JP2005065196A - Filter, signal processor, signal processing method, recording medium, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem wherein a conventional low-pass filter cuts off high-frequency signals independently of the magnitude of the signal level and to process signals selectively corresponding to their levels so as to improve reproduced picture images in picture quality in video apparatuses equipped with gamma correction circuits. <P>SOLUTION: A basic filter is made variable in characteristics corresponding to the level (amplitude)of signals. For instance, the basic filter is kept holding characteristics as well as a low-pass filter, but high-frequency signals having a low signal level are selectively allowed to pass through the basic filter. Or, for instance, the basic filter is kept holding characteristics as well as a high-pass filter, but low-frequency signals having a high signal level are selectively allowed to pass through the basic filter. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a filter. The present invention also relates to a signal processing apparatus using the filter and a signal processing method thereof. The present invention also relates to a program for causing a computer to execute the filter function. The present invention also relates to a computer-readable recording medium on which the program is recorded.

  The filter is one of the basic electronic elements, and is used in combination with other circuits depending on its use and purpose. Therefore, the application range of individual filters is very wide. Here, the background to the development of the filter according to the present invention will be described by taking as an example the case of application to the field of processing video (video) signals.

  In general, various types of signal processing are performed on video signals handled by video cameras, video tape recorders, and other video (video) devices. For example, gamma correction, knee correction, black / white clip, contour enhancement, white balance adjustment, hue adjustment, and the like are performed.

  In particular, in the case of a camera that digitizes a captured video signal and performs digital signal processing, a dedicated signal processing unit is provided to perform the above-described nonlinear processing. Such a signal processing unit performs the above-described processing based on a predetermined control parameter. The control parameters are written in a memory or other storage medium.

  By the way, it is known that when the nonlinear processing described above is performed on a video signal, the waveform of the video signal is distorted. This distortion generates a harmonic component corresponding to an integer multiple of the frequency component included in the video signal before the nonlinear processing in the video signal after the nonlinear processing.

FIG. 1 shows a configuration example of a camera that employs a digital signal processing method. The circuit configuration of this camera will be briefly described.
First, an analog video signal S1 is output from the image sensor 1. The analog video signal S 1 is amplified by the amplifier 2 and then input to the analog / digital conversion circuit 3. The analog / digital conversion circuit 3 outputs a digital video signal S2.
The digital video signal S2 is input to the contour emphasis signal generation circuit 4 and the knee correction circuit 5. The contour enhancement signal generation circuit 4 generates a contour enhancement signal (high frequency component) S3 of the digital video signal S2. The knee correction circuit 5 executes processing for compressing the high luminance signal in order to increase the dynamic range.

  The edge enhancement signal S3 and the digital video signal S4 after knee correction are added by the adder 6. The adder 6 outputs a digital video signal S5 whose contour has been corrected. The low-pass filter 7 extracts only the low frequency component S6 from the digital video signal S5 and supplies it to the gamma correction circuit 8. The gamma correction circuit 8 executes gamma correction processing so that the image is faithfully reproduced in the output device. The adder 9 adds the edge emphasis signal S3 to the digital video signal S7 after the gamma correction. As a result, a digital video signal S8 whose contour is corrected is obtained.

  As described above, the camera having the circuit configuration shown in FIG. 1 employs a method of adding the edge enhancement signal S3 before and after the nonlinear processing unit for the main line signal, specifically, before and after the gamma correction circuit 8, respectively. .

  However, when the contour emphasis signal S3 is added in the previous stage, the high frequency signal is uniformly subjected to nonlinear processing by the gamma correction circuit 8 to generate harmonics. That is, all signals exceeding the Nyquist frequency (fs / 2) that is 1/2 of the sampling frequency fs are folded back as false signals (aliasing) within the 0 to fs / 2 band.

Such aliasing of the false signal generates a low-frequency beat that was not found in the original signal. In addition, the aliasing of the false signal causes various other phenomena, and the image quality is significantly impaired.
Therefore, in order to avoid the generation of such a false signal, the edge enhancement signal S3 is added after the gamma correction in the camera having the circuit configuration shown in FIG.

  However, the outline emphasis signal S3 added after gamma correction is not subjected to gamma correction. On the other hand, when an image is actually displayed on the monitor, the inverse gamma curve of the monitor acts on the image signal. For this reason, as a result, there is a problem in that an image with an insufficient outline on the black side and an excessive outline on the white side is displayed.

Further, even when an edge enhancement signal is input before gamma correction, a method for processing an input signal for each frequency has been proposed as a technique for minimizing the above-described aliasing distortion.
This method makes use of the fact that even if the low frequency component of the digital video signal is distorted, it is difficult to generate a false signal. That is, non-linear processing is performed for low-frequency components, and processing closer to linear is performed for high-frequency components that tend to generate false signals when distorted, thereby suppressing generation of false signals.

