JP2003230025A - Image processing apparatus, solid-state image pickup apparatus, and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing apparatus capable of obtaining an image with excellently high resolution and less noise. <P>SOLUTION: The image processing apparatus for processing an output signal from a solid-state imaging element to produce an image signal includes: a noise control section 24 for calculating a noise signal included in the image signal by each of (N+1) (N is an integer of 2 or over) sub bands and providing an output depending on the operating condition; filter groups 1 to 4 for decomposing the image signal including noise into (N+1) sub band components; noise subtraction circuit groups 25 to 28 for properly subtracting the noise signal of each sub band outputted from the noise control section 24 from the sub band component signal being an output signal of the filter groups 1 to 4; and an adder circuit 29 for summing output signals from the noise subtraction circuit groups 25 to 28 to produce and output the image signal from which the noise is reduced. Since the noise control section estimates the noise component included in the image signal and subtracts the estimated noise component from the image signal, the image processing apparatus can obtain the image signal with less noise. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing device, a solid-state imaging device, and an image processing method for reducing noise from an image signal picked up by a solid-state imaging device such as a CCD.

[0002]

2. Description of the Related Art FIG. 11 is a block diagram of a conventional image processing apparatus. This image processing device includes an AD conversion unit 11 that takes in an analog image signal output from a solid-state image pickup device such as a CCD, performs analog processing, and then converts the analog image signal into a digital image signal, a luminance signal Y and a color difference signal C.
an image signal pre-processing unit 12 which outputs the image data separately into r and Cb,
Low-pass filter 13 for reducing noise of luminance signal Y
, A low-pass filter 14 that reduces noise of the color difference signal Cr, a low-pass filter 15 that reduces noise of the color difference signal Cb, and an image signal post-processing unit 16 that takes in the outputs of the low-pass filters 13, 14, 15 and performs post-processing. With.

FIG. 12 is a schematic diagram showing a pixel array of a solid-state image pickup device. In the solid-state image sensor, hundreds of thousands to millions of pixels are arranged, and the light-receiving charge of each pixel is output as a captured image signal. When, for example, optical shot noise is mixed in a pixel 9 of the solid-state image sensor, the light-receiving charge of the pixel 9 contains a signal component and a noise component.

Therefore, conventionally, the periphery including the pixel 9 (see FIG.
2) LPF1 (the part surrounded by the dotted line) 3 × 3 = 9
The average value of the light-receiving charges of each pixel (5 × 5 = 25 or a wider range) may be calculated (the low-pass filters 13, 14, 15 perform the processing of calculating this average value), and this By making the average value the received light charge of the pixel 9, the noise component with respect to the signal charge of the pixel 9 is reduced. Since noise is generated randomly, and the probability that noise will enter both adjacent pixels is low, it is because the noise component is reduced by taking the average value of multiple pixels within the range to be processed by the low-pass filter. is there.

FIG. 13 is a block diagram of an image processing apparatus according to another conventional example. In this conventional example, median filters 13 ', 14' and 15 'are used instead of the low pass filters. In the low-pass filter, the average value of the received charges of a plurality of pixels is used as the received charges of the pixel 9, but in the median filter, the received charges of, for example, 3 × 3 = 9 are arranged in descending order, and the “central value”, in this example, The fifth largest received light charge is used as the received light charge of the pixel 9.

[0006]

Since the low-pass filter used in the above-described conventional image processing apparatus obtains the light-receiving charge of each pixel by taking the average value of the light-receiving charges of the surrounding pixels, the noise component is reduced. However, there is a problem in that the resolution of the image is deteriorated because the influence of the received light charges of the surrounding pixels is introduced. For example, when an image is captured in which the black and white boundary line is clear, there is a problem that the boundary line is blurred in gray.

Further, in the conventional technique using the median filter, the influence of the received charges of the surrounding pixels does not enter into the received charges which the median filter has adopted as the "center value". Although the resolution does not deteriorate, there is a problem that the signal deterioration may increase. For example, if there is a fine black stripe pattern in a white background image and that image is captured, it actually exists within the filter processing range around that pixel, even though it is the pixel for which "black" was captured. If the "center value" of the received electric charge of each pixel is "white", it is processed as "white". Therefore, there is a problem that the black line disappears although the black line actually exists.

