JPH1063839A - Image processing method and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To remove a granular noise signal included in a non-edge part while operating a desired emphasizing processing at the time of preparing an addition signal to be added to an original image signal for emphasis by integrating a band pass signal indicating a signal in each frequency band of the original image signal. SOLUTION: An edge image signal is prepared by an edge detecting means 2 and an interpolating and enlarging means 3, an edge coefficient which is almost 1 for the signal of an edge part, and which is almost 0 for the signal of a non-edge part is defined by a converting means 5, and this edge coefficient is multiplied by the signal by a frequency emphasis processing means 6. Thus, a noise signal included in the non-edge part can be removed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method and apparatus for enhancing a predetermined frequency component of an image signal, and more particularly to the removal of noise contained in an image signal when performing an enhancement process.

[0002]

2. Description of the Related Art Heretofore, a number of image processing methods and apparatuses have been proposed by the present applicant to perform frequency emphasis processing using an unsharp mask image signal (hereinafter referred to as a blurred image signal) to improve diagnostic performance of a radiation image. It has been proposed (Japanese
-163472 and 55-87953). Here, the blurred image signal is an image signal representing an image having the same number of pixels as the original image signal, but having a lower sharpness than the original image signal, and removing a high-frequency component of a predetermined frequency or more of the original image signal. This is a signal having the frequency response characteristic obtained.

[0003] The above-mentioned frequency emphasis processing is performed by using the original image signal Sorg.
Then, a value obtained by subtracting the blurred image signal Sus from the result and multiplying the result by an enhancement coefficient β is added to the original image signal Sorg to emphasize a predetermined spatial frequency component of the original image signal. This is expressed by the following equation (1).

Sproc = Sorg + β × (Sorg−Sus) (1) (Sproc: signal subjected to frequency emphasis processing, Sorg: original image signal, Sus: blurred image signal, β: emphasis coefficient) Can be solved by adjusting the frequency response characteristic of the addition signal to be added to the original image signal Sorg. As a specific adjustment method for this, The following method has been proposed. (Japanese Patent Application
8-182155).

In this method, first, a plurality of blurred image signals having different sharpness, that is, different frequency response characteristics are created, and a difference between the blurred image signal and two signals in the original image signal is obtained. A plurality of band-limited image signals (hereinafter, referred to as band-pass signals) representing frequency components in a limited frequency band of the original image signal are created, and the band-pass signals each have a desired size by a predetermined function. After the suppression, the added signal is created by integrating the plurality of suppressed bandpass signals. If this processing is expressed as an equation, for example, the following equation (2) Sproc = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2,... SusN) Fusm (Sorg, Sus1, Sus2,... SusN) = f ₁ (Sorg) −Sus1) + f ₂ (Sus1-Sus2) +... + F _k (Susk−1−Susk) +... + F _N (SusN−1−SusN) (2) (where, Sproc: an image signal Sorg in which high-frequency components are emphasized) : Original image signal Susk (k = 1 to N): Blurred image signal f _k (k = 1 to N): Function for converting each bandpass signal β (Sorg): Enhancement coefficient determined based on the original image signal) become that way.

In such processing, functions f _{1 to} f
_N is a function that suppresses the bandpass signal in accordance with the magnitude of the bandpass signal.For example, in order to prevent an artifact, a large bandpass signal that may cause an artifact is strongly suppressed. Conversion processing is performed such that a signal that is not large is not suppressed.

[0007]

However, when the conversion of the band-pass signal is performed based only on the magnitude of the band-pass signal as in the above-described processing, the noise component contained in the image signal should also be originally processed. It is processed in the same way as a signal. This is undesirable because a noise component is included in the added signal added for enhancement, that is, the noise component is enhanced by the enhancement processing.

The present invention has been made in view of the above problems, and provides an image processing method and apparatus capable of removing a noise signal included in an image signal when enhancing a predetermined frequency component of the image signal. It is intended for.