Specifically, a low-pass filter is placed just before nonlinear processing to pass only the low-frequency component of the digital video signal, and then added to the linear processing side, which is the main line, so that a signal that suppresses false signals is added. Generate.
JP-A-9-46554

  However, even when such a method is employed, the edge enhancement signal is filtered in the nonlinear processing. For this reason, gamma correction is not applied to the edge enhancement signal of the video signal after gamma correction. As a result, as in the case of the circuit configuration of FIG. 1 described above, there is a problem that after gamma correction, the black side has insufficient outline and the white side has excessive outline.

  The present invention has been made in consideration of the above problems, and can be solved by applying the present invention to video equipment, and cannot be removed only by basic filter characteristics if applied to other electronic equipment. The purpose is to solve the problem.

(A) First Method In order to achieve this object, one of the present invention proposes a filter that varies the basic filter characteristics in accordance with the signal level (amplitude) of the input signal. That is, a filter that can selectively change the passband according to the signal level of the input signal while maintaining the basic filter characteristics is proposed. FIG. 2 shows a filter 11 having a basic filter characteristic variable function.
Here, the basic filter characteristic means a basic filter characteristic. For example, it is a low-pass filter characteristic, a high-pass filter characteristic, a band-pass filter characteristic, and a band elimination filter characteristic.

  One aspect of the present invention changes the filter characteristics according to the signal level of the input signal while maintaining the basic characteristics of the basic filter characteristics. For example, FIG. 3 shows an example in which the basic filter characteristic is a low-pass filter. In FIG. 3, the horizontal axis represents frequency and the vertical axis represents the signal level of the input signal. In addition, a hatched area represents an area that passes through the filter.

  In the case of FIG. 3, the input signal passes through the low-frequency component (frequency band lower than a certain threshold (the cutoff frequency of the basic filter)) as it is. In this way, basic filter characteristics are maintained. On the other hand, for the high frequency component (frequency band higher than a certain threshold (cutoff frequency)), the pass characteristic is varied according to the signal level. For example, in the case of FIG. 3A, if the signal level is smaller than a predetermined threshold, it indicates that the signal passes through the filter even if the frequency is high. FIG. 3B shows that if the signal level is larger than a predetermined threshold value, the signal passes through the filter even if the frequency is high.

  FIG. 4 shows an example in which the basic filter characteristic is a high-pass filter. FIG. 4A shows a case in which the high-pass filter characteristic is maintained and the low-pass component is allowed to pass through the filter if the signal level is smaller than a predetermined threshold value. FIG. 4B shows a case where the filter is allowed to pass through while maintaining the high-pass filter characteristics, even if the signal level is higher than a predetermined threshold even for low-frequency components. FIG. 5 shows a case where the basic filter characteristic is a band-pass filter, and FIG. 6 shows a case where the basic filter characteristic is a band elimination filter. The description of the figure is omitted since it is the same as the above examples. 3 to 6 illustrate frequency pass characteristics, and various control methods can be considered for the gain of the pass band as will be described later.

(B) Second Method In order to achieve the above-described object, a filter 21 having the configuration shown in FIG. 7 is proposed as one of the present invention. The filter 21 has an inverse characteristic filter stage 22 that passes an input signal with a filter characteristic opposite to the basic filter characteristic, and at least part of the frequency component according to the signal level of the frequency component that has passed through the inverse characteristic filter stage 22. An extraction stage 23 for extraction and a subtraction stage 24 for subtracting the extracted frequency component from the input signal are provided.

  The inverse characteristic filter stage 22 here corresponds to a high-pass filter, for example, if the basic filter characteristic is a low-pass filter. The basic filter characteristic here is a basic filter characteristic to be realized as a whole filter. Therefore, as the basic filter characteristics, the low-pass filter characteristics, the high-pass filter characteristics, the band-pass filter characteristics, and the band elimination filter characteristics can be considered as in the first method.

  Note that the low-pass filter and the high-pass filter have a relationship of filter characteristics opposite to each other. The relationship between the bandpass filter and the band elimination filter is the same. For this reason, the basic filter characteristics of the filter stage here are uniquely determined by any one of the above-described filter stages.

  The extraction stage 23 adaptively extracts a part of the passed frequency component in accordance with the signal level (amplitude) of the frequency component that has passed through the inverse characteristic filter stage 22. The signal extracted here can be a signal of the entire band that has passed through the inverse characteristic filter stage 22, or can be limited to a signal that satisfies a predetermined signal level condition. For example, only signals exceeding a certain signal level (threshold) can be used. Conversely, it is also possible to have only a signal that is smaller than a certain signal level (threshold). Which one is adopted depends on the characteristics required for the filter.