An object of the present invention is to provide an image processing device, a solid-state image pickup device, and an image processing method capable of obtaining a high-resolution and good image with less noise.

[0009]

SUMMARY OF THE INVENTION In the image processing apparatus for processing an output signal from a solid-state image pickup device to generate an image signal, a noise signal included in the image signal is N + 1 (N is 2) depending on operating conditions. A noise control unit that calculates and outputs for each of the above (integer) subbands, a filter group that decomposes an image signal containing noise into N + 1 subband components, and a subband component that is an output signal of the filter group. A noise subtraction circuit group that appropriately subtracts the noise signal for each subband output from the noise control unit from a signal, and an output signal of the noise subtraction circuit group is added to generate and output a noise-reduced image signal. This is achieved by including an adding circuit. With this configuration, the image processing device of the present invention can obtain a good image with less noise and high resolution.

Preferably, a noise subtraction control unit for comparing the subband component signals and controlling the operation of the noise subtraction circuit group is provided. With this configuration, even when the noise of each subband is set to be large, it is possible to suppress the deterioration of the signal component related to the resolution and the image quality, and to significantly reduce the noise of the part that has little influence on the resolution and the image quality. .

The image signal is characterized by being R, G, B signals, or a luminance signal Y and color difference signals Cr, C
It is characterized by being b. The present invention can be applied to any image signal.

A solid-state image pickup device that achieves the above object can be achieved by including the above-described image processing device and solid-state image pickup element.

Further, an image processing method for achieving the above object is an image processing method for processing image signal data containing noise to reduce noise, wherein the noise signal contained in the image signal from the image signal data is N + 1 ( N is an integer greater than or equal to 2) a noise signal generation process for calculating each of the subbands,
A filter processing step of decomposing an image signal containing noise into N + 1 subband components, and a noise signal for each subband obtained in the noise signal generation step from the subband component signal obtained in the filter processing step is appropriately used. It is achieved by including a noise subtraction step of subtracting, and an addition step of adding signals for each subband obtained in the noise subtraction step to generate an image signal with reduced noise.

With this configuration, even after the image pickup image signal from the solid-state image pickup device is stored in the recording medium, the image pickup image signal can be read out from the recording medium and the noise reduction process can be performed by a personal computer or the like. it can.

Further, an image processing method for achieving the above object is an image processing method for processing image signal data including noise to reduce the noise, wherein N + 1 (the noise signal included in the image signal is included in the image signal data). N is an integer greater than or equal to 2) a noise signal generation process for calculating each of the subbands,
A filter processing step of decomposing an image signal containing noise into N + 1 subband components; a noise subtraction control signal generation step of generating a noise subtraction control signal from the subband component signals obtained in the filter processing step; A noise subtraction step of appropriately subtracting the noise signal for each subband obtained in the noise signal generation step from the band component signal based on the noise subtraction control signal, and the signal for each subband obtained in the noise subtraction step are added And an addition step of generating a noise-reduced image signal.

With this configuration, similarly to the above, it is possible to read the image signal stored in the recording medium and perform the noise reduction processing, and further, to significantly reduce the noise without deteriorating the resolution. You can The image signal can be applied to R, G, B signals, luminance signals and color difference signals.

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of an image processing apparatus according to the first embodiment of the present invention. This image processing device is a CCD
AD converter 11 that takes in an analog image signal output from a solid-state image sensor such as a digital image signal and performs analog processing, and then outputs the digital image signal after separating it into a luminance signal Y and color difference signals Cr and Cb. Image signal pre-processing unit 12, noise reduction circuit 21 for reducing noise of luminance signal Y, noise reduction circuit 22 for reducing noise of color difference signal Cr, and noise reduction for reducing noise of color difference signal Cb. A circuit 23 and an image signal post-processing unit 16 that takes in the outputs of the noise reduction circuits 21, 22, and 23 and performs post-processing are provided.

Further, this image processing apparatus has an AD converter 11
Of the control signal (for example, analog gain signal), the control signal of the image signal preprocessing unit 12 (for example, gamma characteristic signal or digital gain signal), and the luminance signal Y are taken in, and the noise reduction circuit 21, 22 and 23, and a noise control unit 24 that outputs a noise signal component to be subtracted.