[0009]

According to an image processing method of the present invention, a plurality of low-resolution image signals having different resolutions are generated based on an original image signal representing an original image, and the plurality of low-resolution image signals are generated for the low-resolution image signals. A plurality of low-resolution edge image signals having different resolutions corresponding to each of the low-resolution image signals by performing a predetermined edge detection process;
Each of the low-resolution image signals is interpolated and expanded so that the number of pixels of the low-resolution image signal is the same as the number of pixels of the original image signal, thereby generating a plurality of different blurred image signals of the original image signal. And interpolating and expanding each of the low-resolution edge image signals so that the number of pixels of the low-resolution edge image signal is the same as the number of pixels of the original image signal. Generating a plurality of band-pass signals representing signals for each of a plurality of frequency bands of the original image signal based on the original image signal and the plurality of blurred image signals. A suppressed image signal is created by converting so as to suppress the absolute value of the path signal, and the edge image signal is substantially 0 when the absolute value of the edge image signal is small,
When the absolute value is large, an edge coefficient is created by conversion using a non-linear function that is converted to be substantially 1 and a converted image signal for each frequency band is created by multiplying the suppressed image signal by the edge coefficient. A predetermined frequency component of the original image signal is emphasized by adding an integrated signal obtained by integrating the converted image signals to the original image signal.

[0010] Here, the "low-resolution image signal" is an image signal obtained by subjecting pixels of the original image signal to predetermined filtering processing at predetermined intervals to thin out the pixels. For the filtering process, various methods that are generally widely used can be used.

As the "predetermined edge detection process", for example, a method of detecting an edge using a difference between a maximum value and a minimum value, a variance value, a density gradient, or the like in a mask of a predetermined size is applied. it can. This may be any processing as long as it is an edge detection processing widely and generally performed, and is not particularly limited.

[0012] The "interpolation enlargement" processing applied to the low-resolution image signal and the low-resolution edge image signal is performed by interpolating pixels by a predetermined interpolation method so as to have the same number of pixels as the original image. Means to do. Regarding this interpolation method, including the B-spline interpolation method,
Various methods generally used widely can be applied. The interpolation enlargement of the low resolution image signal and the interpolation enlargement of the low resolution edge image signal may be performed by the same method, or may be performed by different methods.

The "plurality of band-pass signals representing signals of the original image signal for each of a plurality of frequency bands" may be created by taking a difference between blurred image signals of adjacent frequency bands, for example. It may be created by taking the difference between the image signal and each blurred image signal. Alternatively, the difference may be obtained by another combination of the original image signal and the blurred image signal. Further, "emphasizing a predetermined frequency component" means, for example, emphasizing a high-frequency component in order to emphasize an edge portion of an image.

The creation of the bandpass signal, the creation of the converted image signal, the creation of the integrated signal, and the addition of the integrated signal to the original image signal are specifically performed by the following equation: Sproc = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2, ... SusN) Fusm (Sorg, Sus1, Sus2, ... SusN) = f 1 (Sorg -Sus1) × g (Sedge1) + f 2 (Sus1 -Sus2) × g (Sedge2 ) +... + F _k (Susk-1−Susk) × g (Sedgek) +... + F _N (SusN−1−SusN) × g (SedgeN) (where, Sproc: an image signal in which a predetermined frequency component is emphasized Sorg: Original image signal Susk (k = 1 to N): Blurred image signal Sedgek (k = 1 to N): Edge image signal f _k (k = 1 to N): Convert each bandpass signal to create a suppressed image signal G: a function for converting each edge image signal to generate an edge coefficient β (Sorg): determined based on the original image signal It is desirable to perform in accordance with tone coefficient). Further, it is desirable that the nonlinear function used for generating the edge coefficient is different depending on the result of the histogram analysis of the edge image signal.

