JP4174550B2 - Noise canceller and noise removal method - Google Patents

Description

  The present invention relates to a noise canceller that removes noise mixed in an input signal.

  In the field of signal processing technology, it is important to remove external noise mixed in an input signal to be processed.

  For example, in-vehicle broadcast receivers mounted on automobiles are easily affected by external noise such as ignition noise. Therefore, in general, processing such as detection and demodulation is not immediately performed on an intermediate frequency signal (IF signal) whose frequency is converted by the front end unit of the broadcast receiver, and external noise mixed in the intermediate frequency signal is noise canceller. After the removal, the processing such as detection and demodulation is performed.

  Conventionally, a noise canceller called a gate hold circuit shown in FIG.

  The noise canceller detects a noise mixing period of noise mixed in the input signal Sin and generates a gate signal GT, and sets the noise mixing period as an interpolation period based on the gate signal GT. And an interpolation processing unit 2 that performs an interpolation process for removing noise in the input signal Sin.

  That is, as shown in the waveform diagram of FIG. 1B, when the noise Nz is not mixed in the input signal Sin, the input signal Sin is output as it is via the interpolation processing unit 2, and at time t1, the noise Nz is reduced. When the noise period detection unit 1 detects that it has been mixed, the interpolation processing unit 2 gate-holds the signal component of the input signal Sin, and outputs the held constant level as an interpolation signal. Furthermore, after the interpolation processing unit 2 continuously outputs the interpolation signal in the interpolation period until the noise time detection unit 1 detects the end point t2 of the noise Nz, the input signal Sin is switched and output at the time point t2. Thereby, the output signal Sout output from the interpolation processing unit 2 becomes an input signal Sin in a period other than the noise mixing period, and becomes an interpolation signal at a constant level in the noise mixing period (interpolation period). The input signal Sin and the interpolation signal Is a time-synthesized waveform, thereby eliminating the noise Nz mixed in the input signal Sin.

  By the way, in the noise canceller called the conventional gate hold circuit, if the noise is mixed when the input signal Sin has a large amplitude, the level of the interpolation signal in the interpolation period also increases. When switching to the input signal Sin at the end point of noise (end point of the interpolation period) t2, the waveform of the output signal Sout changes steeply and discontinuously at a large level, and a high energy and high frequency noise component is generated. There is a problem.

  In order to determine the interpolation period, it is necessary to provide a buffer memory or the like for storing noise from the noise detection time t1 to the end time t2 in the noise period detection unit 1 or the like. When long-term noise exceeding the storage capacity of the memory is mixed, a so-called overflow state occurs, and the noise end time t2 shown in FIG. 1B cannot be detected, and the interpolation process is adversely affected. There is a problem that it is difficult to generate a simple interpolation signal.

  Further, in order to avoid such an overflow state, it is conceivable to increase the storage capacity of the buffer memory, but there is a problem that the circuit scale increases.

  The present invention has been made in view of such conventional problems, and an object of the present invention is to provide a noise canceller that does not internally generate noise during interpolation processing for an input signal and that can simplify the circuit scale. .

  The invention according to claim 1 is a noise canceller that removes noise mixed in the input signal via a storage unit that delays the input signal for a predetermined time and outputs the noise, and is mixed in the input signal. Noise detecting means for detecting noise; and an interpolation means for setting an interpolation period based on a detection result detected by the noise detecting means, and generating an interpolation signal by linearly approximating the input signal in the interpolation period. The interpolation means divides the interpolation period into divided interpolation periods shorter than the delay time when the interpolation period is longer than the delay time, and linearly approximates the input signal in each divided interpolation period. Generating an interpolated signal.

  The invention according to claim 2 is a noise canceller that removes noise mixed in the input signal via a storage means for delaying the input signal for a predetermined time and outputs the noise, and is mixed in the input signal. Noise detection means for detecting noise; and interpolation means for setting an interpolation period based on a detection result of the noise detection means, and generating an interpolation signal by performing a gate hold or linear approximation on the input signal in the interpolation period. The interpolation means generates an interpolation signal by holding the signal component immediately before the interpolation period in the interpolation period when the level of the signal component immediately before the interpolation period of the input signal is lower than a predetermined level. When the level of the signal component immediately before the interpolation period is greater than the predetermined level and the interpolation period is within the delay time, the interpolation In the meantime, an interpolation signal is generated by linear approximation of the signal component immediately before the interpolation period and the signal component at the end of the noise, the level of the signal component immediately before the interpolation period is greater than the predetermined level, and the interpolation period is When longer than the delay time, the interpolation period is divided into divided interpolation periods shorter than the delay time, and in each divided interpolation period, the input signal is linearly approximated to generate an interpolation signal.

  A preferred embodiment of the present invention will be described with reference to FIG. FIG. 2 is a block diagram showing the configuration of the noise canceller of this embodiment.

  In FIG. 2, the noise canceller 5 includes a storage unit 6, a noise period detection unit 7, an interpolation unit 8, and a selector unit 9. The storage unit 6 is not configured to be included in the noise canceller 5. For example, a buffer memory or the like provided for temporarily storing a received signal or the like when a digital broadcast receiver or the like performs detection or demodulation is used. May be used.