In the present embodiment, the image processing device is described as being provided separately from the solid-state image pickup device, but the image processing device can be integrated into the solid-state image pickup device.

FIG. 2 is a detailed block diagram of the noise reduction circuit 21 shown in FIG. The configurations of the other noise reduction circuits 22 and 23 are the same as those in FIG. 2, and the signal to be processed is the color difference signal C.
Since only r and Cb are different, the detailed configuration of the noise reduction circuits 22 and 23 will be omitted.

In FIG. 2, the noise reduction circuit 21 is, in this example, four low-pass filters (hereinafter referred to as LPFs) 1, 2, 3, 4 and four difference circuits 5, 6, 7, 8.
And four noise subtraction circuits 25, 26, 27, 28 and 1
And one adder circuit 29.

As shown in FIG. 12, the LPF 1 performs a process of obtaining an average value of the received charges of, for example, 3 × 3 = 9 pixels around each pixel (pixel 9 is illustrated in FIG. 12), and the LPF 2
Is wider than the LPF 1 around the pixel 9, for example, 5 × 5.
= 25 pixels, an average value of the received light charges is calculated. Similarly, the LPF 3 sets 9 × 9 = 81 pixels as a processing target, and
The PF 4 processes 17 × 17 = 289 pixels.

FIG. 3 is a diagram showing the filter characteristics of the LPFs 1 to 4. In the illustrated example, the sub-bands are 1/2, 1/4, 1/8, 1/16, 1 / of the Nyquist frequency.
It is divided by 32, but this is because the processing range of the LPF is the area range described in FIG. 12, and it may be divided by any processing range, that is, any frequency (or band).

The fifth sub-band V has the same band as the filter characteristic of the LPF 4. The example of FIG. 3 shows the filter characteristic only in the x direction in FIG. 12, but the filter characteristic in the y direction is also the same. In addition, LPFi (i = 1
4 to 4), the pixel area (area of the weighted addition processing) is approximately 1 time, 2 times, 4 times, and 8 times in both the x direction and the y direction, and the low frequency subband has a wider pixel area. The average value is calculated with reference.

The difference circuit 5 shown in FIG.
Luminance signal Y from the LP and the LP that takes in the luminance signal Y
The difference from the output signal of F1 (subband I: see FIG. 3) is calculated and output to the noise subtraction circuit 25. Similarly, the difference circuit 6
Is L which takes in the output signal of the LPF 1 and the luminance signal Y.
The difference (subband II) from the output signal of PF2 is taken and output to the noise subtraction circuit 26.

Similarly, the difference circuit 7 takes the difference (subband III) between the output signal of the LPF 2 and the output signal of the LPF 3 which has taken in the luminance signal Y, and outputs it to the noise subtraction circuit 27. , The difference between the output signal of the LPF 3 and the output signal of the LPF 4 that captures the luminance signal Y (subband I
V) and outputs it to the noise subtraction circuit 28.

The noise subtraction circuit 25 subtracts the estimated noise signal component in the subband I sent from the noise control unit 24 (FIG. 1) from the output signal (subband I) of the difference circuit 5 and outputs the subtraction result. Output to the adder circuit 29. The noise subtraction circuit 26 is a subband sent from the noise control unit 24.
The estimated noise signal component in II is subtracted from the output signal (subband II) of the difference circuit 6 and the subtraction result is added by the addition circuit 29.
Output to.

Similarly, the noise subtraction circuit 27 subtracts the estimated noise signal component in the subband III sent from the noise control section 24 from the output signal (subband III) of the difference circuit 7 and adds the subtraction result. Output to the circuit 29. The noise subtraction circuit 28 subtracts the estimated noise signal component in the subband IV sent from the noise control unit 24 from the output signal (subband IV) of the difference circuit 8, and the subtraction result is added by the addition circuit 2
Output to 9.

The adder circuit 29 includes the noise subtraction circuits 25 and 2
The output signals of 6, 27 and 28 are taken in and LPF4
Of the output signal (subband V) of (1) and (2) are added, these signals are added, and the added signal is output to the image signal post-processing unit 16 as a luminance signal Y ′ with a reduced noise signal component.