Further, an image processing apparatus according to the present invention is an apparatus for performing image processing according to the above-described image processing method, wherein a plurality of low-resolution image signals having different resolutions are obtained based on an original image signal representing an original image. A low-resolution image signal generating means for generating, and a plurality of low-resolution edge image signals having different resolutions corresponding to each of the low-resolution image signals by performing a predetermined edge detection process on each of the low-resolution image signals. A low-resolution edge signal generating means for generating the low-resolution image signal, and interpolating and expanding each of the low-resolution image signals so that the number of pixels of the low-resolution image signal is the same as the number of pixels of the original image signal. A blurred image signal generating means for generating a plurality of different blurred image signals of the signal; Each way is the same as the number of pixel signals by interpolating enlargement, the edge image signal producing means for producing a plurality of different edge image signal of the original image signal,
Based on the original image signal and the plurality of blurred image signals, a plurality of bandpass signals representing signals of the original image signal for each of a plurality of frequency bands are created, and each of the bandpass signals is defined as an absolute value of the bandpass signal. And a suppressed image signal creating means for creating a suppressed image signal by converting so as to suppress the edge image signal. When the absolute value of the edge image signal is small, the value is substantially 0, and when the absolute value is large, the value is substantially 1. Edge coefficient creating means for creating an edge coefficient by performing a conversion using a non-linear function, and a converted image signal for each frequency band by multiplying the suppressed image signal by the edge coefficient. Frequency emphasis processing means for emphasizing a predetermined frequency component of the original image signal by adding an integrated signal obtained by accumulating the image signal to the original image signal. It is characterized in.

As described above, the interpolation / enlargement processing of the low-resolution image signal and the low-resolution edge image signal may be performed by the same interpolation / enlargement method. May be the same.

[0017]

According to the image processing method and apparatus of the present invention, a band-pass signal representing a signal in each frequency band of an original image signal is integrated, and an addition signal to be added to the original image signal for enhancement is created. At this time, edge information of the image signal is extracted, and based on this edge information, an edge coefficient is defined that is substantially 1 for an edge signal and substantially 0 for a non-edge signal. , It is possible to remove a granular noise signal included in a non-edge portion while performing a desired enhancement process, and to obtain a high-quality processed image signal.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of an image processing method and apparatus according to the present invention will be described in detail with reference to the drawings. The image processing apparatus described below uses a blurred image signal for an image signal obtained by reading a radiation image of a human body recorded on a stimulable phosphor sheet so that the image becomes an image suitable for diagnosis. Then, the processed image signal is mainly recorded on a film and used for diagnosis.

FIG. 1 is a diagram schematically showing the image processing apparatus. The image processing apparatus performs a filtering process on the original image signal Sorg to perform low-resolution image signal B _k (k = 1 to n).
, An edge detecting means 2 for detecting edges of the low-resolution image signal B _k obtained by the filtering process and generating a low-resolution edge image signal E _k (k = 1 to n), A predetermined interpolation enlargement process is performed on the resolution image signal B _k and the low resolution edge image signal E _k , and the blurred image signal Susk is _obtained from the low resolution image signal B _k.
And also the low-resolution edge image signal and an interpolation enlarging means 3 for creating an edge image signal Sedge respectively from E _k, the original image signal and the unsharp image signals Susk create a band-pass signals for each frequency band from the band-pass signal Is converted by a predetermined function to generate a suppressed image signal, a second converting unit 5 that generates an edge coefficient from the edge image signal Sedge, and a predetermined conversion is performed by using the edge coefficient and the suppressed image signal. Frequency emphasis processing means 6 for emphasizing a specific frequency component by creating a signal and adding the signal to the original image signal.

Here, the filtering means 1 corresponds to the low-resolution image signal generating means, and the edge detecting means 2 corresponds to the low-resolution edge image signal generating means.
The interpolation enlargement unit 3 corresponds to a unit having both functions of the blurred image signal creation unit and the edge image signal creation unit. Further, the first conversion means 4 is provided to the suppressed image signal creation means,
The second conversion means 5 corresponds to the edge coefficient creation means.

The original image signal Sorg, the low resolution image signal B _k , the low resolution edge image signal E _k , and the blurred image signal Sus
The five types of signals, k and the edge image signal Sedge, have a relationship as shown in FIG. That is, the low-resolution image signal B _k is a signal whose resolution has been reduced by reducing the number of pixels of the original image signal Sorg, and the blurred image signal Sus
k is an image signal having the same number of pixels as the original image signal obtained by interpolating and expanding the low-resolution image signal B _k . The low-resolution edge image signal E _k is a signal _obtained by extracting only the edge portion from the low-resolution image signal B _k , and the edge image signal Sedge is such that the low-resolution edge image signal E _k has the same number of pixels as the original image signal. This is a signal obtained by interpolation and expansion.

