteristics to output; an arithmetic unit that performs weighted addition on each output of the nonlinear converter corresponding to each data in the input memory; and an addition that adds the output of the arithmetic unit and the output of the register. A smoothing device with a container. (3) a register that temporarily stores the results of the previous smoothing on a series of digitized signals; an input memory that stores a certain number of input data used for smoothing; A subtracter subtracts the contents of the above register from each data stored in the input memory, and the output value of the above subtracter is input, and if the input value is within a set range, it is output linearly, and if it exceeds the set range, it is output. A nonlinear converter having a characteristic of holding a boundary value, an arithmetic unit that performs weighted addition on each output of the nonlinear converter corresponding to each data in the input memory, and an output of the arithmetic unit and an output of the register. a smoothing device comprising an adder for adding; (4) a register that temporarily stores the results of the previous smoothing on a series of digitized signals; an input memory that stores a certain number of input data used for smoothing; a subtracter that subtracts the contents of the register from each data stored in the input memory; a nonlinear converter that inputs the output of the subtracter and whose input and output have characteristics of a hyperbolic function; A smoothing device comprising: an arithmetic unit that performs weighted addition on each output of a nonlinear converter corresponding to each data; and an adder that adds the output of the arithmetic unit and the output of a register.