  The subtraction stage 24 subtracts (removes) the frequency component extracted from the input signal. That is, the subtraction stage 24 removes only a part of the frequency components that are not passed by the basic filter characteristics from the final output stage. In other words, a frequency component that satisfies the basic filter characteristics and a part of the frequency component that satisfies the inverse filter characteristics are output as final outputs.

(C) Third Method The extraction stage 23 described in the second method is preferably configured as follows, for example. FIG. 8 shows an example of the configuration. In FIG. 8, parts corresponding to those in FIG. The extraction stage 23 shown in FIG. 8 includes a threshold processing stage 23A that selects a frequency component that has passed the inverse characteristic filter stage 22 and whose signal level satisfies a predetermined condition, and a gain that changes the gain of the selected frequency component. And a change stage 23B. Note that the output of the threshold processing stage 23A may be output as it is. In other words, the gain change stage 23B may not be used.

  The extraction stage 23 shown in FIG. 8 is a configuration example in the case of adopting a method of further extracting only the frequency components that pass through the inverse characteristic filter stage 22 that satisfy a certain signal level and changing the gain. . The threshold value may be provided to the threshold processing stage 23A from the outside, or the threshold value may be stored in a storage unit prepared in advance in the threshold processing stage 23A.

  Here, the threshold value may be fixed, or may be changed as appropriate by the user. In addition, even when variable, an arbitrary value may be given, or a method of giving selectively from a plurality of prepared threshold values may be adopted. In any case, if the threshold value can be changed, the filter characteristics can be finely adjusted.

  If it is desired to obtain the filter characteristics shown in FIG. 3A, the threshold processing stage 23A extracts only a component having a high signal level (that is, a white area in the figure). If it is desired to obtain the filter characteristics shown in FIG. 3B, the threshold processing stage 23A extracts only a component having a low signal level (that is, a white area in the figure). The same applies when the basic filter characteristics are other characteristics.

  The gain changing stage 23B is provided to adjust the gain (amplitude) given to the extracted frequency component. The gain can be set to 1, that is, the output can be set without changing the amplitude of the input signal. In this case, the gain changing stage 23B functions as a buffer.

  The gain can also be obtained by calculation each time according to the signal level of the frequency component that has passed through the threshold processing stage 23A. The gain can also be read from a lookup table in which the relationship between the signal level and the gain is recorded in advance. The gain here may be fixedly given by a predetermined value or an arithmetic expression, or may be appropriately changed as described above.

  The gain may be fixed for each band. For example, the gain may be set large for a signal having a low signal level among the passed frequency components, and the gain may be set small for a signal having a high signal level.

(D) Fourth Method The extraction stage 23 described in the second method can be configured as follows, for example. FIG. 9 shows an example of the configuration. In FIG. 9, parts corresponding to those in FIG. The extraction stage 23 shown in FIG. 9 includes a detection stage 23C that detects the signal level of the frequency component that has passed through the inverse characteristic filter stage 22, and a gain setting stage 23D that adaptively sets the gain according to the detected signal level. The gain changing stage 23E changes the gain of the frequency component that has passed through the inverse characteristic filter stage 22 in accordance with the value set in the gain setting stage 23D.

The extraction stage 23 shown in FIG. 9 is a configuration example in the case of adopting a method of changing the gain according to the signal level of the frequency component that has passed through the inverse characteristic filter stage 22. Even with such a configuration, it is possible to output only a signal component that is finally desired to be removed with a sufficient gain.
For example, by setting a large gain for a signal having a low signal level and by setting a small gain for a signal having a high signal level, substantially the same result as that of the third method can be obtained. That is, even in the case of the extraction stage 23 having this configuration, it is possible to give only a signal having a substantially low signal level to the subtraction stage 24 among the signals in the entire band that has passed through the inverse characteristic filter stage 22. On the other hand, it is also possible to give only a signal having a substantially high signal level to the subtracter 24.

  Here, the detection stage 23C can output all the signal levels that can be taken by the frequency component, or can output it as a value of a plurality of stages. By limiting the output value to a plurality of values, it is possible to reduce the circuit scale of the subsequent stage and reduce the calculation amount.

  Further, the gain setting stage 23D can be obtained by calculation each time in accordance with the detected signal level. The gain can also be read from a lookup table in which the relationship between the signal level and the gain is recorded in advance. The gain here may be fixedly given by a predetermined value or an arithmetic expression, or may be appropriately changed as described above.

  The gain setting stage 23D acts so as to give a sufficient gain only to the signal component to be finally removed. If it is desired to obtain the same processing result as the threshold processing stage, the gain of the signal component to be finally passed may be set to zero. However, it is possible to obtain substantially the same effect by making the gain of the signal component finally passed very small. How to give the gain depends on the set value.

