KR100628878B1 - Noise filtering device and method - Google Patents

Abstract

The present invention discloses a noise filtering apparatus and method. In the noise filter method, when a target signal including a noise signal is received, a signal input step of determining a sampling group signal period for identifying a period of the target signal and a period of the noise signal and dividing a sampling signal group and the target signal A sampling signal group generating step of generating a plurality of sampling signal groups by classifying the signal according to the sampling group signal period, obtaining a maximum value of each of the plurality of sampling signal groups, and obtaining the maximum value. Calculate the average value of the maximum values (minimum values) having a relatively large value (minimum value) among the maximum values (minimum values), and then calculate the average value of the maximum values (minimum values) that are not deleted. And clipping the target signal according to the average value. Therefore, the harmonic components of the noise signal which were not separated in the frequency domain can be easily separated. This increases the reliability of the filtering operation.

Description

Noise filtering apparatus and method

1 is a diagram illustrating an example of a target signal including a noise signal;

2 is a block diagram illustrating an internal noise filtering apparatus of the present invention.

3 is a view for explaining the concept of noise occurrence probability of FIG.

4 is a flowchart illustrating a noise filtering method of the present invention.

5 is a flowchart for explaining an actual application example of the noise filtering method of the present invention.

The present invention relates to a noise filtering device, and more particularly, to a noise filtering device and a method for removing harmonic components of a noise signal from a target signal which are generally known to be impossible to resolve.

In general, when generating and transmitting a signal, that is, a target signal, various noise signals may be mixed with the original target signal by a peripheral device or a communication environment.

FIG. 1 is a diagram illustrating an example of an input signal including a noise signal NS and a target signal TS. The target signal TS of FIG. 1 has a period of PT, and the noise signal NS is a target signal. With the period of PN multiplied by TS, the period PN of the noise signal has an integer multiple N of the period PT of the target signal.

In this case, if the noise signal NS is not a signal having a single frequency, the harmonic components of the noise signal NS overlap with the target signal TS in the frequency domain.

However, the noise filter according to the related art is mostly a frequency domain filter, and the frequency domain filter cannot detect harmonic components of the noise signal NS using the same frequency band as the target signal TS. It was not possible to remove the harmonic components of the noise signal NS from the TS).

That is, the noise filter according to the prior art has a limit in separating the noise signal from the target signal.

For example, if the target signal is cos (t), the noise signal is a0cos (t / 5) + a1cos (t) + a2cos (5t) and a0, a1, a2 are coefficients in the frequency domain, then the noise signal ( It can be seen that the harmonic component a1cos (t) of NS) uses the same frequency domain as cos (t) of the target signal.

However, conventional frequency domain filters such as a Fast Fourier Transform (FFT) filter and a Band Pass Filter (BPF) filter cannot detect harmonic components of the noise signal NS that use the same frequency band of the target signal TS. The harmonic component of the noise signal could not be removed from the target signal TS.

Accordingly, the present invention proposes a new filter that can separate harmonic components of the noise signal NS from the target signal TS.

An object of the present invention is directed to a noise filtering apparatus and method for removing the harmonic components of a noise signal from a target signal to increase the reliability of the filtering operation.

The noise filtering apparatus of the present invention for achieving the above object is a signal input unit for analyzing the target signal, to determine the period of the target signal, the period of the noise signal and the sampling group signal period, and the target signal to the sampling group signal A sampling signal group generator for classifying according to a period to generate a plurality of sampling signal groups, a representative sampling signal acquisition unit for obtaining a maximum value of each of the plurality of sampling signal groups, and the representative sampling according to a noise signal generation probability An average value calculator which calculates an average value of the undeleted maximum values after deleting the maximum values having a relatively large value (small value) among the maximum values obtained through the signal obtaining unit, and the target signal according to the average value. And a signal filter for clipping.

According to an aspect of the present invention, a noise filter method includes: a signal input step of determining a period of the target signal, a period detection of the noise signal, and a sampling group signal period when receiving a target signal including a noise signal; Generating a plurality of sampling signal groups by classifying the target signals according to the number of repetitions of the target signal period, and obtaining a maximum value for obtaining a maximum value (minimum value) of each of the plurality of sampling signal groups. And deleting the maximum values (minimum values) having a relatively large value (small value) among the obtained maximum values (minimum values) according to the noise signal generation probability, and then averaging the undeleted maximum values (minimum values). Calculating a mean value and clipping the target signal according to the mean value. The.

