JP2005039659A - Device and method for correcting signal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an integrating filter circuit capable of eliminating noise from an input signal with high accuracy while suppressing the deterioration of dynamic resolution of signal components. <P>SOLUTION: A differential signal D(n) between a delay signal Y(n-1) obtained by delaying an output signal Y(n) for a prescribed period of time and an input signal X(n) is calculated. A multiplication signal A(n) obtained by multiplying a nonlinear signal S(n) obtained by performing nonlinear processing of the differential signal D(n) in accordance with amplitude by a cyclic coefficient (a) is added to the input signal X(n) to be the output signal Y(n). Since noise of the input signal X(n) can be detected with high accuracy in accordance with amplitude of the differential signal D(n), the noise can be eliminated from the input signal X(n) with high accuracy and easily while suppressing the deterioration of the dynamic resolution of the signal components. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a signal correction apparatus and a signal correction method for correcting an input signal and outputting it as an output signal.

2. Description of the Related Art Conventionally, as this type of signal correction device, a configuration including a cyclic filter that removes noise mixed in an input signal such as an image signal from the input signal is known. This cyclic filter has a delay unit that outputs a delayed signal obtained by delaying the output signal for a predetermined time. The delay unit is connected to a multiplier that multiplies the delay signal output from the delay unit by a cyclic coefficient and outputs the product as a multiplication signal. Further, an adder is connected to the multiplier to add the multiplication signal output from the multiplier to the input signal to produce an output signal (see, for example, Patent Document 1).
JP-A-9-284174 (page 3-4, FIG. 1)

  However, since the signal correction apparatus described above only adds the multiplied signal obtained by multiplying the delayed signal by the cyclic coefficient by the multiplier to the input signal, even the signal component that is not originally removed even if the cyclic coefficient is adjusted is noisy. In addition, there is a problem that the dynamic resolution of the signal component may be reduced.

  The present invention has been made in view of these points, and an object of the present invention is to provide a signal correction apparatus that can accurately remove noise from an input signal while suppressing a decrease in dynamic resolution of signal components.

  The signal correction device according to claim 1 is a signal correction device that corrects an input signal having a signal component and noise and outputs the signal as an output signal, and a delay unit that outputs a delay signal obtained by delaying the output signal by a predetermined time; A subtractor for calculating a differential signal between the delayed signal output from the delay unit and the input signal; and a non-linear process on the differential signal calculated by the subtractor according to the amplitude and outputting the result as a non-linear signal A non-linear processing unit, a multiplying unit that multiplies the non-linear signal output from the non-linear processing unit by a cyclic coefficient and outputs the result as a multiplication signal, and adds the multiplication signal output from the multiplication unit to the input signal. And an adder that serves as an output signal.

  Then, the differential signal between the delay signal delayed by a predetermined time by the delay unit and the input signal is subjected to nonlinear processing by the nonlinear processing unit according to the amplitude of the difference signal, the multiplication unit is multiplied by the cyclic coefficient, and the input signal is input by the addition unit. By adding to the output signal, the noise of the input signal can be accurately detected according to the amplitude of the difference signal, so that the noise can be accurately removed from the input signal while suppressing the decrease in dynamic resolution of the signal component. Can easily be done.

  The signal correction device according to claim 2 is the signal correction device according to claim 1, wherein the nonlinear processing unit multiplies the difference signal between the delayed signal delayed by the delay unit and the input signal by an ε function and outputs the resultant signal. It is.

  Then, by multiplying the difference signal by the ε function, it is possible to arbitrarily set the amplitude of noise to be removed from the input signal according to the ε function. It is possible to easily remove noise from the signal with higher accuracy.

  The signal correction method according to claim 3 is a signal correction method for correcting an input signal having a signal component and noise and outputting the output signal as an output signal, delaying the output signal for a predetermined time, and the delayed output signal and The differential signal with respect to the input signal is subjected to nonlinear processing, the differential signal subjected to the nonlinear processing is multiplied by a cyclic coefficient and added to the input signal to obtain an output signal.

  Then, the differential signal between the output signal delayed by a predetermined time and the input signal is nonlinearly processed according to the amplitude of the differential signal, and the cyclic processing unit multiplies the cyclic coefficient to add to the input signal to obtain the output signal. As a result, the noise of the input signal can be accurately detected in accordance with the amplitude of the differential signal, so that it is possible to easily remove the noise from the input signal while suppressing a decrease in the dynamic resolution of the signal component. .

  A signal correction method according to a fourth aspect is the signal correction method according to the third aspect, wherein the differential signal between the delayed output signal and the input signal is subjected to non-linear processing and multiplied by an ε function.

  Then, by multiplying the difference signal by the ε function, it is possible to arbitrarily set the amplitude of noise to be removed from the input signal according to the ε function. It is possible to easily remove noise from the signal with higher accuracy.