Although there is noise in the subband V,
Since the noise is inconspicuous due to the low frequency, the noise signal component is not subtracted from the subband V signal in this embodiment. However, if it is desired to reduce the noise component of the subband V, another noise subtraction circuit may be provided.

The noise control section 24 estimates how much the noise signal contained in the luminance signal Y and the color difference signals Cr and Cb is, divides the estimated noise signal into components for each sub-band, and separates each noise. It outputs to the subtraction circuits 25 to 28.

Random noise of an image sensor such as a CCD is dominated by optical shot noise (white noise) in the case of a large output signal, and is small in the case of a small output signal.
The kTC noise and the noise of the output circuit are noticeable. The noise included in the image obtained by the image signal processing depends on the image sensor, the analog processing unit, and the digital processing unit.

In the case of an image processing device (or a solid-state image pickup device) in which an image sensor, an analog processing unit, and a digital processing unit are fixed, the output level of the image sensor, the analog gain of the analog signal processing circuit, and the digital gain of the image signal processing unit. From the gamma characteristics, it is possible to calculate (estimate) how much noise is generated in a theoretical manner.

Therefore, the noise control section 24 of the present embodiment is
A signal indicating the output level of the image sensor, the analog gain of the analog signal processing circuit, the digital gain of the image signal processing unit, the gamma characteristic, etc. is fetched and the noise signal is calculated (estimated). I have decided. If it is difficult to estimate, noise may be measured and calculated (estimated) from the measured value.

The image signal is a luminance signal Y and color difference signals Cr and C.
In the case of b, with respect to the luminance signal Y and the color difference signals Cr and Cb,
A noise table corresponding to the signal level is created for each subband. Even if the calculation (estimation) of the noise signal is roughly performed, there is no practical problem, and thus the noise table can be made simple.

For example, as shown in FIG.
The noise level of the color difference signals Cr and Cb is determined by the luminance signal Y. For example, if the noise of the color difference signal Cr is twice the noise of the luminance signal Y and the noise of the color difference signal Cb is three times the noise of the luminance signal Y, there is no practical problem.

Next, the processing operation of the image processing apparatus shown in FIGS. 1 and 2 will be described. When an analog image signal output captured by an image sensor such as a CCD is input to the AD conversion unit 11, the AD conversion unit 11 performs analog processing on this analog image signal to convert it into a digital image signal, and performs image signal preprocessing. Output to the unit 12.

The image signal preprocessing unit 12 digitally processes the digital image signal to obtain a luminance signal Y, a color difference signal Cr,
Cb is separated, the luminance signal Y is output to the noise reduction circuit 21, the color difference signal Cr is output to the noise reduction circuit 22, and the color difference signal Cb is output to the noise reduction circuit 23.

The noise control section 24 takes in the analog gain used by the AD conversion section 11 for the analog processing, the image signal preprocessing section 12 takes in the digital gain and the gamma characteristic signal used in the digital processing, and the brightness. The signal Y is fetched and, as described above, the noise signal components N _Yi , N _Cri ,
_Estimate N _Cbi (i = 1 to 4). Then, the noise signal component N _Yi is output to the noise reduction circuit 21, and the noise signal component N _Yi is output.
_Cri is output to the noise reduction circuit 22, and the noise signal component N _Cbi is output to the noise reduction circuit 23.

In the noise reduction circuit 21, the noise signal components N _Y1 , N _Y2 , N _Y3 and N _Y4 output from the noise control unit 24.
Noise subtraction circuits 25, 26, 27, and 28, respectively, and the noise signal components are subtracted from the difference circuits 5, 6, according to the following equation.
Luminance signals Y of subbands output from 7 and 8 respectively
Subtract from _H1 , Y _H2 , Y _H3 , Y _H4 . The following formula Y _Hi 'is
It is an output signal from each of the noise subtraction circuits 25 to 28.