Hereinafter, the noise filtering apparatus and method of the present invention will be described with reference to the accompanying drawings.

2 is a block diagram illustrating the internal structure of the noise filtering apparatus of the present invention.

Referring to the drawings, the noise filtering apparatus includes a signal input unit 1, a sampling signal group generation unit 2, a representative sampling signal acquisition unit 3, an average value calculation unit 4, and a signal filter unit 5.

Looking at the function of each component as follows.

The signal input unit 1 receives an input signal including the noise signal NS and the target signal TS, and grasps signal characteristics and sampling characteristics of the target signal TS and the noise signal NS.

More specifically, the period PT of the target signal, the period PN of the noise signal, and the number of repetitions R of the target signal period are determined within the sampling group signal period, and the target is determined within the period PT of the target signal. The number of times (NP) to sample the signal, that is, the sampling period, is determined. In this case, the sampling period may be arbitrarily set by the user.

The sampling signal group generation unit 2 generates a plurality of sampling signals by sampling the input signal according to the sampling characteristics set through the signal input unit 1, and target signals identified through the signal input unit 1. The plurality of sampling signal groups SSG [1] to SSG [R] are generated by classifying according to the number of repetitions of the period.

The representative sampling signal acquisition unit 3 acquires a sampling signal having a maximum value (minimum value) of each of the plurality of sampling signal groups SSG [1] to SSG [R], and converts the obtained sampling signal into a corresponding sampling signal group. Set as a representative sampling signal for (SSG [1]-SSG [R]).

The average value calculator 4 removes representative sampling signals having a relatively large signal value (small signal value) among the plurality of representative sampling signals acquired through the representative sampling signal acquisition unit 3 according to a noise occurrence probability, The average value M of representative sampling signals that are not removed is calculated. The noise occurrence probability at this time will be described in detail with reference to FIG. 3.

The signal filter unit 5 acquires a signal region having a signal value (small signal value) larger than the average value M calculated by the average value calculator 4, and replaces the obtained signal region with an average value. Then print. That is, the input signal is clipped according to the average value M calculated by the average value calculator 4 and then output.

In this case, it can be seen that the signal region obtained through the signal filter 5 is a region including harmonic components of the noise signal in the target signal.

FIG. 3 is a diagram for describing a concept of noise occurrence probability of FIG. 2.

In this case, the target signal TS of FIG. 3 is the same signal as the target signal TS of FIG. 1.

(a) shows the noise occurrence probability when the start point and the end point of the period PT and PN of the target signal and the noise signal are set as the maximum occurrence points, and (b) the period PT of the target signal and the noise signal. A diagram for describing the noise occurrence probability when the start point and the end point of PN) are set as minimum value start points.

When generating a plurality of sampling signal groups from the target signal NS of FIG. 3, the probability that the plurality of sampling signal groups include the noise signal NS as the representative sampling signal is as follows.

First, the noise occurrence probability when the start point and the end point of the period PT and PN of the target signal and the noise signal are set to the maximum value generation points will be described with reference to (a).

In the case of (a), N target signals TS are included in the period PN of the noise signal.

In the period PN of the noise signal, N sampling signal groups SSG [1] to SSG [N] are included, and the first to Nth sampling signal groups SSG [1] to SSG [N]. ) Sets the noise signal having the largest signal value within the corresponding sampling signal group as the representative sampling signals SSMAX1 to SSMAXN, and sets the second to N-first sampling signal groups SSG [2] to SSG [N. Each sets the target signal having the largest signal value in the corresponding sampling signal group as the representative sampling signals SSMAX2 to SSMAX (N-1).

Therefore, in the case of (A), two representative sampling signals SSG [1] to SSG [N] of the N representative sampling signals SSMAX1 to SSMAXN are generated from the noise signal NS.

Since the probability is equally applied to all target signals, the probability that the plurality of sampling signal groups include the noise signal as the representative sampling signal may be generalized to 2 / N, that is, 2 / (PN / PT).

On the other hand, the probability of noise occurrence when the start point and the end point of the period PT and PN of the target signal and the noise signal are set as the minimum occurrence points will be described with reference to (b) as follows.

In the case of (b), as in (a), N target signals are included in the period PN of the noise signal. That is, N sampling signal groups are included in the period PN of the noise signal.