  The storage unit 6 buffers the input signal Sina, delays it by a predetermined delay time TD, and outputs the delayed input signal Sinb. Further, it has a function of outputting the buffered input signal Sina as a reference signal Sinnz in response to a request from the interpolation unit 8.

  The noise period detection unit 7 detects the noise Nz mixed in the input signal Sina, and outputs a gate signal GT indicating a noise mixing period from the noise start point to the end point of the noise Nz.

  The selector unit 9 is a switching unit that performs switching operation according to the switching control signal SW from the interpolation unit 8, and switches between the input signal Sinb output from the storage unit 6 and the interpolation signals CA and CB output from the interpolation unit 8. Output as output signal Sout.

  The interpolation unit 8 detects a noise mixing period of the noise Nz mixed in the input signal Sina by the gate signal GT, and sets an interpolation period Tc corresponding to the noise mixing period. That is, when the noise start time of the noise Nz is expressed as ts and the end time is expressed as te, the start time of the interpolation period Tc is shifted by the delay time TD (ts + TD), and the end time of the interpolation period Tc is shifted by the delay time TD. By setting the time (te + TD), the interpolation period Tc is set.

  Further, when detecting the noise start time ts by the gate signal GT, the interpolation unit 8 acquires the input signal Sina at the noise start time ts as the reference signal Sinnz, and compares the level of the reference signal Sinnz with a predetermined threshold value THD.

  When it is determined that the reference signal Sinnz is smaller than the threshold value THD, it is determined that the level of the input signal Sina is also small, the reference signal Sinnz compared with the threshold value THD is gate-held, and the held constant level is maintained during the interpolation period Tc. The interpolation signal CA is output as an output signal Sout through the selector unit 9.

  On the other hand, if it is determined that the reference signal Sinnz is greater than the threshold value THD, it is determined that the level of the input signal Sina is also large, and the following processing is performed without performing gate hold. That is, the input signal Sina at the noise start time ts is acquired as the reference signal Sinnz (ts), and the input signal Sina at the end time te of the noise Nz is acquired as the reference signal Sinnz (te). The rate of change ((Sinnz (te) −Sinnz (ts)) / Tc) of the interpolation period Tc with respect to the difference between (ts) and Sinnz (te) (Sinnz (te) −Sinnz (ts)) is calculated. Further, an interpolation signal (that is, an interpolation signal that is linearly approximated) CB that changes at the rate of change (Sinnz (te) −Sinnz (ts)) / Tc, with the level of the reference signal Sinnz (ts) as an initial value, is converted into an interpolation period. During Tc, the output signal Sout is output via the selector unit 9.

  Further, the interpolation unit 8 has a longer noise mixing period of the noise Nz than the delay time TD of the storage unit 6, so that the above-described reference signals Sinnz (ts) and Sinnz (te) are not buffered in the storage unit 6. In this case, the reference signal Sinnz (tp) corresponding to the noise at the mid-point tp that has been input to the storage unit 6 and buffered after the noise start time ts is obtained. Then, the difference (Sinnz (tp) −Sinnz (ts)) between the reference signal Sinnz (ts) at the noise start time ts and the reference signal Sinnz (tp) at the midway time tp is calculated, and at the midpoint from the noise start time ts. The period up to tp is calculated as the divided interpolation period (tp−ts), and the rate of change of the divided interpolation period (tp−ts) with respect to the difference (Sinnz (tp) −Sinnz (ts)) ((Sinnz (tp) −Sinnz (ts)) / (tp-ts)) is calculated. An interpolation signal that changes at the rate of change ((Sinnz (tp) −Sinnz (ts)) / (tp−ts)) with the level of the reference signal Sinnz (ts) as an initial value (ie, an interpolation signal that is linearly approximated) CB is output as an output signal Sout through the selector unit 9.

  Furthermore, when the reference signal Sinnz (te) corresponding to the noise end time te can be acquired by the end of the first divided interpolation period, the interpolation unit 8 ends with the reference signal Sinnz (tp) at the midpoint tp. The difference (Sinnz (te) −Sinnz (tp)) from the reference signal Sinna (te) at the time point te is calculated, and the period from the intermediate time point tp to the end time point te is calculated as the second divided interpolation period (te−tp). And the rate of change ((Sinnz (te) −Sinnz (tp)) / (te−tp) of the second divided interpolation period (te−tp) with respect to the difference (Sinnz (te) −Sinnz (tp)) ) Is calculated. An interpolation signal that changes at the rate of change (Sinnz (te) −Sinnz (tp)) / (te−tp) with the level of the reference signal Sinnz (tp) as an initial value (ie, an interpolation signal that is linearly approximated) CB Is output as an output signal Sout via the selector unit 9. As a result, when the noise Nz is mixed for a longer time than the delay time TD, the entire interpolation period Tc corresponding to the noise mixing period is divided into a plurality of divided interpolation periods. In each divided interpolation period, the noise Nz Interpolation is performed by generating an interpolation signal CB by linear approximation following the change in level.