When N _Yi ≧ | Y _Hi | Y _Hi '= 0 | Y _Hi |> N _Yi , Y _Hi > 0 Y _Hi ' = Y _Hi −N _Yi | Y _Hi |> N _Yi , Y _Hi In case of <0 Y _Hi '= Y _Hi + N _Yi

Then, the adder circuit 29 outputs the output signals Y _Hi 'of the noise subtraction circuits 25 to 28 and the signal Y of the subband V.
_The luminance signal _{Y'incorporating L4} and removing the noise component is output according to the following equation. Y '= _YH1 ' + _YH2 '+ _YH3 ' + _YH4 '+ _YL4

Similarly, the noise reduction circuit 22 outputs the color difference signal Cr 'from which the noise component is removed according to the following equation, and when N _Cri ≥ | Cr _Hi | Cr _Hi ' = 0 | Cr _Hi |> N _Cri , When Cr _Hi ＞ 0 Cr _Hi ′ ＝ Cr _Hi −N _Cri ｜ Cr _Hi ｜> N _Cri , Cr _Hi 〈0 Cr _Hi ′ = Cr _Hi ＋ N _Cri Cr ′ = Cr _H1 ′ + Cr _H2 ′ + Cr _H3 ′ + Cr _H4 '+ Cr _L4

Similarly, the noise reduction circuit 23 also outputs the color difference signal Cb 'from which the noise component has been removed according to the following equation. In the case of N _Cbi ≧ | Cb _Hi | In the case of Cb _Hi '= 0 | Cb _Hi |> N _Cbi , Cb _Hi > 0 Cb _Hi ' = Cb _Hi −N _Cbi | Cb _Hi |> N _Cbi , Cb _Hi <0 Case Cb _Hi '= Cb _Hi + N _Cbi Cb' = Cb _H1 '+ Cb _H2 ' + Cb _H3 '+ Cb _H4 ' + Cb _L4

The image signal post-processing unit 16 takes in the signals output from the noise reduction circuits 21, 22, and 23, performs post-processing, and outputs the processing result as an image-processed signal.

As described above, in the present embodiment, since the noise of each subband is theoretically estimated and only the noise component is subtracted from the image signal, the signal component is actually slightly reduced, but the resolution is deteriorated. It is possible to reduce noise without causing noise. It also estimates (calculates) the noise of each subband
Even if there is an error in, the noise component is significantly subtracted compared to the signal component, so that resolution degradation is small and noise can be reduced.

In the above-described embodiment, the image processing apparatus shown in FIG. 1 is composed of hardware.
It is also possible to realize the same function as in the first embodiment by software processing by PU). FIG. 5 is a flowchart according to the second embodiment of the present invention in which the noise reduction processing is performed by software.

When performing noise reduction by software processing, first, in step S1, the number of subbands (N + 1)
Is set, and in the next step S2, the parameters of the low-pass filter corresponding to each subband are set. Next, the noise components of the luminance signal Y and the color difference signals Cr and Cb of the image to be processed are estimated (step S3), and a noise table (luminance signal Y for each luminance signal Y and color difference signals Cr and Cb for each subband is estimated. 4) is created (step S4).

In the next step S5, the picked-up image signal of the image sensor is fetched, and in step S6, i = 1 to N
Up to Y _Li (output of low-pass filtering process) and Y _Hi
(Output of difference circuit processing step) is calculated, Cr _Li and Cr _Hi are calculated in the same manner, and Cb _Li and Cb _Hi are calculated.

In the next step S7, for the pixel (x, y), N _Yi , N _Cri , and N _Cbi are _calculated from the values of the luminance signal Y and the color difference signals Cr and Cb and the noise table, and Y _Li and Y _Hi. And N
_Y'is calculated from _Yi and C is calculated from Cr _Li , Cr _Hi and N _Cri.
r ′ is calculated, and Cb ′ is calculated from Cb _Li , Cb _Hi, and N _Cbi .

In the next step S8, the luminance signal Y'and the color difference signal C in which the noise component for the pixel (x, y) is reduced.
r ′ and Cb ′ are output and recorded, and it is determined in step S 9 whether or not the processing has been completed for all pixels of the processing target image. If the processing has not been completed for all pixels, Returning to step S7, when the processing has been completed for all pixels, the noise reduction processing of FIG. 5 is completed.

