JP2000293680A - Signal processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To emphasize the spectrum of an original signal without depending on the existence/absence of the temporal fluctuation of a frequency or blurring the spectrum by generating the grayscale display picture of spectrum intensity at each time and in every frequency, generating the plurality of grayscale display pictures by changing a frequency resolution and synthesizing them. SOLUTION: A sound signal received from a sound source 22 is subjected to preprocessing such as filter processing and AD conversion (S201) and as a frequency analysis, a signal waveform that has undergone the preprocessing is divided into short time sections. Then, a fast Fourier transform is executed in each section, and an output obtained as the result of an analysis is a time series spectrum and can be shown as a grayscale display picture 23 (S202). In signal enhancement processing to the result of a frequency analysis, grayscale display pictures whose frequency resolutions are different are inputted, and their sizes are made equal by expanding or reducing these grayscale display pictures. Subsequently, synthetic parts on the pictures are cut out, the cut images are synthesized, and the spectrum of an original signal is emphasized (S203).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention belongs to the field of signal processing.

[0002]

2. Description of the Related Art When performing frequency analysis on a signal containing noise, a method of removing noise from the signal before performing a fast Fourier transform or reducing the noise spectrum after the conversion in order to reduce the influence of the noise. Reducing methods have been used.

[0003] As a processing method before the fast Fourier transform,
A method of calculating the average of signals by repeatedly observing the same signal, and a method of setting a signal value at a certain time to an average value of signal values at times before and after the same as described in Japanese Patent Application Laid-Open No. 8-86819, etc. is there.

Further, as a processing method after the fast Fourier transform, a method of averaging the results of the fast Fourier transform executed for each short time interval, or a method of reducing noise as described in JP-A-10-240294. There is a method using a filter designed based on the estimated value of the amplitude spectrum.

[0005]

In the above-mentioned prior art in which noise is processed before fast Fourier transform, signal components other than noise (hereinafter referred to as "original signals") are attenuated at the same time as noise components are removed. There is a problem that is blurred.

Further, in the above-described conventional technology in which noise is processed after fast Fourier transform, it is difficult to reduce noise when the frequency of a signal fluctuates with time, and it is not easy to design a high-performance filter. There is a problem.

SUMMARY OF THE INVENTION It is an object of the present invention to emphasize the spectrum of an original signal without blurring the spectrum irrespective of the presence or absence of a temporal change in frequency.

[0008]

In order to solve the above-mentioned problems, in the present invention, a gray scale display image of a spectral intensity at each time and each frequency (hereinafter, a gray scale display image) is generated. A plurality of such gray-scale display images are generated by changing the frequency resolution.
By combining, the original signal can be emphasized without blurring, regardless of the presence or absence of a temporal change in frequency. Here, “combining” means replacing the density of the pixels of the new image with the sum or product of the density of the corresponding pixels on the image to be synthesized. Further, as another synthesis method, there is a method of replacing a product of a certain frequency component and the overtone component with a spectrum intensity at that frequency, that is, a pixel value, in order to emphasize a signal having an overtone component. Hereinafter, the reason why the above-described problem can be solved by combining the grayscale display images will be briefly described.

FIG. 1 is a conceptual diagram of the synthesizing process. As shown in FIG. 1, in a plurality of grayscale display images (11, 14) obtained for different frequency resolutions, the spectra (12, 15, etc.) of the original signal strongly appear at almost equal positions. On the other hand, the appearance position and intensity of noise (13, 16, etc.) vary from image to image. Therefore, as a result of the synthesis processing that emphasizes the overlapping parts of the parts where the spectrum appears strongly,
The spectrum of the original signal is emphasized compared to the noise (17, 1
8). Since it only emphasizes the overlapping part of the spectrum,
The spectrum does not blur.

In addition, in the synthesizing process, in order to calculate the spectrum intensity at each frequency at a certain time, only the spectrum intensity at the corresponding time and frequency of the original grayscale display image is used. That is, in the synthesis processing, unlike the method of taking the time average of the spectrum, the calculation in the time axis direction of the spectrum is not required at all, and even if the frequency changes over time, the processing result is not affected. Therefore, the original signal can be emphasized regardless of whether or not there is a temporal change in the frequency.

[0011]

Embodiments of the present invention will be described below in detail with reference to the drawings.