  Furthermore, when the interpolation unit 8 cannot acquire the reference signal Sinnz corresponding to the noise end time te even in the second and subsequent divided interpolation periods, the reference signal corresponding to the noise mixed at the next intermediate time tn. Get Sinnz (tn). Then, the difference (Sinnz (tn) −Sinnz (tp)) between the reference signal Sinnz (tp) corresponding to the time point tp and the reference signal Sinnz (tn) corresponding to the intermediate time point tn is calculated, and the intermediate time point from the time point tp. The period up to tn is calculated as a divided interpolation period (tn-tp), and the rate of change of the divided interpolation period (tn-tp) with respect to the difference (Sinnz (tn) -Sinnn (tp)) ((Sinnz (tn) -Sinnz (tp)) / (tn-tp)). Then, an interpolation signal CB is generated and output by linear approximation between the levels of the reference signals Sinnz (tp) and Sinnz (tn) at intermediate points tp and tn, and finally, the reference signal Sinnz corresponding to the noise end point te. If (te) can be acquired, the same process as the second interpolation described above is performed to complete the interpolation process. As a result, even when the noise Nz is mixed for a longer time than the delay time TD, the entire interpolation period Tc corresponding to the noise mixing period is divided into a plurality of divided interpolation periods. Interpolation is performed by generating an interpolation signal CB by linear approximation that follows the change in the level of Nz.

  Further, when the interpolation unit 8 determines that it is a period other than the noise mixing period based on the gate signal GT, the selector unit 9 switches and controls the input signal Sinb, which is a delayed output of the storage unit 6, to the selector unit 9. Output via.

  As described above, the noise canceller 5 of the present embodiment is an input in which noise output from the storage unit 6 is not mixed during a period in which noise is not mixed in the input signal Sina (period other than the interpolation period Tc). When the signal Sinb is output and the level of the input signal Sina is low, the interpolation signal CA is generated by gate hold and output during the interpolation period Tc. When the level of the input signal Sina is high, one or more linear approximations are performed. Since the interpolation signal CB to be generated is output during the interpolation period Tc, noise can be removed.

  Further, the noise canceller 5 of the present embodiment is not a simple one that temporally synthesizes the input signal Sinb not mixed with noise and the interpolation signal CA or CB into the output signal Sout, but the level of the input signal Sina. When the signal is small, the gate hold is performed. When the level of the input signal Sina is large, the interpolation is performed using the interpolation signal generated by the linear approximation. As a result, when the amplitude of the input signal Sina is small, the level of the interpolation signal CA by the gate hold also decreases, and is input from the interpolation signal CA at the end of the noise mixing period (in other words, at the end of the interpolation period Tc). Even if switching to a signal is performed, it is possible to prevent internally generating a high-energy noise component. Further, when the amplitude of the input signal Sina is large, interpolation is performed with the interpolation signal CB by linear approximation. Internal generation of components can be prevented.

  Furthermore, when the level of the input signal is large, or when noise Nz longer than the delay time TD of the storage unit 6 having a limited storage capacity is mixed, the above-described divided interpolation periods are successively connected. Thus, the interpolation period Tc is used, and interpolation period signals are generated by linear approximation in each divided interpolation period, and interpolation is performed. Therefore, even when the storage capacity of the storage unit 6 is limited, interpolation processing can be performed.

  And even if the storage capacity of the storage unit 6 is limited, the fact that the interpolation process can be performed eliminates the need to increase the storage capacity of the storage unit 6 and reduces the storage capacity of the storage unit 6. Therefore, a noise canceller having a simple configuration can be provided.

  Note that the noise canceller 5 according to the embodiment described above has a so-called program configuration configured by a microprocessor, a digital signal processor, or the like, but may be realized by a hardware configuration. In the case of a hardware configuration, a digital circuit configuration or an analog circuit configuration may be used.

  In the present embodiment, in order to determine the level of the input signal Sina, the determination is performed using the reference signal Sinnz corresponding to the input signal not mixed with noise. However, the level of the input signal Sina is determined by other methods. May be determined.

  Next, an embodiment will be described with reference to FIGS. 3A is a block diagram showing the configuration of the noise canceller of the present embodiment, FIG. 3B is a timing chart for explaining the operation of the noise canceller, and FIG. 4 is a waveform diagram for explaining the operation of the noise canceller. It is.

  In FIG. 3A, the noise canceller 5 is formed by a microprocessor (MPU) or a digital signal processor (DSP) that operates according to a computer program, and includes a storage unit 6, a noise period detection unit 7, and a first interpolation. A processing unit 8a, a second interpolation processing unit 8b, a control unit 8c, and a selector unit 9 are included. Here, the interpolation unit 8 shown in FIG. 3 includes a first interpolation processing unit 8a, a second interpolation processing unit 8b, and a control unit 8c.

<Configuration of storage unit 6>
The storage unit 6 is formed by a buffer memory or the like that outputs the input signal Sina after being delayed by a predetermined delay time TD by a FIFO (first in first out) process, and the selector unit 9 switches the delayed input signal Sinb. Supply to contact a.

Further, the storage unit 6 outputs the input signal being processed by the FIFO addressed from the second interpolation processing unit 8b to the second interpolation processing unit 8b as the reference signal Sinnz, and the FIFO addressed by the control unit 8c. The input signal being processed is output to the control unit 8c as a reference signal Sinnz.
That is, the storage unit 6 has a random access function.