  According to the signal correction device of the first aspect, the differential signal between the delay signal delayed by the delay unit for a predetermined time and the input signal is nonlinearly processed by the nonlinear processor according to the amplitude of the difference signal, and the multiplier By multiplying the cyclic coefficient and adding it to the input signal by the adder to produce an output signal, the noise of the input signal can be accurately detected according to the amplitude of the differential signal, while suppressing the reduction in the dynamic resolution of the signal component Noise can be accurately and easily removed from the input signal.

  According to the signal correction apparatus of the second aspect, in addition to the effect of the signal correction apparatus of the first aspect, the difference signal is multiplied by the ε function, whereby the amplitude of noise to be removed from the input signal is determined according to the ε function. Therefore, noise can be removed from the input signal more accurately and easily while suppressing a decrease in the dynamic resolution of the signal component.

  According to the signal correction method of the third aspect, the differential signal between the output signal delayed by a predetermined time and the input signal is nonlinearly processed according to the amplitude of the differential signal, and the cyclic processing unit multiplies the cyclic coefficient. By adding the input signal to the output signal, the noise of the input signal can be accurately detected according to the amplitude of the difference signal, so the noise from the input signal can be accurately detected while suppressing the decrease in the dynamic resolution of the signal component. And can be easily removed.

  According to the signal correction method of the fourth aspect, in addition to the effect of the signal correction method according to the third aspect, the difference signal is multiplied by the ε function so that the amplitude of noise to be removed from the input signal depends on the ε function. Therefore, noise can be removed from the input signal more accurately and easily while suppressing a decrease in the dynamic resolution of the signal component.

  Hereinafter, the configuration of an embodiment of a signal correction apparatus of the present invention will be described with reference to the drawings.

  In FIG. 1, reference numeral 1 denotes an integration filter circuit as a signal correction device. The integration filter circuit 1 is used in an image signal recording / reproducing device such as a hard disk recorder, for example, and is used for a time axis (frame / field) of a video signal or the like. The signal is input in the direction, and the video signal or the like is corrected and output.

  The integration filter circuit 1 includes an input unit 2 such as an input terminal to which an input signal X [n] having a signal component and noise is input, and an output unit 3 such as an output terminal from which the output signal Y [n] is output. And.

  A subtracting unit 4 is connected to the input unit 2, and the subtracting unit 4 outputs a delay signal Y [n-1] obtained by delaying the output signal Y [n] by a predetermined time, for example, one frame. Part 5 is connected. The subtracting unit 4 outputs a difference signal D [n] between the input signal X [n] from the input unit 2 and the delayed signal Y [n−1]. That is, D [n] = Y [n−1] −X [n].

  Further, the subtracting unit 4 is connected to a non-linear processing unit 6 that outputs a non-linear signal S [n] obtained by performing non-linear processing on the difference signal D [n]. The nonlinear processing unit 6 multiplies the difference signal D [n] by, for example, an ε (epsilon) function. That is, S [n] = ε × D [n] = ε × (Y [n−1] −X [n]). Here, the ε function refers to a nonlinear function that satisfies | ε | ≦ C with respect to an arbitrary threshold value C. In the present embodiment, for example, when the signal amplitude of the input signal X [n] is 8 bits, the ε function is a digital signal expressed by two's complement, that is, −128 to 127, so that the function is as shown in FIG. Set to.

  The nonlinear processing unit 6 is connected to a multiplication unit 7 that outputs a multiplication signal A [n] obtained by multiplying the nonlinear signal S [n] by a cyclic coefficient a. That is, A [n] = a × S [n] = a × ε × D [n] = a × ε × (Y [n−1] −X [n]). Here, the cyclic coefficient a is a number between 0 and 1. That is, 0 ≦ a ≦ 1.

  The multiplication unit 7 is connected to an addition unit 8 that adds the multiplication signal A [n] and the input signal X [n] and outputs the result to the output unit 3 as an output signal Y [n]. That is, the difference equation is Y [n] = A [n] + X [n] = a × ε × (Y [n−1] −X [n]) + X [n].

Therefore, the integral filter circuit 1 has a filter characteristic as shown in FIG. 3, and the transfer function H (z) is H (z) = (1−a × ε) / (1−a × ε × z ^−1). ).

  Next, the operation of the above embodiment will be described.

  First, the output signal Y [n] is delayed by one frame by the delay unit 5 to calculate the delay signal Y [n−1], and the input signal input to the input unit 2 from the delay signal Y [n−1]. X [n] is subtracted by the subtracting unit 4 and output to the nonlinear processing unit 6 as a differential signal D [n]. That is, the difference signal D [n] is calculated by D [n] = Y [n−1] −X [n].