Next, a detailed description will be given of the processing for creating each of the above signals. FIG. 3 is a block diagram showing an outline of the blurred image signal creation processing. As shown in FIG. 3, the blurred image signal is generated by first performing filtering processing on the original image signal Sorg in the x direction and the y direction of the pixels of the original image by the filtering unit 1 to obtain the blurred image signal. also creates a low-resolution low-resolution image signal B _1, then the same is subjected to filtering processing the low-resolution image signal B ₁ low-resolution image signal further has lower resolution than for the low resolution image signal B ₁ create a B _2, those to superimpose successive similar filtering process later. Then, the low-resolution image signal B _k obtained at each stage of the filtering process is
A plurality of blurred image signals Sus1 to SusN having different sharpness levels are obtained by performing interpolation enlargement processing, respectively.

In the present embodiment, a filter substantially corresponding to a one-dimensional Gaussian distribution is used as a filter for the filtering process. That is, the filter coefficient of the filter is calculated by the following equation (3) for a Gaussian signal.

[0024]

(Equation 1)

Is determined according to the following. This is because the Gaussian signal has good localization in both the frequency space and the real space. For example, a 5 × 1 one-dimensional filter when σ = 1 in the above equation (3) is shown in FIG. It will be something like

The filtering process is performed on the original image signal Sorg or every other pixel on the low-resolution image signal as shown in FIG. By performing such filtering processing for every other pixel in the x direction and the y direction, the number of pixels of the low-resolution image signal B ₁ becomes １ ／ of the original image, and the low-resolution image signal obtained by the filtering processing is reduced. By repeatedly performing this filtering process, each of the n low-resolution image signals B _{k obtained} is an image signal whose number of pixels is 2 ^{2 k} of the original image signal.

A description will now be given of the interpolation enlargement process performed on the low-resolution image signal B _k obtained in this manner. As the method of the interpolation operation, various methods such as a method using a B-spline can be mentioned. In the present embodiment, since the low-pass filter based on the Gaussian signal is used in the filtering process, the Gaussian signal is also used for the interpolation operation. Shall be used. Specifically, the following equation (4)

[0028]

(Equation 2)

In the above, an approximation of σ = 2 ^k−1 is used.

When interpolating the image signal B ₁ , σ = 1 because k = 1. In this case, the filter for performing the interpolation processing is a 5 × 1 one-dimensional filter as shown in FIG. In this interpolation processing, first, the low-resolution image signal B ₁
, The low-resolution image signal B ₁ is enlarged to the same size as the original image by interpolating one pixel having a value of 0 every other pixel, and then the interpolated low-resolution image signal B
_This is performed by subjecting 1 to a filtering process using the above-described one-dimensional filter shown in FIG.

Similarly, this interpolation enlargement processing is performed on all the low-resolution image signals _Bk . When interpolating the low-resolution image signal B _k , 3 × 2 ^k −1 based on the above equation (4).
By creating a filter of length and interpolating 2 ^k -1 pixels having a value of 0 between each pixel of the image signal B _k ,
Expanding the original image and the same size, the length of 3 × 2 ^k -1 filtering the image signals B _k having pixels interpolated in this value is 0, the interpolation based expansion by performing a filtering process.

Further, for each low resolution image signals B _k obtained by the filtering processing, edge detection means 2
Is also performed. In the present embodiment,
If the difference between the maximum value and the minimum value among the values of the pixels falling within a mask of a predetermined size is larger than a predetermined threshold, the edge of the image signal is detected by judging that the edge is an edge. the detected edge information is set to the low-resolution edge image signal E _k. However, the method of edge detection is not limited to this. For example, instead of the difference between the maximum value and the minimum value, detection may be performed using a variance value or a density gradient.

The low-resolution edge image signal E _k obtained as described above is also interpolated and expanded similarly to the low-resolution image signal B _k . The method of interpolation and enlargement may be the same as or different from the method of interpolation and enlargement of a low-resolution image signal. However, since the same method is used in the present embodiment, the description is omitted here. .