  The gain changing stage 23E amplifies the frequency component that has passed through the inverse characteristic filter stage 22 with the gain given by the gain setting stage 23D. The output of the gain changing stage 23E is output to the subtracting stage 24 as a signal component to be removed.

  According to the present invention, a frequency component that should be removed originally can be included in the filter output in accordance with the signal level, and the basic filter characteristics can be partially expanded. In addition, by using a filter having the characteristics, an electronic device or a signal processing circuit capable of more adaptive control can be realized.

Hereinafter, exemplary embodiments of the present invention will be described. It should be noted that characteristics not particularly shown or described herein are selected from those known in the art.
In the following description, a case where the preferred embodiment is realized as hardware will be described. However, it can also be realized by a computer program equivalent to such hardware.

When the present invention is implemented as a computer program, the program is stored in a computer-readable storage medium.
Examples of the storage medium include a magnetic storage medium such as a magnetic disk (flexible disk or hard disk) or magnetic tape, an optical storage medium such as an optical disk, an optical tape, or a machine-readable barcode, and a random access memory (RAM). In addition to semiconductor storage devices such as read only memory (ROM), other physical devices or media used to store computer programs are included.

  Where the present invention is implemented in hardware, it may be implemented as an integrated circuit such as an application specific integrated circuit (ASIC) or other device known in the art. The present invention can be implemented not only as a digital circuit but also as an analog circuit. The filter may be an active filter or a passive filter. Of course, a mixed circuit of an active type and a passive type may be used.

(A) First Embodiment (a-1) Overall Configuration of Digital Signal Processing Device FIG. 10 shows an example of a digital signal processing device to which a filter according to one embodiment of the present invention is applied. This digital signal processing apparatus is suitable for being mounted on an imaging apparatus.
First, light from a subject enters through an optical system and is picked up by an image pickup device 31 including an image sensor or the like. The image pickup device 31 includes three image sensors corresponding to three channels of R (red), G (green), and B (blue) of the three primary colors of light. From this image sensor, imaging signals S31 of three channels of R, G, and B are output and sent to the amplifier circuit 32.

The amplification circuit 32 performs various processes such as a black / white balance adjustment process, a black / white shading distortion correction process, a flare correction process, and the like on the three-channel imaging signal S31, and also performs signal amplification. The output of the amplifier circuit 32 is converted into a digital video signal S32 by an analog / digital (A / D) conversion circuit 33.
The digital video signal S32 is branched into two signal paths, one of which is sent to the knee correction circuit 35 which is the main line system, and the other is sent to the contour emphasis signal generation circuit 34 which is the interpolation system.

  The contour emphasis signal generation circuit 34 performs processing for enhancing the contour in each of the horizontal direction and the vertical direction, that is, processing for correcting the contour portion of the image on the video signal and increasing the resolution. Specifically, the contour enhancement signal generation circuit 34 generates a contour enhancement signal S33 that is a high-frequency signal. The contour enhancement signal S33 is generated for each of the vertical direction and the horizontal direction. The generated contour emphasis signal S33 is input to the pre-gamma processing adder 36 and the post-gamma processing adder 39, respectively.

  The knee correction circuit 35 performs a predetermined knee correction process on the digital video signal S32 using the slope of the input / output characteristics and the coefficient that gives the change point, and outputs the digital video signal S32. The digital video signal S34 after knee correction is added to the above-described contour emphasizing signal S33 in an adder 36 before gamma processing. The digital video signal S36 that has passed through the low-pass filter 37 out of the digital video signal S35 whose contour has been corrected in this way is sent to the gamma correction circuit 38. As the low-pass filter 37, a low-pass filter having filter characteristics according to the present invention is applied. Incidentally, in the application of the present embodiment, a filter having a characteristic of allowing a high-frequency component to pass a low signal level is used.

The gamma correction circuit 38 performs gamma correction using a gamma curve as shown in FIG. The digital video signal S37 after the gamma correction is added to the contour emphasis signal S33 in the post-gamma processing adder 39. The digital video signal S38 whose contour is corrected in this way is finally output.
Although not shown in the figure, a gain adjustment circuit for the outline emphasis signal S33 may be provided. That is, the distribution of the contour emphasis signal S33 to be added may be adjusted before and after gamma correction. By providing such a gain adjustment circuit, it is possible to finely adjust the degree of contouring and the occurrence of aliasing distortion.