However, only the N th sampling signal group SSG [N] obtains a noise signal having the largest signal value in the sampling signal group as the representative sampling signal SSMAXN, and the first to N th sampling signal groups ( SSMAX1 to SSMAX (N-1)) obtain a target signal having the largest signal value in the corresponding sampling signal group as the sampling signal.

Therefore, in the case of (b), one representative sampling signal SSG [N] of the N representative sampling signals SSMAX1 to SSMAXN is generated from the noise signal. In addition, since the probability is equally applied to all target signals, the probability that the plurality of sampling signal groups include the noise signal as the representative sampling signal may be generalized to 1 / N, that is, 1 / (PN / PT).

The noise occurrence probability obtained when the noise signal and the target period start point and the end point are moved right or left with respect to the time axis is necessarily included in 1 / (PN / PT) to 2 / (PN / PT).

Therefore, the maximum probability that the plurality of sampling signal groups include the noise signal as the representative sampling signal is 2 / (PN / PT), and the minimum probability is 1 / (PN / PT).

Accordingly, in the present invention, some representative sampling signals of the plurality of representative sampling signals of the plurality of sampling signal groups are deleted according to the maximum noise occurrence probability.

4 is a flowchart illustrating a noise filtering method of the present invention.

When an input signal including the target signal TS and the noise signal NS is transmitted, the input signal is received and the received signal is analyzed to determine signal characteristics and sampling characteristics of the target signal TS and the noise signal NS. do. More specifically, the number of samplings for sampling the target signal TS within the period PT of the target signal while determining the sampling group signal period and determining the repetition number R of the target signal period within the sampling group signal period. (NP) is determined (step S1).

R x NP sampling signals are generated by sampling the input signal according to the sampling frequency (NP) determined in step S1, and the R x NP sampling signals are equally divided by the repetition number (R) of the target signal period, and R sampling is performed. Acquire signal groups (step S2). In this case, each sampling signal group includes NP sampling signals.

A sampling signal having a maximum value (minimum value) is obtained for each sampling signal group, and each of the obtained sampling signals is regarded as a representative sampling signal of the corresponding sampling signal group (step S3).

The R representative sampling signals are sorted in ascending order (descending order), and the sorted R representative sampling signals are deleted in descending order (ascending order) according to the noise occurrence probability (step S4). Accordingly, some representative sampling signals having a relatively large signal value (small signal value) among the plurality of representative sampling signals are deleted.

The average value of the remaining representative sampling signals not deleted is calculated (step S5), the input signal is clipped according to the average value calculated through step S5, and the clipped signal is output (step S6).

Through the above method, the present invention detects harmonic components of a noise signal having the same frequency as the target signal, and removes harmonic components of the noise signal from the target signal.

Hereinafter, the practical application of the present invention will be described with reference to FIGS. 5A to 5F.

FIG. 5A illustrates an input signal including a target signal TS and a noise signal NS, FIG. 5B illustrates a plurality of sampling signal groups, FIG. 5C illustrates representative sampling signals of each of the plurality of sampling signal groups, and FIG. The remaining representative sampling signals from which the representative sampling signals due to noise are removed, FIG. 5E illustrates an average value of the remaining representative sampling signals, and FIG. 5F illustrates an output signal from which harmonic components of the input signal are removed using the average value of the remaining representative sampling signals.

First, when the input signal of FIG. 5A is received, the sampling group signal period of the input signal is determined, the number of repetitions R of the target signal period is determined within the sampling group signal period, and simultaneously within the period PT of the target signal. The number NP of sampling the target signal is determined (step S1).

The input signal is sampled by the number NP set in step S1 to generate the first to RNP sampling signals SS [1] to SS [RNP], where R is the number of repetitions of the target signal period and NP is the number of samplings. The generated sampling signals SS [1] to SS [RNP] are equally divided by the number of repetitions R of the target signal period within the sampling group signal period, and the first to Rth sampling signal groups of FIG. (SSG [1] to SSG [R]) are generated (step S2).

Accordingly, the first to R th sampling signal groups SSG [1] to SSG [R] are generated, and the first sampling signal group SSG [1] is the first to NP sampling signals SS [1]. SS [NP]), and the second sampling signal group SSG [2] includes first NP + 1 to second NP sampling signals SS [NP + 1] to SS [2NP]. In the same manner, the third to R th sampling signal groups SSG [3] to SSG [R] include a plurality of sampling signals.