<Configuration of Noise Period Detection Unit 7>
The noise period detector 7 detects a noise Nz mixed in the input signal Sina, and indicates a gate signal indicating a noise start time ts when the signal component of the input signal Sina is switched to the noise Nz and an end time te of the noise Nz. GT is supplied to the control unit 8c. Therefore, when the noise Nz is mixed for a long time, the gate signal GT indicating that the interval between the noise start time ts and the end time te is long, and when the noise Nz for a short time is mixed, the noise start A gate signal GT indicating that the interval between the time point ts and the end time point te is short is supplied to the control unit 8c.

<Configuration of Control Unit 8c>
Based on the gate signal GT, the control unit 8c detects a period from the noise start time ts to the end time te as a noise mixing period, and ends from the time (ts + TD) after the delay time TD has elapsed from the noise start time ts. A period up to a point (te + TD) when the delay time TD has elapsed is set as the interpolation period Tc.

  Further, the control unit 8c connects the selector unit 9 to the switching contact point a by the switching control signal SW during a period other than the interpolation period Tc, thereby inputting the delay output of the storage unit 6 that is not mixed with the noise Nz. The signal Sinb is output as the output signal Sout.

  Further, when detecting the noise start time ts by the gate signal GT, the control unit 8c acquires the reference signal Sinnz corresponding to the input signal Sina at the noise start time ts, and the level (amplitude) of the reference signal Sinnz is predetermined. It is determined whether it is smaller than the threshold value THD. Here, the threshold value THD is determined in advance in consideration of the amplitude of the input signal interpolated by the first interpolation processing unit 8a.

  When the control unit 8c determines that the level of the reference signal Sinnz is smaller than the threshold value THD, the switching control signal SW at the time (ts + TD) after the delay time TD has elapsed from the noise start time ts, that is, at the start time of the interpolation period Tc. At the same time as the selector unit 9 is connected to the switching contact b by the interpolation control signal CNT1, the first interpolation processing unit 8a starts the interpolation process, and further until the end of the interpolation period Tc specified by the interpolation control signal CNT1. The interpolation signal CA generated by the one interpolation processing unit 8a is output as the output signal Sout.

  If the control unit 8c determines that the level of the reference signal Sinnz is greater than the threshold value THD, the control unit 8c connects the selector unit 9 to the switching contact c by the switching control signal SW at the start of the interpolation period Tc. Simultaneously, the interpolation control signal CNT2 causes the second interpolation processing unit 8b to start interpolation processing, and further, the interpolation signal CB generated by the second interpolation processing unit 8b is output until the end of the interpolation period Tc specified by the interpolation control signal CNT2. The output signal Sout is output.

  As described above, when the control unit 8c controls the first and second interpolation processing units 8a and 8b and the selector unit 9, as shown in FIG. 3B, the time shift between the input signal Sina and the output signal Sout is performed. Is set according to the delay time TD of the storage unit 6, and when no noise is mixed in the input signal Sina, the input signal Sina (in other words, the input signal Sinb) is output as it is as the output signal Sout. Further, when the noise Nz is mixed in the input signal Sina having a small amplitude, the interpolation signal generated by the first interpolation processing unit 8a in the interpolation period Tc (indicated by “Tc = Tca” in FIG. 5B). When the CA is output as the output signal Sout and the noise Nz is mixed in the input signal Sina having a large amplitude, the second interpolation is performed in the interpolation period Tc (indicated by “Tc = Tcb” in FIG. 5B). The interpolation signal CB generated by the processing unit 8b is output as the output signal Sout.

  When the selector unit 9 is switched, the input signal Sinb that is not mixed with the noise Nz and the interpolation signal CA or CB generated by the first interpolation processing unit 8a or the second interpolation processing unit 8b are time-series. Therefore, the output signal Sout can be generated by removing the noise Nz from the input signal Sina.

<Configuration of First Interpolation Processing Unit 8a>
The first interpolation processing unit 8a performs interpolation processing by gate hold within the interpolation period Tc (interpolation period Tca in FIG. 3B) according to the interpolation control signal CNT1 supplied from the control unit 8c. In other words, as described above, when the control unit 8c determines that the level of the reference signal Sinnz is low and outputs the interpolation control signal CNT1 simultaneously with the switching control signal SW, the first interpolation processing unit 8a is configured to start the interpolation period Tc. The reference signal Sinnz corresponding to the input signal Sina input at (ts + TD) is gate-held, and the held constant level interpolation signal CA is supplied to the switching contact b of the selector unit 9 during the interpolation period Tca. By this interpolation processing, the signal component of the input signal Sina at the noise start time ts is gate-held, and the interpolation signal CA of a certain level from which the noise Nz has been removed within the interpolation period Tc is the switching contact b of the selector unit 9. Is output as an output signal Sout.

That is, when the first interpolation processing unit 8a performs the interpolation processing by the gate hold, as shown in FIG. 4A, in the interpolation period Tc corresponding to the noise mixing period from the noise start time ts to the end time te, the noise The signal component Xs of the input signal Sina at the start time ts is held, and a constant level of the interpolation signal CA is output as the output signal Sout until the end of the interpolation period Tc.
For this reason, the noise Nz is replaced with the interpolation signal CA of a certain level, and the noise Nz can be removed.