(A-2) Configuration of Low-Pass Filter 37 and Gamma Correction Circuit 38 Next, detailed configurations of the low-pass filter 37 and the gamma correction circuit 38 will be described. The knee correction circuit 36 and the gamma correction circuit 38 described above are both non-linear processing circuits. The knee correction circuit 36 functions as a level compression unit, and the gamma correction circuit 38 functions as a level expansion unit. Therefore, in any circuit, there is a possibility that a harmonic component due to waveform distortion is generated.
However, in this specification, the configuration of the low-pass filter will be described particularly from the viewpoint of suppressing the aliasing component due to the gamma correction processing.

FIG. 12 shows components of the low-pass filter 37 and the gamma correction circuit 38. In general, when an input is represented as an x-axis and an output is represented as a y-axis, a narrow-section curve (here, a gamma curve) can be approximated by a linear expression of y = ax + b. The gamma correction circuit 38 uses this property. In this example, the coefficient a indicates the slope of the tangent line of the gamma curve in the target section, and the coefficient b indicates the y-axis intercept of the target section.
In an actual circuit, a gamma curve is realized by continuously connecting approximate straight lines represented by the linear expression.

  In the case of FIG. 12, the values of the slope a and the intercept b are stored in advance in a look-up table (LUT) 38A using a memory or the like. A value of the digital video signal S35 is given to the lookup table 38A as an address, and appropriate coefficients a and b corresponding to the value are read out. Of the read coefficients, the slope a is multiplied by the digital video signal S35 by the multiplier 38B, and the output and the intercept b are added by the adder 38C.

  At this time, since the digital video signal S36 given to the lookup table 38A has passed through the low-pass filter 37, the level change becomes smaller as the frequency becomes higher. That is, almost only the DC component is input to the look-up table 38A, the fluctuation range of the inclination a and the intercept b is reduced, and linear processing is performed. As a result, distortion is reduced and false signals are suppressed.

By the way, the aliasing distortion generated by the gamma correction becomes conspicuous on the image not only when the frequency of the input signal is high but also when the signal level is high. In other words, even if the frequency of the input signal is high, even if the signal level is low, there is little effect on the image even if aliasing occurs in gamma correction. On the other hand, such high frequency components contribute to enhancing the contour.
Therefore, when the low-pass filter 37 is configured, the high frequency is not limited simply, but a signal component having a low signal level even though it is a high-frequency signal component is configured to pass through the low-pass filter. That is, the filter is configured so as to have the filter characteristics shown in FIG.

FIG. 13 shows a specific configuration example of the low-pass filter 37. The low-pass filter 37 belongs to the filter having the structure shown in FIG. Therefore, a high-pass filter 37A having characteristics opposite to the basic filter characteristics is used for the input stage. The high-pass filter 37A passes only the high-frequency component of the digital video signal S35, which is input data, and restricts the passage of the low-frequency component.
The high-frequency component that has passed through the high-pass filter 37A is given to the threshold processing circuit 37B.

The threshold processing circuit 37B has the input / output characteristics shown in FIG. That is, the threshold processing circuit 37B passes only signal components having a signal level equal to or higher than the set threshold Sth, and blocks signal components having a signal level equal to or lower than the set threshold Sth. The output signal after the threshold processing is subtracted from the digital video signal S35 by the subtractor 37D, thereby realizing a filter through which some high-frequency components pass while maintaining the basic characteristics as a low-pass filter.
In this example, a multiplier 37C is arranged at the subsequent stage of the threshold processing circuit 37B. By arranging the multiplier 37C, the amplitude of the signal component input to the subtractor 37D can be adjusted. The amplitude is adjusted by changing the magnitude of the gain signal Sg.

Here, how the characteristics change between when the threshold processing circuit 37B is provided and when it is not provided will be described.
When the threshold value Sth is set to zero, the characteristics of the low-pass filter 37 are the inverse characteristics of the high-pass filter 37A used in the input stage. Accordingly, the filter characteristics in that case are symmetrical as shown in FIG. In FIG. 15, the horizontal axis represents frequency and the vertical axis represents gain.

  On the other hand, when a certain value is set for the threshold value Sth, the output of the high-pass filter 37A is shown by a solid line in FIG. The characteristic curve changes. That is, the curve indicating the frequency characteristic (high-pass filter characteristic) after the threshold processing is shifted to the right as compared with FIG. That is, the passing region shifts to the high frequency side. On the other hand, since a signal with a low signal level is cut off in the high frequency region, the curve has a reduced pass characteristic even in the high frequency region.

Also in this case, the filter characteristic of the low-pass filter 37 is given as the inverse characteristic of the high-pass filter 37A. For this reason, the filter characteristic of the low-pass filter 37 shifts to the high frequency side as indicated by a broken line in FIG. That is, it has a characteristic that the signal component passes to a higher frequency than in FIG. In addition, since the signal component with a low signal level can pass through the low pass filter 37 even in the high frequency component, the right end of the figure does not completely converge to zero.
Since the sampling frequency fs / 2 or more at the right end of the figure is not completely limited, the aliasing distortion naturally occurs.