FIG. 6 is a flowchart showing another embodiment of the software processing (the third embodiment of the present invention). In the software processing shown in FIG. 5, the components of each subband are first calculated for the image signal of the entire screen, and then
Although the noise reduction processing is performed for each pixel, in the present embodiment (FIG. 6), the subband component is calculated for each pixel and the noise reduction processing is performed.

That is, in this embodiment, similarly to the embodiment of FIG. 5, the process proceeds to steps S1, S2, S3, S4 and S5, and at the next step S6 '(FIG. 6), the pixel (x, y) is selected. On the other hand, for i = 1 to N, Y _Li (output of low-pass filter processing step) and Y _Hi (output of difference circuit processing step) are calculated,
Similarly, Cr _Li and Cr _Hi are calculated, and Cb _Li and Cb _Hi are calculated. Then, similarly to FIG. 5, the process proceeds to steps S7 and S8, and if the processing for all pixels is not completed in the determination of step S9, the process returns to step S6 ′.

According to the present embodiment, since the sub-band component is calculated for each pixel, there is an advantage that the memory capacity required for processing can be small.

FIG. 7 is a block diagram of a noise reduction circuit of an image processing apparatus (basic configuration is the same as that of FIG. 1) according to the fourth embodiment of the present invention. This example also shows a noise reduction circuit for the luminance signal Y, but the noise reduction circuit for the color difference signals has the same configuration as in the first embodiment.

The configuration of FIG. 7 is basically the same as the configuration of FIG. 2, except that the input signal of each LPFi (i = 1 to 4) is the output signal of the preceding LPF. is there. That is, in the present embodiment, the luminance signal Y is LPF1.
The luminance signal Y and the output of LPF1 are input to the difference circuit 5, the output of LPF1 is input to the LPF2, and the output of LPF1 and the output of LPF2 are input to the difference circuit 6. Similarly, the output of LPF2 is input to the LPF3, the outputs of LPF2 and LPF3 are input to the difference circuit 7, the output of LPF3 is input to the LPF4, and the outputs of LPF3 and LPF4 are input to the difference circuit 8. Is entered.

The cutoff characteristic of the LPF is not limited to the case where the cutoff characteristic is good (the transmissivity is close to zero) as in the example shown in FIG. 3, but, for example, the cutoff characteristic is the bad characteristic shown in FIG. 8 (exemplified as LPF2). In some cases. In such a case, by adopting the fourth embodiment shown in FIG. 7, the output of the LPF2 is the LPF1.
And LPF2 are multiplied, and the advantage that the cutoff characteristic is improved can be obtained, and the selection range and accuracy of the weighting parameter of the filter can be widened.

FIG. 9 is a block diagram of the noise reduction circuit of the image processing apparatus according to the fifth embodiment of the present invention. This example also shows a noise reduction circuit for the luminance signal Y, but the noise reduction circuit for the color difference signals has the same configuration as in the first embodiment.

LPFs 1 to 4 and difference circuit 5 of this embodiment
8, the adder circuit 29 is the same as that of the first embodiment shown in FIG. 2, but in this embodiment, it is determined whether or not the output of each of the difference circuits 5 to 8 is taken in and the noise component subtraction process is performed. The difference is that the noise subtraction control unit 30 is provided.
The only difference is that instead of the noise subtraction circuits 25 to 28 in FIG. 2, noise subtraction circuits 31 to 34 that can receive a command from the noise subtraction control unit 30 are used.

The noise subtraction control unit 30 instructs the noise subtraction circuits 31 to 34 to turn on their operations to perform noise subtraction or to turn them off to not perform noise subtraction based on the following conditions. The control signal CYi is generated and output. This condition is the case where i = 1 where the influence of noise is large,
It is different from the case of i = 2, 3, 4. still,
T with a subscript is a predetermined threshold, and A with a subscript
Is a predetermined constant.

In the case of | Y _H1 |> T _{YH1, in} the case of C _Y1 = OFF signal | Y _H1 | ≦ T _YH1 , in the case of C _Y1 = ON signal | Y _Hi |> A _Y * | Y _Hi-1 | i = 2,3,4), C _Yi = OFF signal | Y _Hi | ≦ A _Y * | Y _Hi-1 | (i = 2, 3, 4), C _Yi = ON signal and OFF signal If the smallest subband is K, then K
All of the above C _Yi are OFF signals.