  Further, as shown in FIG. 4B, even when the noise Nz is mixed for a long period of time, the noise Nz is replaced with the interpolation signal CA at a certain level, and the noise Nz can be removed.

<Configuration of Second Interpolation Processing Unit 8b>
The second interpolation processing unit 8b performs interpolation processing within the interpolation period Tc (interpolation period Tcb in FIG. 3B) according to the interpolation control signal CNT2 supplied from the control unit 8c. Here, the second interpolation processing unit 8b has a function of switching between basic interpolation processing and adaptive interpolation processing according to the length of the noise mixing period of the noise Nz.

[Basic interpolation processing]
First, basic interpolation processing will be described. This basic interpolation process is performed when the noise mixing period from the noise start time ts to the end time te is relatively short compared to the storage capacity of the storage unit 6.

  That is, the second interpolation processing unit 8b detects the noise start time ts based on the gate signal GT before the interpolation control signal CNT2 is supplied, and acquires the input signal Sina at the noise start time ts as the reference signal Sinnz. The level of the reference signal Sinnz (hereinafter referred to as “initial level Xs”) is held. Further, the noise end time te is detected based on the gate signal GT within the preparation period TPR determined to be shorter than the delay time TD from the noise start time ts, and corresponds to the input signal Sina of the noise end time te. The reference signal Sinnz to be acquired is acquired, and the level of the reference signal Sinnz (hereinafter referred to as “final level Xe”) is held.

  Then, as expressed by the following expression (1), the interpolation control signal CNT2 is supplied by calculating the change rate α of the noise mixing period (te−ts) with respect to the difference between the initial level Xs and the final level Xe. Prepare for the interpolation process before it is done.

  That is, assuming that the input signal Sina at the noise mixing time ts is the reference signal Sinz (ts) and the input signal Sina at the noise end time te is represented by the reference signal Sinz (te), the second interpolation processing unit 8b The change rate α is obtained by performing an operation equivalent to (2).

  Next, when the interpolation control signal CNT2 is output at the same time as the switching control signal SW is output from the control unit 8c, the second interpolation processing unit 8b determines the initial value at the start time (ts + TD) of the interpolation period Tc. Is set to the initial level Xs, and as shown in the following equation (3), an interpolation signal CB whose level changes according to the rate of change α is generated as time t passes, and the interpolation period Tc is the end point (te + TD). Output until

  Thus, when the second interpolation processing unit 8b performs the basic interpolation processing, instead of the noise Nz, one linearly approximated interpolation signal CB is output as the output signal Sout through the selector unit 9, and the noise Nz is removed. can do.

  That is, when the second interpolation processing unit 8b performs the basic interpolation process, as shown in FIG. 4C, in the interpolation period Tc corresponding to the noise mixing period from the noise start time ts to the end time te, the noise start time Interpolation is performed by the interpolation signal CB that linearly approximates the signal component Xs of the input signal Sina of ts and the signal component Xe of the end time te of the interpolation period Tc, so that the noise Nz is replaced by the interpolation signal CB. Noise Nz can be removed.

[Adaptive interpolation processing]
As described above, the basic interpolation process is performed when noise Nz for a relatively short time is mixed. That is, when the noise Nz that continues from the noise start time ts to the end time te is shorter than the delay time TD of the storage unit 6, the basic interpolation processing based on the above equations (1) to (3) is performed. When Nz continues for a long period of time and becomes a so-called overflow state exceeding the delay time TD of the storage unit 6, the final level Xe of the input signal Sina at the noise end time te cannot be acquired as the reference signal Sinnz. The change rate α shown in 1) cannot be calculated, and the basic interpolation processing based on the above equations (1) to (3) cannot be performed.

  Therefore, the second interpolation processing unit 8b holds only the above-described initial level Xs in the preparation period TPR after starting the basic interpolation processing, and refers to the final level Xe of the input signal Sina at the noise end time te. If it is determined that the signal cannot be acquired as the signal Sinnz, the shift to the interpolation processing based on the above formulas (1) to (3) is stopped, and the adaptive interpolation processing is performed.

  When the second interpolation processing unit 8b shifts to the adaptive interpolation processing, the second interpolation processing unit 8b accesses the storage unit 6 within the preparation period TPR, and the reference signal having the noise Nz stored in the storage unit 6 temporally after the noise start time ts. Get Sinnz. Here, any reference signal Sinnz having the noise Nz stored in the storage unit 6 after the noise start time ts may be used, but the second interpolation processing unit 8b of the present embodiment is the storage unit 6. The reference signal Sinnz having the latest noise Nz stored in is acquired.

  The level of the reference signal Sinnz having the latest noise Nz (that is, the level of the noise Nz) is detected and set to the intermediate level Xp, and based on the address information in which the reference signal Sinnz having the latest noise Nz is stored. Then, the noise intermediate time point tp in which the latest noise Nz is actually mixed in the input signal Sina is detected, and the delay time TD from the noise intermediate time point tp from the time point (ts + TD) after the delay time TD has elapsed from the noise start time point ts. A period until the time point (tp + TD) after elapse is defined as a first interpolation period Tc1. Then, as expressed by the following equation (4), by calculating the change rate β of the first interpolation period Tc1 with respect to the difference between the initial level Xs and the intermediate level Xp, before the interpolation control signal CNT2 is supplied. Prepare for interpolation processing.