[1] First Embodiment FIG. 2 shows a configuration example of a sound signal processing system (21) to which the present invention is applied. In the figure, there is only one sound source (22),
There may be multiple sound sources. In a signal processing system,
First, preprocessing such as filter processing and AD conversion is performed on a sound signal received from a sound source (201), and frequency analysis is performed (202). Next, signal enhancement processing is performed on the result of the frequency analysis to enhance the spectrum of the original signal (20
3). As a display format of the result, a signal waveform display from which noise has been removed or a grayscale display image is used (23). For example,
In the gray-scale display image 23 of FIG. 2, the strong part of the spectrum is expressed in black, and the weak part is expressed in white. Hereinafter, specific examples of the frequency analysis unit (202) and the signal enhancement unit (203) will be described.

First, the frequency analysis unit (202) will be described with reference to the block diagram of FIG. FIG. 3 is a specific example of the frequency analysis unit. The frequency analysis unit divides the signal waveform (31) after the pre-processing (201) into short time sections (30
1). Then, fast Fourier transform is performed for each section (3
02), all sections are processed (303). The output obtained as a result of the analysis is a time-series spectrum and can be represented as a grayscale display image (32).

Next, the signal emphasizing section (203) will be described with reference to the block diagrams of FIGS. FIG. 4 is a specific example of the signal emphasizing unit, and FIG. 5 is an explanatory diagram of the image synthesizing process (403) in FIG.

In the signal emphasizing section, gray scale display images having different frequency resolutions are input (41, 42). Although FIGS. 4 and 5 show only two grayscale display images having different resolutions, a plurality of grayscale display images can be used. Next,
These gray-scale display images are enlarged or reduced to adjust the size (401, 51, 52). The ratio of the number of samples used for the fast Fourier transform is used as the scaling ratio. In the example of FIG. 5, 41 is expanded in the time axis direction (resampling, the same applies hereinafter), and 42 is expanded in the frequency axis direction. However, 41 is reduced in the frequency axis direction and 42 is reduced in the time axis direction. The size of the grayscale display image may be adjusted by a combination of reduction or enlargement or reduction of either 41 or 42.

After adjusting the size of the image, a portion to be synthesized on the image is cut out (402). In FIG. 5, since the entire image is simply processed, the image is not cut out.
The method of extracting an image will be described in [4].

Finally, the cut images are combined (40).
3, 43). Specifically, as shown in FIG. 5, for example, a sum or a product of the pixel values of the corresponding pixels on the two images (51, 52) is calculated, and the calculated value is held as the pixel value of the corresponding pixel. , Create one image (53).

As a result of the above-described processing, the spectrum of the original signal can be emphasized without blurring the time-dependent fluctuation of the frequency for the above-described reason.

[2] Second Embodiment As a second embodiment of the present invention, a synthesizing method for enhancing a predetermined harmonic component will be described.

The processing procedure for enhancing the signal is the same as the procedure shown in FIG. The difference from [1] is that the inputs (41, 42) used and the combining process (403) are different.
As the input, one or a plurality of gray scale display images having different frequency resolutions, gray scale display images having different frequency bands, gray scale display images having different frequency resolutions and frequency bands, and the like are used. If an input that differs only in frequency resolution is used, there is no need to adjust the size (401). Less than,
An example of a combining process when one gray scale display image is used as an input, that is, when the same gray scale display image is synthesized.

FIG. 10 shows an example of a filter used in the synthesizing process to emphasize a signal component having a certain frequency and a frequency component twice as high. In the figure, 101 is a harmonic emphasis filter, 102 is a grayscale display image of a signal composed of noise (102a) and original signals (102b, 102c), and 103 is a grayscale display image after filtering.

In FIG. 10, attention is paid to a spectrum at a certain time t. At time t, the noise includes the component of the frequency fa but does not include the component of the frequency 2fa (102). Therefore, when the harmonic overtone filter of FIG. 10 is applied, the frequency fa
And the spectral intensity of the noise at 2fa is zero (1
03). On the other hand, since the original signal includes components of the frequency fb and the frequency 2fb, the spectrum of the original signal at the frequencies fb and 2fb is emphasized.

As another example of the overtone emphasizing filter, the spectral intensity at a certain frequency and its double frequency is
If both are larger than a certain threshold value, there is a filter that sets the spectrum intensity at those frequencies to a predetermined value, and otherwise sets the spectrum intensity to zero.

Generally, as for noise, the frequency at which a spectrum appears strongly at a certain frequency and twice its frequency is small at the same time, so that the noise is removed by the above-described filtering. Conversely, a signal having a certain frequency and a frequency component twice as high as that of a certain frequency is more emphasized. The above is not limited to the double frequency component, and generally the same can be said for the frequency component of n times (n is an integer). In the example of FIG. 10, only a certain frequency and a frequency component twice the frequency are used. However, a signal is emphasized by simultaneously using a plurality of frequency components such as a triple frequency component and a quadruple frequency component. Can also.