[0063] | Cr _H1 |> For T _CrH1, C _Cr1 = OFF signal | Cr _H1 | For ≦ T _CrH1, C _Cr1 = ON signal _{| Cr Hi |> A Cr *} | Cr Hi-1 | if ( i = 2,3,4), C _Cri = OFF signal | Cr _Hi | ≦ A _Cr * | Cr _Hi-1 | (i = 2,3,4), C _Cri = ON signal and OFF signal If the smallest subband is K, then K
The above C _Cri are all OFF signals.

[0064] | Cb _H1 |> For T _CbH1, C _Cb1 = OFF signal | Cb _H1 | For ≦ T _CbH1, C _Cb1 = ON signal _{| Cb Hi |> A Cb *} | Cb Hi-1 | if ( i = 2,3,4), C _Cbi = OFF signal | Cb _Hi | ≦ A _Cb * | Cb _Hi-1 | (i = 2,3,4), C _Cbi = ON signal and OFF signal If the smallest subband is K, then K
The above C _Cbi are all OFF signals.

The adder circuit 29 of each noise reduction circuit calculates and outputs the noise reduced luminance signal Y'and the color difference signals Cr ', Cb' as follows.

[Luminance signal Y] When C _Yi = OFF signal, Y _Hi '= Y _Hi C _Yi = ON signal, When N _Yi ≧ | Y _Hi |, Y _Hi ' = 0 C _Yi = ON signal , | Y _Hi |> N _Yi , Y _Hi > 0, Y _Hi '= Y _Hi −N _Yi C _Yi = ON signal, | Y _Hi |> N _Yi , Y _Hi <0, Y _Hi ' = Y _Hi + N _Yi Y '= Y _H1 ' + Y _H2 '+ Y _H3 ' + Y _H4 '+ Y _L4

[When Color Difference Signal Cr] C _Cri = OFF Signal, Cr _Hi '= Cr _Hi C _Cri = ON Signal, N _Cri ≧ | Cr _Hi |, Cr _Hi ' = 0 C _Cri = ON Signal , | Cr _Hi |> N _Cri , Cr _Hi > 0, Cr _Hi '= Cr _Hi- N _Cri C _Cri = ON signal, | Cr _Hi |> N _Cri , Cr _Hi <0, Cr _Hi ' = Cr _Hi + N _Cri Cr '= Cr _H1 ' + Cr _H2 '+ Cr _H3 ' + Cr _H4 '+ Cr _L4

[Case of Color Difference Signal Cb] When C _Cbi = OFF Signal, Cb _Hi '= Cb _Hi C _Cbi = ON Signal, When N _Cbi ≧ | Cb _Hi |, Cb _Hi ' = 0 C _Cbi = ON Signal , | Cb _Hi |> N _Cbi , Cb _Hi > 0, Cb _Hi '= Cb _Hi −N _Cbi C _Cbi = ON signal, | Cb _Hi |> N _Cbi , Cb _Hi <0, Cb _Hi ' = Cb _Hi + N _Cbi Cb '= Cb _H1 ' + Cb _H2 '+ Cb _H3 ' + Cb _H4 '+ Cb _L4

In the fifth embodiment, even when the noise of each sub-band is set to be large, the reduction of the signal component related to the resolution and the image quality is suppressed, and the noise of the part having a small influence on the resolution and the image quality is greatly reduced. There is an advantage that it can be reduced.
That is, resolution deterioration is small and it is possible to significantly reduce noise.

In the fifth embodiment, the noise subtraction OFF processing determination condition means that when there is a sharp image change in a distant pixel area, the noise reduction processing in the pixel area including it is avoided. That is, the noise reduction process is performed on the signal component having a strong correlation with the pixel to be processed, and the noise reduction process is not performed on the signal component having no correlation. This is because the noise reduction process for a signal having a small correlation is an erroneous process, which lowers the resolution and the image quality.

In addition, in the fifth embodiment, it is possible to combine the following two operations. The first operation is an operation that performs high-frequency noise reduction and does not perform low-frequency noise reduction processing. In this case, for example, the noise subtraction control signals of the third and fourth subbands are always OFF signals. The second operation is an operation that performs low-frequency noise reduction and does not perform high-frequency noise reduction. In this case, for example, the noise subtraction control signals of the first and second subbands are always turned off.
Signal. With this operation, low-frequency noise that particularly affects image quality is reduced, and high-frequency signal components related to resolution are not cut at all, so that high resolution can be maintained.