  That is, if the input signal Sina at the noise mixing time point ts is represented by the reference signal Sinz (ts) and the input signal Sina at the noise intermediate time point tp is represented by the reference signal Sinz (tp), the second interpolation processing unit 8b The change rate β is obtained by performing an operation equivalent to (5).

  Next, when the interpolation control signal CNT2 is output at the same time as the switching control signal SW is output from the control unit 8c, the second interpolation processing unit 8b determines the initial value at the start time (ts + TD) of the interpolation period Tc. Is set to the initial level Xs, and as shown in the following equation (6), an interpolation signal CB whose level changes in accordance with the rate of change β is generated with the lapse of time t, and from the time point (ts + TD) to the time point (tp + TD). ) During the first interpolation period Tc1, which is the period until.

  When interpolation processing is performed within the first interpolation period Tc1, the level Xs of the input signal Sina at the noise start time ts and the noise intermediate level Xp at the noise intermediate time tp as shown in FIG. Is output as the output signal Sout, so that the noise Nz in the first interpolation period Tc1 can be removed.

  Further, the second interpolation processing unit 8b determines one by one whether the noise end time te can be detected from the gate signal GT in parallel with the generation of the above-described interpolation signal CB within the first interpolation period Tc1. If it is determined that it has been detected, the convergence interpolation processing described below is performed. If it is determined that it cannot be detected, extended interpolation processing is performed.

[Convergent interpolation processing]
As described above, the convergence interpolation process is performed when the second interpolation processing unit 8b detects the noise end time te from the gate signal GT within the first interpolation period Tc1. The second interpolation processing unit 8b determines that the noise end time te has been detected, accesses the storage unit 6, and stores the reference signal Sinnz of the noise end time te already stored in the storage unit 6 within the first interpolation period Tc1. Level (final level) Xe is acquired. The second interpolation period Tc2 is a period from the time point (tp + TD) after the delay time TD has elapsed from the noise intermediate time point tp already determined to the time point (te + TD) after the delay time TD has elapsed since the noise end time te. And Further, as represented by the following equation (7), a change rate γ of the second interpolation period Tc2 with respect to the difference between the intermediate level Xp and the final level Xe is calculated.

  Then, when the first interpolation period Tc1 ends, the second interpolation processing unit 8b continues to set the second interpolation period Tc2, sets the initial value at the start point (tp + TD) of the second interpolation period Tc2 to the intermediate level Xp, and then As represented by Expression (8), an interpolation signal CB whose level changes according to the change rate γ is generated as time t passes, and is output within the period of the second interpolation Tc2.

  Thus, when the second interpolation processing unit 8b sets the first interpolation period Tc1 and the second interpolation period Tc2 and performs the adaptive interpolation process, as shown in FIG. 4D, in the first interpolation period Tc1, the interpolation signal CB1. In the second interpolation period Tc2, interpolation by the interpolation signal CB2 is performed, and further, the interpolation signals CB1 and CB2 are interpolated with continuity, so that the input signal Sina is generated without generating noise internally. It is possible to remove long-time noise Nz mixed in the.

  That is, in the adaptive interpolation process, the entire interpolation period Tc corresponding to the noise mixing period is time-divided into a first interpolation period Tc1 and a second interpolation period Tc2, and interpolation is performed using the interpolation signal CB1 in the first interpolation period Tc1. Since the convergence interpolation process is performed in the two interpolation period Tc2 and the interpolation by the interpolation signal CB2 is performed, even when a long-time noise Nz is mixed as compared with the storage capacity of the storage unit 6, the noise Nz can be removed. .

[Extended interpolation processing]
As described above, the extended interpolation process is performed when the second interpolation processing unit 8b cannot detect the noise end time te from the gate signal GT within the first interpolation period Tc1. If the second interpolation processing unit 8b determines that the noise end time te cannot be detected,
The reference signal Sinnz corresponding to the latest input signal Sina having the noise Nz is acquired, the level of the reference signal Sinnz is detected and set to the second intermediate level Xn, and the reference signal Sinnz having the latest noise Nz is stored. Based on the address information, the latest noise Nz is detected as a noise intermediate time tn when it is actually mixed in the input signal Sina, and the noise is detected from the time (ts + TD) after the delay time TD has elapsed from the noise start time ts. A period from the intermediate time point tn to the time point (tp + TD) after the delay time TD has elapsed is defined as a second interpolation period Tc2. Then, instead of the final level Xe in the above formulas (7) and (8), the second intermediate level Xn is substituted, and the same processing as the above formulas (7) and (8), that is, In the second interpolation period Tc2, the interpolation signal CB is generated and the interpolation process is performed.

  Thus, when the second interpolation processing unit 8b performs the extended interpolation process, even if the noise Nz still remains at the end of the second interpolation period Tc2, as shown in FIG. Since interpolation is performed by the signal CB2, the noise Nz in the second interpolation period Tc2 can be removed.

  Further, when the noise end time te can be detected from the gate signal GT within the second interpolation period Tc2, the second interpolation processing unit 8b performs the same process as the above-described convergence interpolation process, thereby performing the second interpolation period. A third interpolation period Tc3 is set subsequent to Tc2, and the interpolation signal CB3 is generated and output until the end of the third interpolation period Tc3, thereby completing the adaptive interpolation process.