[3] Third Embodiment As a third embodiment of the present invention, a method of synthesizing a grayscale display image of signals received by different sensors will be described.

FIG. 11 shows an example of processing for enhancing an original signal in signals received by different sensors. In FIG.
The inputs 111 and 112 are grayscale display images obtained as a result of frequency analysis of signals received by different sensors.

The processing shown in FIG. 11 is almost the same as the configuration example of the signal emphasizing unit shown in FIG. The difference between the two is that a process (1101) for equalizing the frequency band of the received signal has been added before the size adjustment (401) is performed. That is, after the difference of the frequency band between the sensors is corrected by 1101, the same processing as the processing shown in FIG. 4 is performed. As a method of correcting the frequency band of a signal, there is a method of cutting out a corresponding frequency portion on a grayscale display image.

In signals received by different sensors, the original signal has a correlation, but the noise has little correlation. Therefore, the same effect as [1] can be obtained by the above processing.

[4] Fourth Embodiment As a fourth embodiment of the present invention, an image clipping process will be described.

FIG. 6 shows an example of execution of the image cutting process (402) in FIG. In FIG. 6, reference numerals 61 and 62 denote images to be combined, and a solid line frame in the image is a range selected using a mouse.

The cut-out processing section selects a cut-out range on one image (61), and automatically selects a corresponding area for another image (62). In FIG. 6, two images are combined, but two or more images can be processed in the same manner.

According to the above-described method, it is not necessary to separately select corresponding ranges on a plurality of images, so that work efficiency is improved. In addition, since only the part that needs to be processed can be extracted, the processing efficiency is improved.

[5] Fifth Embodiment As a fifth embodiment of the present invention, a description will be given of a graphical user interface (hereinafter, GUI) for freely combining and executing filters.

FIG. 7 shows an example of the configuration of the filter processing unit.
The filter processing unit is a process inside the signal emphasizing unit (203), and is performed after the frequency analysis (202) or after the synthesis process (403). In FIG. 7, reference numeral 71 denotes a library in which various filters are prepared, reference numerals 72 and 73 denote examples of a GUI for selecting a filter for combining filters, and reference numeral 74 denotes an example of a filter to be executed.

In a signal analysis system to which filter processing is added, various filters are prepared as a library (see FIG. 7).
1). The filter library includes a basic library (71a) and a composite library (71b), and filters are registered in the basic library. The composite library stores composite filters obtained by combining the filters in the basic library, and information on how to combine the filters in the composite filters. By pressing the register button (72e), a new filter or a composite filter frequently used by the user can be registered or deleted from the filter library. Then, a list of filters is generated based on these libraries, and usable filters are presented to the user of the signal analysis system (701, 72). In the illustrated example, the list 72a on the left of 72 is a list of available filters, and the list 72b on the right is a list of filters selected by the user.

The user selects a filter to be used from the presented list of filters. At this time, the user
The two operations of adding (702, 703) and deleting (704, 705) the filter are repeated. FIG. 8 shows an example of an addition operation and a deletion operation.

The adding operation is an operation of pressing a mouse button on a desired filter in the list on the left side and adding a filter to the list on the right side (FIG. 8A). Reference numeral 81 denotes an example in which filter 3 and filter 5 are sequentially selected, and then filter 2 is selected with a mouse (thick arrow). Reference numeral 82 denotes a result obtained by adding a filter.

The delete operation is performed by pressing the mouse button on an unused filter in the list of selected filters on the right side,
This is an operation to delete the filter from the list on the right side (FIG. 8).
(b)). 83 is an example in which the filter 5 is deleted from the selected filter as in 82, and 84 is the result of deleting the filter.

After a filter to be used is determined by repeatedly adding and deleting a filter, filter parameters can be set as required (72d, 706). Then, when the execute button (72c) is pressed, the selected filters are executed in the order of selection (707). For example, if you finally select a filter like 73, filter 3,
Filter processing is performed in the order of filter 2 (74).

By using the above GUI, the user can freely combine any number of desired filters, so that a complex filter that realizes complicated processing can be easily configured. Further, since the operation is simple, it is easy to redo the filter processing, and the work efficiency is improved.

FIG. 9 shows a GUI for combining filters.
Here is another example. In FIG. 9, reference numeral 91 denotes a graphic representing an available filter, which corresponds to the list of available filters in FIG. 8, and 92 and 93 are screens for determining a processing flow. Corresponding to the list.