The first sampling signal SS [1] (see FIG. 5B) having the largest value in the first sampling signal group SSG [1] is the first representative sampling signal SSMAX1, and the second sampling signal group SSG [ 2)) is the second NP + 1 sampling signal SS [NP + 1] having the largest signal value as the second representative sampling signal SSMAX2, and the third sampling signal group SSG [3] is the largest signal. A second NP + 1 sampling signal SS [2NP + 1] having a value is obtained as the third representative sampling signal SSMAX3, and the fourth sampling signal group SSG [4] is a third NP having the largest signal value. The sampling signals SS [4NP] are acquired as the fourth representative sampling signals SSMAX4, respectively, and the remaining fifth to Rth sampling signal groups SSG [5] to SSG [R] are similarly applied to the fifth to fourth sampling signals SSMAX4. Obtains R-th representative sampling signals SSMAX5 to SSMAXR. That is, representative sampling signals of FIG. 5C are acquired (step S3).

After arranging the plurality of representative sampling signals SSMAX1 to SSMAXR obtained in step S3 in descending order, a noise signal is included to remove some representative sampling signals having a relatively high signal value. The period PN / the period PT x 100% " of the target signal is deleted from the aligned representative sampling signals (step S4).

Accordingly, as shown in (d), all of the representative sampling signals including the noise signal among the plurality of representative sampling signals SSMAX1 to SSMAXR are removed, and the representative sampling signal obtained by the input signal as shown in (d). Only the fields remain.

As shown in FIG. 5E, the average value M of the remaining representative sampling signals is calculated (step S5), and the input signal is clipped according to the average value M calculated through step S5 as shown in FIG. 5F. The output signal is output (step S6).

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art will be variously modified and changed within the scope of the invention without departing from the spirit and scope of the invention described in the claims below. I can understand that you can.

The noise filtering apparatus and method of the present invention filter target signals in a time domain to easily separate harmonic components of a noise signal that are not separated in a frequency domain. This increases the reliability of the filtering operation.

Claims (8)

A signal input unit which analyzes an input signal including a target signal and a noise signal, determines a period of the target signal, a period of the noise signal, and a sampling group signal period, and determines the number of repetitions of the target signal period within the sampling group signal period; A sampling signal group generation unit classifying the input signal according to the number of repetitions of the target signal period to generate a plurality of sampling signal groups; A representative sampling signal obtaining unit obtaining a maximum value of each of the plurality of sampling signal groups; The maximum value (minimum value) having a relatively large value (small value) among the maximum values obtained through the representative sampling signal acquisition unit is deleted according to the probability of generating a noise signal, and then the average value of the maximum value (minimum value) that is not deleted is deleted. An average value calculator for calculating; And And a signal filter for clipping the input signal according to the average value. The method of claim 1, wherein the noise signal generation probability And a "2 x noise signal period / target signal period". The method of claim 1, wherein the signal input unit And a function of setting a sampling period of the target signal. The method of claim 3, wherein the sampling signal group generation unit And generating a plurality of sampling signal groups by generating a plurality of sampling signals sampled according to the sampling period and classifying the plurality of sampling signals according to the number of repetitions of the target signal period. The method of claim 4, wherein the representative sampling signal acquisition unit And a maximum value (minimum value) of signal values of a plurality of sampling signals included in each sampling signal group. When receiving an input signal including a noise signal and a target signal, a signal input for determining the period of the target signal, the period of the noise signal and the sampling group signal period, and the number of repetitions of the target signal period within the sampling group signal period step; Generating a plurality of sampling signal groups by classifying the input signals according to the number of repetitions of the target signal period; A maximum value obtaining step of obtaining a maximum value (minimum value) of each of the plurality of sampling signal groups; An average value for calculating average values of the maximum values (minimum values) that are not deleted after deleting the maximum values (minimum values) having relatively large values (minimum values) among the obtained maximum values (minimum values) according to a noise signal occurrence probability Calculating step; And clipping the target signal according to the average value. The method of claim 6, wherein the noise signal generation probability And a "2 x noise signal period / target signal period". The method of claim 7, wherein the average value calculating step Sorting the obtained maximum values (minimum values) in descending order (ascending order); Deleting predetermined maximums (minimum values) in descending order (ascending order) according to the noise signal occurrence probability; And And calculating an average value of the maximum values (minimum values) that are not deleted.

Images