  If the noise end point te cannot be detected from the gate signal GT within the second interpolation period Tc2, the above-described extended interpolation process is further repeated until the noise end point te can be detected from the gate signal GT. The process is performed to complete the adaptive interpolation process.

  As described above, when the second interpolation processing unit 8b performs the extended interpolation process, it is possible to remove the noise Nz mixed for a longer time. That is, the entire interpolation period Tc corresponding to the noise mixing period is time-divided into a number of interpolation periods such as the first interpolation period Tc1, the second interpolation period Tc2, and the third interpolation period Tc3, and the previous interpolation period is excluded. Interpolation that follows noise in a plurality of interpolation periods is performed, and then convergence interpolation processing is performed in the last interpolation period. Therefore, even when noise Nz that is longer than the storage capacity of the storage unit 6 is mixed, the noise Nz can be removed.

  As described above, according to the noise canceller 5 of the present embodiment, the input signal is output as the output signal Sout during the period when no noise is mixed in the input signal Sina, and the input has an amplitude smaller than the predetermined threshold value THD. When noise is mixed in the signal Sina, the interpolation signal CA generated by performing the gate hold in the first interpolation processing unit 8a is output as the output signal Sout, and the input signal Sina having an amplitude larger than a predetermined threshold value THD is output. When noise is mixed, the second interpolation processing unit 8b performs basic interpolation processing and adaptive interpolation processing to output an interpolation signal CB based on one or a plurality of linear approximations as the output signal Sout, so that noise is removed. be able to.

  Furthermore, the noise canceller 5 of the present embodiment is not simply switched by the first and second interpolation processing units 8a and 8b, and the first interpolation processing unit 8a is used when the amplitude of the input signal Sina is smaller than the threshold value THD. When the amplitude of the input signal Sina is larger than the threshold value THD, the second interpolation processing unit 8b performs an interpolation process.

  For this reason, when the amplitude of the input signal Sina is smaller than the threshold value THD, the level of the interpolation signal CA generated by the first interpolation processing unit 8a also decreases, and at the end of the noise mixing period (in other words, the interpolation period Tc). Even when the interpolation signal CA is switched to the input signal at the time of completion, it is possible to prevent internally generating a high energy noise component.

  On the other hand, when the amplitude of the input signal Sina is larger than the threshold value THD, the above-described high-energy noise component can be prevented from being internally generated by switching from the first interpolation processing unit 8a to the second interpolation processing unit 8b. .

  Further, by performing basic interpolation processing and adaptive interpolation processing, even if the storage capacity of the storage unit 6 is limited, it is possible to perform interpolation processing.

  And even if the storage capacity of the storage unit 6 is limited, the fact that the interpolation process can be performed eliminates the need to increase the storage capacity of the storage unit 6 and reduces the storage capacity of the storage unit 6. Therefore, a noise canceller having a simple configuration can be provided.

  Next, an embodiment of a noise canceller provided in an in-vehicle FM receiver that receives FM broadcast will be described with reference to FIG. FIG. 5 is a block diagram showing the configuration of the noise canceller of this embodiment, and the same or corresponding parts as those in FIG.

  In FIG. 5, the noise canceller 5 of this embodiment inputs a baseband signal detected by the front end unit FE provided in the FM receiver as an input signal Sina, removes external noise such as ignition noise, An output signal Sout, which is a baseband signal after noise removal, is output to the demodulator.

  That is, the front end unit FE of the FM receiver mixes a high-frequency reception signal (RF reception signal) generated in the reception antenna ANT and a local oscillation signal of the tuning frequency to perform frequency conversion, thereby performing an intermediate frequency signal (IF signal). , And the intermediate frequency signal is band-limited and amplified to an amplitude level that allows signal processing, thereby generating a baseband signal as a desired signal component. Then, the baseband signal is supplied to the noise period detection unit 7 and the storage unit 6 as the input signal Sina.

  The noise period detector 7 includes a high-pass filter 7a, a smoothing circuit 7b, and a comparator 7c. The filter 7a passes the external noise mixed in the input signal Sina, and the smoothing circuit 7b smoothes the noise. A smoothing noise component is generated by the comparator 7a having a hysteresis and the level of the smoothing noise component is compared with a predetermined comparison voltage, and a period in which the level of the smoothing noise component is larger than the comparison voltage is detected as a noise mixing period. Then, the gate signal GT is supplied to the control unit 8c and the second interpolation processing unit 8b.

  The storage unit 6 outputs the input signal Sinb obtained by buffering the input signal Sina and delaying the input signal Sina by a predetermined delay time TD.

  The first and second interpolation processing units 8a and 8b and the selector unit 9 have the same functions as the first and second interpolation processing units 8a and 8b and the selector unit 9 shown in FIG.