However, by adjusting the threshold value S to an appropriate value, the low-pass filter process can be applied to the contour emphasis signal S33 having a signal level equal to or higher than the set threshold value. That is, the aliasing distortion caused by the contour enhancement signal S33 having such a signal level can be suppressed.
Further, the low-pass filter processing can be prevented from acting on the contour emphasis signal S33 having a signal level equal to or lower than the threshold value (for a signal component that does not cause deterioration in image quality due to aliasing distortion). That is, it is possible to apply gamma correction to the contour enhancement signal S33 so that an appropriate contour appears at the time of image reproduction.

Finally, the technical effect will be described with reference to the drawings. FIG. 17 shows the input / output characteristics of the gamma correction circuit 38 when a sawtooth wave is input to the gamma correction circuit 38 combined with the low-pass filter 37. Note that an edge emphasis waveform S33 having the same height above and below at an appropriate interval is added to the sawtooth wave.
FIG. 18 is an enlarged view of the contour portion of this waveform. In FIG. 18, the contour is set when the threshold setting is zero (normal low-pass filter configuration) (indicated by a broken line in the figure), and when a threshold value is set (indicated by a solid line in the figure). The waveform when the same level as the level of the emphasized waveform is set is shown.

As can be seen from FIG. 18, when the threshold is set on the upper side and the lower side of the contour emphasizing waveform, the waveform is closer to the lower side. This is the best evidence that gamma correction works effectively.
For example, when a low-pass filter having a general configuration is used as the input filter of the gamma correction circuit 38, no high-frequency component passes through the non-linear processing portion, and therefore no gamma correction is applied to the contour enhancement signal S33. This is as indicated by a broken line in the figure.

  On the other hand, as shown by the solid line in the figure, when the low-pass filter 37 having the configuration of the present embodiment is used, the contour emphasizing signal S33 having a small signal level is subjected to gamma correction, so that the contour emphasizing waveform below the upper side is longer. Become. That is, the waveform is more emphasized by the gamma correction. Thus, it is possible to perform gamma correction on the contour emphasis signal S33 for emphasizing the contour while suppressing the influence of aliasing distortion.

(A-3) Effect of Embodiment As described above, by using the low-pass filter 37 according to the present embodiment, it is possible to selectively pass a component having a small signal level even with a high-frequency component. Then, by inputting the output of the low-pass filter 37 to the gamma correction circuit 38, it is possible to apply gamma correction to the contour emphasis signal S33 having a signal level smaller than the set threshold value Sth.

  Thus, in the digital signal processing apparatus according to the present embodiment, filtering is performed on a signal having a large signal level to suppress aliasing distortion, and at the same time, gamma correction can be applied to a signal having a small level. That is, it is possible to improve the quality of the reproduced image by executing selective processing according to the signal level.

(B) Second Embodiment (b-1) Overall Configuration of Digital Signal Processing Device FIG. 19 shows an example of a digital signal processing device to which a filter according to one aspect of the present invention is applied. This digital signal processing apparatus is suitable for application to a sampling rate conversion apparatus. One aspect of the invention applies to this filter portion.

In general, the sampling rate conversion apparatus includes an up sampler 41, a low pass filter 42, and a down sampler 43.
The upsampler 41 samples the input signal at a frequency higher than the frequency before conversion. The sampling frequency at this time is set to the least common multiple of the frequency before conversion and the frequency after conversion.

The low-pass filter 42 performs low-pass filtering on the signal after upsampling in accordance with predetermined filter characteristics and outputs the filtered signal. At this time, the filter characteristic is set so as to cut off the lower Nyquist frequency or higher before and after the sampling rate conversion.
The down sampler 43 down-samples the low-pass filtered signal to a desired frequency and outputs it.
The conversion of the sampling rate is realized through such a series of processes.

The characteristic of the low-pass filter 42 determines the performance of the sampling rate converter. Depending on the characteristics of the low-pass filter 42, the waveform characteristics and frequency characteristics of the signal after the sampling rate conversion are determined.
Usually, the low-pass filter 42 is required to have a flat frequency characteristic up to the vicinity of the Nyquist frequency so as to extend to a frequency as high as possible.

  However, the fact that the frequency characteristic extends to the vicinity of the Nyquist frequency means that the cutoff characteristic of the low-pass filter becomes a steep characteristic. On the other hand, when the cutoff characteristic becomes steep, the order of the low-pass filter increases. For this reason, ringing occurs in the waveform after the sampling rate conversion, and the waveform characteristics are greatly impaired.