Each image signal obtained as described above is
The following equation (5) Sproc = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2, ... SusN) Fusm (Sorg, Sus1, Sus2, ... SusN) = f 1 (Sorg -Sus1) × g (Sedge1) + f ₂ (Sus1−Sus2) × g (Sedge2) +... + F _k (Susk−1−Susk) × g (Sedgek) +... + F _N (SusN−1−SusN) × g (SedgeN) (5) (However, Sproc: an image signal in which a predetermined frequency component is emphasized Sorg: an original image signal Susk (k = 1 to N): an unsharp mask image signal Sedgek (k = 1 to N): an edge image signal f _k (k = 1 to N): Function for converting each bandpass signal to generate a suppressed image signal g: Function for converting each edge image signal to generate an edge coefficient β (Sorg): Determined based on the original image signal Is performed according to the following enhancement coefficient.

Here, the functions f _{1 to} f _N use, for example, functions as shown in FIG. f _{1 to} f _N may all be the same function, or may be different functions for each bandpass signal.

As the function g, for example, a function as shown in FIG. 8 is used. This is to determine the edge coefficient so that it is 1 when the edge signal is large and 0 when it is small. This is because the portion where the edge signal is large is a true edge, but the portion where the edge signal is small is not a true edge but a granular noise component included in a flat portion is detected as an edge. From
For the part containing such granular noise,
In order to reduce or eliminate noise by suppressing the signal, the edge coefficient corresponding to the portion is set to 0 or a value close to 0, and the band coefficient obtained by converting such a small value of the edge coefficient is used. By multiplying the signal, the signal of that portion is prevented from being added as a signal for emphasis.

It is to be noted that a histogram analysis of the edge image signal may be separately performed, and the function g may be changed according to the amount of noise in the image. That is, as shown in FIG. 9, when the histogram is spread (b),
Since the histogram is more noisy than when the histogram is localized near 0 (a), the edge image signal is relatively large for the noisy image (b) as shown in the figure. If the edge coefficient is set to 0, noise can be more appropriately removed.

According to the present invention, as described above, a noise signal which has been conventionally emphasized together with an edge image signal is separated from the edge image signal so that the noise signal is not emphasized. The desired enhancement processing can be performed on only the signal to obtain a high-quality processed image.

[Brief description of the drawings]

FIG. 1 is a diagram schematically showing an image processing apparatus according to the present invention.

FIG. 2 is a diagram showing the relationship between various signals obtained in the course of image processing according to the present invention.

FIG. 3 is a block diagram showing an outline of a blurred image signal creation process.

FIG. 4 is a diagram showing an example of a filter used for a filtering process.

FIG. 5 is a diagram showing details of a low-resolution image signal creation process;

FIG. 6 is a diagram illustrating an example of a filter used for interpolation enlargement processing;

FIG. 7 is a diagram illustrating an example of a function for suppressing a bandpass signal.

FIG. 8 is a diagram illustrating an example of a nonlinear function for determining an edge coefficient.

FIG. 9 is a diagram showing a relationship between a histogram analysis result of an edge image signal and a nonlinear function for determining an edge coefficient.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Filtering means 2 Edge detection means 3 Interpolation enlargement means 4 Conversion means 5 Conversion means 6 Frequency emphasis processing means

Claims (6)