In the GUI shown in FIG. 9, a filter represented as a graphic is selected with a mouse and arranged on
a), determine the filter to be used. 92a is an example of selecting the filter 2. In addition, by pressing the mouse button in the order of processing on the filter of continuous processing (92b),
Determine the processing order of each filter. 92b shows an example in which the mouse button is first pressed on the filter 2 and then the mouse button is pressed on the filter 3;
Processing is performed in the order of the filter 3 (93a).

By using the above GUI, G in FIG.
The same effect as in the case of using the UI can be obtained. In addition, by setting a branch point such as 93b, a plurality of processing procedures having the same halfway processing can be efficiently determined.

[0044]

According to the present invention, as described in the section of the means for solving the problems, the spectrum of the original signal is emphasized without blurring the spectrum irrespective of the presence or absence of the temporal fluctuation of the frequency. Is obtained.

[Brief description of the drawings]

FIG. 1 is a conceptual diagram of a composition process of a grayscale display image.

FIG. 2 is a basic configuration diagram of a sound signal processing system.

FIG. 3 is a detailed explanatory diagram of a frequency analysis process of FIG. 2;

FIG. 4 is a detailed explanatory diagram of the signal emphasizing process of FIG. 2;

FIG. 5 is a synthesis example of a grayscale display image.

FIG. 6 is an explanatory diagram of a method of partially selecting an image when combining images.

FIG. 7 is a configuration example of a filter processing unit.

FIG. 8 is an explanatory diagram of a filter adding operation and a deleting operation using a GUI in a filter processing unit.

FIG. 9 illustrates a GUI different from that of FIG. 8 in a filter processing unit.
Example.

FIG. 10 is an example of a filter that emphasizes harmonic components.

FIG. 11 is an example of processing for enhancing an original signal in signals received by different sensors.

[Explanation of symbols]

 41: Results of fast Fourier transform with different frequency resolutions 51: Image of 41 expanded in the time axis and frequency axis 72: Example of GUI used in filter processing unit 101: Example of filter to emphasize harmonic components .

Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat II (Reference) G10L 15/02 (72) Inventor Nobuo Makimoto 1099 Ozenji Temple, Aso-ku, Aso-ku, Kawasaki City, Kanagawa Prefecture Inside of Hitachi Systems Development Laboratory ( 72) Inventor Junichi Taguchi 1099 Ozenji Temple, Aso-ku, Kawasaki City, Kanagawa Prefecture F-Term in Hitachi, Ltd.System Development Laboratory (Reference) 5B057 BA04 CD05 CE08 CG05 5D015 CC03 EE05 LL05 5D045 AB30 9A001 BB06 DD13 GG03 HH16 HH18 HH24H25

Claims (5)

[Claims] In a signal analyzing apparatus comprising a pre-processing unit, a frequency analyzing unit, a signal emphasizing unit, and a display unit, the pre-processing unit performs pre-processing such as signal conversion in order to analyze a received signal. The frequency analysis unit generates a plurality of gray scale display images of the spectrum intensity at each time and each frequency while changing the frequency resolution at the time of the frequency analysis. The signal emphasis unit generates the same gray scale display image at the same time and the same frequency for each gray scale display image. A new image is generated by combining the values at, for example, geometric averaging.
A signal processing device for enhancing the spectrum of an original signal by displaying the image. 2. The method according to claim 1, wherein the input is the input
, But the frequency band may be different, and the signal emphasizing unit synthesizes the values at the same time and at a frequency having a predetermined harmonic relation in the same manner as in claim 1, A signal processing device for enhancing a spectrum of an original signal including harmonics. 3. A method according to claim 1, wherein a grayscale display image of a signal received by a different sensor is used as an input. A signal processing device characterized in that the spectrum of an original signal is emphasized by compensating the signals by resampling or the like and then combining the corrected signals. 4. A signal emphasizing unit according to claim 1, wherein, when selecting and synthesizing a part or the whole of the image, a part or the whole of the image is selected using an input device on an arbitrary one of the images. Then, a corresponding portion of another image is selected at the same time, and the signal processing device has a function of synthesizing only the selected portion. 5. A filter processing unit for storing various filters in a filter library and emphasizing the spectrum of the original signal by using them in combination inside the signal emphasizing unit according to claim 1 to 3. A signal processing system, characterized in that the signal processing system has a function to easily realize more complicated processing by selecting and combining desired filters from a filter library. apparatus.