For example, when sampling rate conversion processing is performed on an image signal, ringing occurs before and after the edge of the subject, and the image quality is significantly impaired.
At this time, the larger the level difference before and after the edge, the more ringing occurs and the image quality deteriorates.

  In other words, if the level difference is small even with high frequency components such as edges, it means that the level of ringing that occurs is low and the effect on image quality is small. Therefore, it is considered that a low-pass filter that does not have a very steep cutoff characteristic is considered at a certain level or less, assuming that the influence on the image quality is small even if ringing occurs.

  In other words, a low-pass filter whose cutoff frequency extends to the Nyquist frequency as much as possible for a certain level or lower, that is, a low-pass filter having a steep characteristic, is applied, while a low-pass filter having a gentle characteristic that suppresses ringing as much as possible is switched to a certain level or higher. Think about it.

  This means that the characteristics of the low-pass filter are dynamically changed according to the level of the input signal. An example of the characteristics of the low-pass filter in this case is shown in FIG. The two low-pass filters LPF1 and LPF2 shown in FIG. 20 have the same cut-off frequency (cut-off frequency), but the characteristics of the transition region are steeper in the low-pass filter 2 than in the low-pass filter 1.

  Usually, when there is such a difference in characteristics, the orders of the two low-pass filters are different and the delay amount is different. Therefore, when applied to an actual circuit, a process for compensating for the delay amount of each low-pass filter and a process for preventing discontinuity near the switching of the low-pass filter are required.

  For reference, a specific example of this type of low-pass filter is shown in FIG. The low-pass filter 42A corresponds to the low-pass filter LPF1 in FIG. That is, the low-pass filter 42A has a characteristic in which the transition region indicated by the broken line is gentle. On the other hand, the low pass filter 42B corresponds to the low pass filter LPF2 of FIG. That is, the low-pass filter 42B has a characteristic that the transition region indicated by the solid line is steep.

  The blend circuit 42C receives two filter outputs and outputs either one or a blend value. The ratio here is determined by the threshold processing circuit 42E based on the input signal that has been delay compensated by the delay circuit 42D. The threshold processing circuit 42E applies LPF2 (steep characteristics) below the lower threshold 1, and applies LPF1 (slow characteristics) above the higher threshold 2. Note that the threshold processing circuit 42E applies two blend values in an intermediate range between the threshold 1 and the threshold 2. As described above, various processes are required in a normal circuit.

Therefore, one of the present invention is applied to the low-pass filter 42. The basic characteristics required for the low-pass filter 42 are the same as those in FIG. Therefore, the circuit configuration described in FIG. 13 is used.
The filter characteristic based on the low-pass filter 42 is set to LPF 2 (characteristic indicated by a solid line in FIG. 20) having a pass band up to the vicinity of the Nyquist frequency. Therefore, the high-pass filter 37A having a reverse characteristic of the above-mentioned LPF2 is used.

  The threshold processing circuit 37B passes only signal components having a signal level equal to or higher than the set threshold Sth among the passed high frequency components, and blocks signal components having a signal level equal to or lower than the set Sth. When the signal component after the threshold processing is subtracted from the input signal by the subtractor 37D, a high frequency component having a low signal level can be selectively passed while satisfying the pass band of the LPF2.

  That is, above the Nyquist frequency, it is possible to obtain the same output characteristics as when the LPF 1 having a gradual frequency characteristic is selected. Even in the case of the low-pass filter 42 according to this embodiment, the final output characteristics can be adjusted by arranging the multiplier 37D after the threshold processing circuit 37B.

(B-2) Effects of Embodiment As described above, if the low-pass filter 42 that selectively passes a component having a low signal level even in a high-frequency component is used, it is possible to have flat characteristics up to near Nyquist and the frequency. Moreover, no significant ringing occurs in the output after the sampling rate conversion. Therefore, the sampling rate can be converted without deteriorating the image quality.

The present invention can be applied to various video apparatuses having a gamma correction circuit. In addition, the electronic device using the filter having the above-described configuration can be applied to, for example, an imaging camera, a camera-integrated video recorder, and other devices having an imaging unit and its signal processing circuit.
Further, for example, the present invention can be applied to a recording device that records content (including audio, video, data, and the same applies hereinafter) on a video tape recorder, a disk recorder, or other recording medium, and its signal processing circuit.

Further, for example, the present invention can be applied to a playback device that plays back content from a video player, a disc player, a game device, or other recording media, and its signal processing circuit.
Further, for example, the present invention can be applied to a printer, a television receiver, a terminal with a display function, other output devices, and signal processing circuits thereof.

Further, for example, the present invention can be applied to a telephone, a transmission device, a reception device, other communication devices, and signal processing circuits thereof.
Further, for example, the present invention can be applied to various electronic devices that perform signal processing such as compression / decompression, encoding / decoding, and the like of content.