[Claims] 1. A plurality of low-resolution image signals having different resolutions are created based on an original image signal representing an original image, and a predetermined edge detection process is performed on each of the low-resolution image signals. A plurality of low-resolution edge image signals having different resolutions corresponding to each low-resolution image signal are created, and the number of pixels of each of the low-resolution image signals is the same as the number of pixels of the original image signal. Respectively, and a plurality of different unsharp mask image signals of the original image signal are created, and each of the low-resolution edge image signals is converted into the original image signal by the number of pixels of the low-resolution edge image signal. A plurality of different edge image signals of the original image signal are created by interpolation and enlargement so as to have the same number of pixels of the signal, and the original image signal and the plurality of unsharp masks Creating a plurality of band-limited image signals representing signals for each of a plurality of frequency bands of the original image signal based on the image signal, and converting each of the band-limited image signals so as to suppress the absolute value of the band-limited image signal. Then, a suppressed image signal is created. The edge image signal is converted by a non-linear function that converts the edge image signal to be substantially 0 when the absolute value of the edge image signal is small and to be substantially 1 when the absolute value of the edge image signal is large. A coefficient is created, a converted image signal for each frequency band is created by multiplying the suppressed image signal by the edge coefficient, and an integrated signal obtained by integrating the converted image signals is added to the original image signal. An image processing method for emphasizing a predetermined frequency component of the original image signal. 2. Creating the band-limited image signal, creating the converted image signal, creating the integrated signal, and adding the integrated signal to the original image signal is represented by the following equation: Sproc = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2 , ... SusN) Fusm (Sorg, Sus1, Sus2, ... SusN) = f 1 (Sorg -Sus1) × g (Sedge1) + f 2 (Sus1 -Sus2) × g (Sedge2) + ... + F _k (Susk-1−Susk) × g (Sedgek) +... + F _N (SusN−1−SusN) × g (SedgeN) (where Sproc: an image signal in which a predetermined frequency component is emphasized Sorg: an original image signal Susk (k = 1 to N): Unsharp mask image signal Sedgek (k = 1 to N): Edge image signal f _k (k = 1 to N): Convert each band-limited image signal to create a suppressed image signal G: function for converting each edge image signal to create an edge coefficient β (Sorg): emphasis determined based on the original image signal 2. The image processing method according to claim 1, wherein the image processing is performed according to: 3. The image processing method according to claim 1, wherein the non-linear function used to generate the edge coefficient is different according to a result of a histogram analysis of the edge image signal. 4. A low-resolution image signal generating means for generating a plurality of low-resolution image signals having different resolutions based on an original image signal representing an original image, and a predetermined edge for each of the low-resolution image signals. Low-resolution edge image signal creating means for creating a plurality of low-resolution edge image signals having different resolutions corresponding to the low-resolution image signals by performing a detection process; Unsharp mask image signal creating means for creating a plurality of different unsharp mask image signals of the original image signal by performing interpolation interpolation so that the number of pixels of the image signal is the same as the number of pixels of the original image signal, The low-resolution edge image signals are interpolated and enlarged so that the number of pixels of the low-resolution edge image signal is the same as the number of pixels of the original image signal, and Edge image signal creating means for creating a plurality of different edge image signals of the signal; and a plurality of signals representing a signal for each of a plurality of frequency bands of the original image signal based on the original image signal and the plurality of unsharp mask image signals. A band-limited image signal is created, the band-limited image signal is converted so as to suppress the absolute value of the band-limited image signal, and a suppressed image signal creating unit that creates a suppressed image signal, and the edge image signal, Edge coefficient creating means for creating an edge coefficient by converting with a non-linear function for converting the absolute value of the edge image signal to be substantially 0 when the absolute value is small and to be substantially 1 when the absolute value is large; By multiplying the edge coefficient, a converted image signal for each frequency band is created, and an integrated signal obtained by integrating the converted image signals is added to the original image signal. The image processing apparatus characterized by comprising a emphasizing frequency enhancement processing means a predetermined frequency component of the original image signal by calculation to. 5. The process of creating the band-limited image signal, creating the converted image signal, creating the integrated signal, and adding the integrated signal to the original image signal is represented by the following equation: Sproc = Sorg + β (Sorg) × Fusm (Sorg, Sus1, Sus2 , ... SusN) Fusm (Sorg, Sus1, Sus2, ... SusN) = f 1 (Sorg -Sus1) × g (Sedge1) + f 2 (Sus1 -Sus2) × g (Sedge2) + ... + F _k (Susk-1−Susk) × g (Sedgek) +... + F _N (SusN−1−SusN) × g (SedgeN) (where Sproc: an image signal in which a predetermined frequency component is emphasized Sorg: an original image signal Susk (k = 1 to N): Unsharp mask image signal Sedgek (k = 1 to N): Edge image signal f _k (k = 1 to N): Convert each band-limited image signal to create a suppressed image signal G: function for converting each edge image signal to create an edge coefficient β (Sorg): emphasis determined based on the original image signal The image processing apparatus according to claim 4, wherein the image processing is performed in accordance with a coefficient. 6. The image processing apparatus according to claim 4, wherein the non-linear function used for generating the edge coefficient differs according to a result of a histogram analysis of the edge image signal.