  The control unit 8c has basically the same function as the control unit 8c shown in FIG. However, instead of comparing the input signal Sina with the threshold value THD and determining the amplitude of the input signal Sina, a reception sensitivity signal CN indicating the reception sensitivity (C / N) output from the front end unit FE is input, The reception sensitivity signal CN is compared with a predetermined threshold value THD. Then, when the control unit 8c detects the mixing of the noise Nz based on the gate signal GT, when the reception sensitivity signal CN is smaller than the threshold value THD, the control unit 8c determines that the amplitude of the input signal Sina is small, and the first interpolation processing unit When the reception sensitivity signal CN is larger than the threshold value THD, it is determined that the amplitude of the input signal Sina is large, and the second interpolation processing unit 8b performs the interpolation processing.

  In the noise canceller 5 of this embodiment, the control unit 8c controls the switching of the selector unit 9 and the switching control of the first and second interpolation processing units 8a and 8b, similarly to the noise canceller 5 shown in FIG. As a result, the noise Nz mixed in the input signal Sina, which is a baseband signal, is removed, and the output signal Sout after the noise removal is supplied to a demodulator or the like.

  According to the noise canceller 5 of the present embodiment, when noise is not mixed in the baseband signal, the baseband signal is output as an output signal Sout to a demodulator or the like. Since the interpolation signal CA or CB generated by the interpolation processing unit 8a or the second interpolation processing unit 8b is output to the demodulator or the like as the output signal Sout, noise can be removed.

  Further, when the reception sensitivity is smaller than the threshold value THD, the first interpolation processing unit 8a performs an interpolation process for a baseband signal having a small amplitude, thereby preventing a high energy noise component from being generated internally, and the reception sensitivity is reduced. When it is larger than the threshold value THD, switching from the first interpolation processing unit 8a to the second interpolation processing unit 8b can prevent the above-described high energy noise component from being generated internally. Furthermore, when long-term noise is mixed, the second interpolation processing unit 8b performs basic interpolation processing and adaptive interpolation processing to perform interpolation processing even if the storage capacity of the storage unit 6 is limited. Making it possible.

  And even if the storage capacity of the storage unit 6 is limited, the fact that the interpolation process can be performed eliminates the need to increase the storage capacity of the storage unit 6 and reduces the storage capacity of the storage unit 6. Therefore, it is possible to provide a noise canceller with a simple configuration, and to contribute to simplification of the configuration of the receiver. In particular, since it is possible to have a simple configuration, the present invention is effective when applied to an in-vehicle type or a portable type receiver.

It is a figure for demonstrating the structure and function of the conventional noise canceller. It is a block diagram showing the structure of the noise canceller which concerns on embodiment of this invention. 3 is a block diagram illustrating a configuration of a noise canceller according to an embodiment and a timing chart for explaining an operation. FIG. It is a wave form diagram for demonstrating operation | movement of the noise canceller shown in FIG. It is a block diagram showing the structure of the noise canceller which concerns on another Example.

Explanation of symbols

5 ... Noise canceller 6 ... Storage unit 7 ... Noise period detection unit 8 ... Interpolation unit

Claims (4)

A noise canceller that removes noise mixed in the input signal through a storage means that delays the input signal for a predetermined time, and outputs the noise.
Noise detection means for detecting noise mixed in the input signal;
An interpolation unit configured to set an interpolation period based on a detection result detected by the noise detection unit, and generate an interpolation signal by linearly approximating the input signal in the interpolation period;
When the interpolation period is longer than the delay time, the interpolation means divides the interpolation period into divided interpolation periods shorter than the delay time, and linearly approximates the input signal in each divided interpolation period to obtain an interpolation signal. Generating,
Noise canceller characterized by A noise canceller that removes noise mixed in the input signal through a storage means that delays the input signal for a predetermined time, and outputs the noise.
Noise detection means for detecting noise mixed in the input signal;
Interpolation means for setting an interpolation period based on the detection result of the noise detection means, and generating an interpolation signal by gate holding or linear approximation of the input signal in the interpolation period,
The interpolation means includes
When the level of the signal component immediately before the interpolation period of the input signal is lower than a predetermined level, an interpolation signal is generated by holding the signal component immediately before the interpolation period in the interpolation period,
When the level of the signal component immediately before the interpolation period is greater than the predetermined level and the interpolation period is within the delay time, a straight line between the signal component immediately before the interpolation period and the signal component at the end of noise in the interpolation period. Generate an interpolated signal by approximation,
When the level of the signal component immediately before the interpolation period is greater than the predetermined level and the interpolation period is longer than the delay time, the interpolation period is divided into divided interpolation periods shorter than the delay time, and in each divided interpolation period, Generating an interpolation signal by linearly approximating the input signal,
Noise canceller characterized by The storage means and the noise detection means input a reception signal generated by a receiver as the input signal,
The interpolation means determines a level of a signal component immediately before the interpolation period of the input signal based on a reception sensitivity signal detected by the receiver;
The noise canceller according to claim 2. A noise removal method for removing and outputting noise mixed in the input signal through a storage means for delaying the input signal for a predetermined time,
A noise detection step of detecting noise mixed in the input signal;
An interpolation period is set based on a detection result detected in the noise detection process, and an interpolation signal is generated by linearly approximating the input signal in the interpolation period,
In the interpolation step, when the interpolation period is longer than the delay time, the interpolation period is divided into divided interpolation periods shorter than the delay time, and in each divided interpolation period, the input signal is linearly approximated to obtain an interpolation signal. Generating,
The noise removal method characterized by this.

Images