It is a figure which shows the structural example of the digital signal processing apparatus using a normal filter. It is a figure which shows the structure of the filter which concerns on a 1st method. It is a figure which shows the filter characteristic example in case a basic filter characteristic is a low-pass filter. It is a figure which shows the filter characteristic example in case a basic filter characteristic is a high pass filter. It is a figure which shows the filter characteristic example in case a basic filter characteristic is a band pass filter. It is a figure which shows the example of a filter characteristic in case a basic filter characteristic is a band elimination filter. It is a figure which shows the structural example of the filter which concerns on a 2nd method. It is a figure which shows the structural example of the filter which concerns on a 3rd method. It is a figure which shows the structural example of the filter which concerns on the method of FIG. It is a figure which shows the structural example of the digital signal processing apparatus using the filter of the embodiment. It is a figure which concerns on description of a gamma correction process. It is a figure which shows the detailed structure of a low-pass filter and a gamma correction circuit. It is a figure which shows the detailed structure of a low-pass filter. It is a figure which shows the input / output characteristic in a threshold value processing circuit. It is a figure which shows the frequency characteristic of a low-pass filter and a high-pass filter in case a threshold value is zero. It is a figure which shows the frequency characteristic of a low pass filter and a high pass filter in the case of performing a threshold value process. It is a figure which shows the input-output characteristic in a gamma correction circuit. It is an enlarged view of the input / output characteristics shown in FIG. It is a figure which shows the circuit structure of a sampling rate converter. It is a figure which shows two filter characteristics requested | required of a low-pass filter. It is a figure which shows the circuit structure normally used as a low-pass filter of FIG.

Explanation of symbols

11 Filter (conventional type)
21 Filter (present invention)
22 Inverse characteristic filter stage 23 Extraction stage 23A Threshold processing stage 23B Gain change stage 23C Detection stage 23D Gain setting stage 24 Subtraction stage 37 Low pass filter (present invention)
37A high-pass filter (conventional type)
37B Threshold processing circuit 37C Multiplier 37D Subtractor 38 Gamma correction circuit 38A Look-up table 38B Multiplier 38C Adder

Claims (12)

A filter comprising means for adaptively varying basic filter characteristics in accordance with the signal level of an input signal. An inverse characteristic filter stage for passing an input signal with a filter characteristic opposite to the basic filter characteristic;
An extraction stage that adaptively extracts at least a part of the frequency component according to the signal level of the frequency component that has passed through the inverse characteristic filter stage;
A filter comprising: a subtracting stage for subtracting the extracted frequency component from the input signal. The filter according to claim 2, wherein
The extraction stage includes
A filter comprising: a threshold processing stage that selects a frequency component having a signal level smaller than a threshold among frequency components that have passed through the inverse characteristic filter stage. The filter according to claim 2, wherein
The extraction stage includes
A filter comprising: a threshold processing stage that selects a frequency component having a signal level greater than a threshold among frequency components that have passed through the inverse characteristic filter stage. The filter according to claim 2, wherein
The extraction stage includes
A detection stage for detecting a signal level of a frequency component that has passed through the filter;
A gain setting stage for adaptively setting the gain according to the detected signal level;
And a gain changing stage for changing a gain of the frequency component that has passed through the inverse characteristic filter stage in accordance with a value set in the gain setting stage. The filter according to claim 1 or 2,
A signal processing apparatus comprising: gamma correction means for gamma correcting an output signal of the filter. A signal processing method comprising the step of adaptively varying the basic filter characteristics in accordance with the signal level of the input signal. Passing the input signal with a filter characteristic opposite to the basic filter characteristic;
Adaptively extracting at least part of the frequency component according to the signal level of the frequency component that has passed through the filtering step;
And subtracting the extracted frequency component from the input signal. On the computer,
A computer-readable recording medium storing a program for executing the function of adaptively changing the basic filter characteristics according to the signal level of the input signal.
On the computer,
A filter function that allows an input signal to pass by a filter characteristic opposite to the basic filter characteristic;
An extraction function for adaptively extracting at least part of the frequency component according to the signal level of the frequency component obtained by the filter function;
A computer-readable recording medium storing a program for executing a subtraction function for subtracting an extracted frequency component from the input signal. On the computer,
A program for executing a function that adaptively varies the basic filter characteristics according to the signal level of the input signal. On the computer,
A filter function that allows an input signal to pass by a filter characteristic opposite to the basic filter characteristic;
An extraction function for adaptively extracting at least part of the frequency component according to the signal level of the frequency component obtained by the filter function;
A subtracting function for subtracting the extracted frequency component from the input signal.

Images