  Next, the non-linear processing unit 6 performs non-linear processing on the differential signal D [n], for example, multiplies the differential signal D [n] by the ε function and outputs the result to the multiplying unit 7 as a non-linear signal S [n]. That is, the nonlinear signal S [n] is calculated by S [n] = ε × D [n].

  Further, the multiplication unit 7 multiplies the nonlinear signal S [n] by the cyclic coefficient a and outputs the multiplication signal A [n] to the addition unit 8. That is, the multiplication signal A [n] is calculated by A [n] = a × S [n].

  Then, the adder 8 adds the multiplication signal A [n] and the input signal X [n] and outputs the result as an output signal Y [n] from the output unit 3. That is, the output signal Y [n] is calculated by Y [n] = A [n] + X [n].

  As described above, in the above embodiment, the differential signal D [n] between the delayed signal Y [n−1] delayed by the delay unit 5 by one frame and the input signal X [n] is used as the differential signal. The nonlinear processing unit 6 performs nonlinear processing according to the amplitude of D [n], the multiplication unit 7 multiplies the cyclic coefficient a, and the addition unit 8 adds to the input signal X [n] to obtain the output signal Y [n]. .

  As a result, by setting the cyclic coefficient a and the non-linear processing according to the desired noise amplitude in the time axis direction to be reduced from the input signal X [n], the noise of the input signal X [n] is changed to the difference signal D [ n] can be detected accurately according to the amplitude of n], so that the dynamic resolution of the signal component of the input signal X [n], that is, the noise in the time axis direction from the input signal X [n] is accurately suppressed while suppressing a decrease in the f characteristic. And can be easily removed.

  In addition, by multiplying the differential signal D [n] by the ε function as the nonlinear processing in the nonlinear processing unit 6, the integral filter circuit 1 has an f characteristic determined by the cyclic coefficient a up to the input threshold of the ε function. However, it does not act on a signal having an amplitude exceeding the input threshold value of the ε function. As a result, the f characteristic of the high frequency component can be adjusted with respect to the input signal amplitude by the cyclic coefficient a and the input / output threshold value of the ε function, and the amplitude of noise to be removed from the input signal X [n] can be adjusted according to the ε function. Since it can be set arbitrarily, noise can be removed from the input signal X [n] more accurately and easily while suppressing the f-characteristic drop of the signal component of the input signal X [n].

  In particular, when the input signal X [n] is a video signal, the f characteristic can be arbitrarily adjusted by separating the amplitude change of the input signal X [n] in the time axis direction according to the set value of the ε function.

  The integration filter circuit 1 of the above embodiment is combined with, for example, a skin color detection circuit, and the cyclic coefficient a of the area where the skin color is detected is set to 1, so that the dynamic resolution with respect to the human face in the video signal can be obtained. It can be improved.

  In the above embodiment, the nonlinear processing unit 6 can also perform nonlinear processing using a nonlinear function other than the ε function.

  Also, the ε function can be used with various other threshold values.

  Furthermore, in the above embodiment, the integration filter circuit 1 is a digital circuit. However, the time constant until the horizontal scanning time is adjusted by using a CCD delay line or the like, so that it is within the delay time range of the delay element. It is also possible to construct an analog circuit that performs the same processing.

It is a block diagram which shows one Embodiment of the signal correction apparatus of this invention. It is a graph which shows an example of the input-output characteristic of an epsilon function same as the above. It is a graph which shows the characteristic of a signal correction apparatus same as the above.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Integral filter circuit as signal correction device 4 Subtraction unit 5 Delay unit 6 Non-linear processing unit 7 Multiplication unit 8 Addition unit

Claims (4)

A signal correction device that corrects an input signal having a signal component and noise and outputs it as an output signal,
A delay unit that outputs a delayed signal obtained by delaying the output signal for a predetermined time;
A subtractor for calculating a difference signal between the delayed signal and the input signal output from the delay unit;
A non-linear processing unit that non-linearly processes the difference signal calculated by the subtracting unit according to an amplitude and outputs the non-linear signal;
A multiplication unit that multiplies the nonlinear signal output from the nonlinear processing unit by a cyclic coefficient and outputs the multiplication signal;
A signal correction apparatus comprising: an addition unit that adds the multiplication signal output from the multiplication unit to the input signal to generate an output signal. The signal correction apparatus according to claim 1, wherein the nonlinear processing unit multiplies the difference signal between the delay signal output from the delay unit and the input signal by an ε function and outputs the result as a nonlinear signal. A signal correction method for correcting an input signal having a signal component and noise and outputting it as an output signal,
Delay the output signal for a predetermined time,
Non-linear processing of the differential signal between the delayed output signal and the input signal,
A signal correction method characterized by multiplying the non-linearly processed difference signal by a cyclic coefficient and adding it to the input signal to produce an output signal. The signal correction method according to claim 3, wherein when the difference signal between the delayed output signal and the input signal is subjected to nonlinear processing, an ε function is multiplied.

Images