FIG. 10 is a flow chart according to the sixth embodiment of the present invention. Although the processing contents are the same as in the fifth embodiment, in this embodiment, image processing is performed by software processing by the CPU, as in the embodiments of FIGS. 5 and 6. The flow chart of the present embodiment is basically the same as the flow chart of FIG. 5, step S10 is provided between step S6 and step S7 of FIG. 5, and the determination result of step 9 indicates that all pixel processing has not been completed. When step S1
The only difference is that it is set back to 0.

In step S10, the pixel (x, y)
To the noise control signal C_Yi, Cr _i, Cb_iIs judged,
In the calculation of steps S7 and S8, the fifth embodiment described above is used.
The calculation is performed according to the calculation formula described in the section. to this
Therefore, the same effect as that of the fifth embodiment can be obtained.

The noise subtraction control unit 30 provided in the fifth embodiment of the present invention can be provided in the noise reduction circuit of the embodiment shown in FIG. Similarly, in the processing procedure of FIG. 6, it is possible to provide processing steps for performing the same processing as the noise subtraction control unit.

In the noise reduction processing by the software processing of FIGS. 5, 6 and 10, the sub-band component may be directly calculated from the luminance signal Y and the color difference signals Cr and Cb, or
Similar to the embodiment shown in FIG. 7, Y _Li-1 , Cr _{Li- 1} , and Cb _Li-1 may be used for the calculation for subband components of i = 2 or more. Furthermore, in the embodiment described above, L
Although the subband component is calculated by the PF and the difference circuit, it may be directly calculated from the bandpass filter having the frequency characteristic of the subband.

Further, in the above-described embodiment, an example having 5 subbands (4 LPFs) is shown, but the present invention can be applied as long as the number of subbands is 3 or more. Further, an example in which the noise reduction processing is performed on the image signals of the luminance signal Y and the color difference signals Cr and Cb has been described, but it goes without saying that the image signals of RGB (three primary colors) (signals before YC conversion).
A similar noise reduction process can be applied to. Further, in the case of a black and white image, noise reduction processing of only the luminance signal Y is sufficient.

Furthermore, in the above-mentioned embodiments, the image processing apparatus has a built-in digital still camera, and the noise reduction processing is performed before the image signal picked up by the solid-state image pickup apparatus is stored in the recording medium. However, it is also possible to store the image signal picked up by the digital still camera in the recording medium, read the image signal from the recording medium, and perform the noise reduction processing described above, and the same effect can be obtained. In this case, the software described with reference to FIGS. 5, 6, and 10 is installed in a personal computer or the like, and the user is the same as in the first, fourth, and fifth embodiments according to his / her preference. It is possible to perform the noise reduction processing by selecting the processing of (1) and the value of each parameter (threshold value, constant, etc.).

[0078]

According to the present invention, since the noise component can be greatly reduced while securing the signal component related to the resolution, it is possible to obtain a good image with less noise and high resolution.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an image processing apparatus according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of the noise reduction circuit shown in FIG.

FIG. 3 is a filter characteristic diagram of the noise reduction circuit shown in FIG.

FIG. 4 is a diagram showing an example of a noise table.

FIG. 5 is a flowchart showing an image processing procedure according to the second embodiment of the present invention.

FIG. 6 is a flowchart showing an image processing procedure according to the third embodiment of the present invention.

FIG. 7 is a configuration diagram of a noise reduction circuit of an image processing apparatus according to a fourth embodiment of the present invention.

FIG. 8 is a diagram showing an example of filter characteristics in which the noise reduction circuit shown in FIG. 7 is effective.

FIG. 9 is a configuration diagram of a noise reduction circuit of an image processing apparatus according to a fifth embodiment of the present invention.

FIG. 10 is a flowchart showing an image processing procedure according to the sixth embodiment of the present invention.

FIG. 11 is a configuration diagram of a conventional image processing apparatus.

FIG. 12 is an explanatory diagram of a low pass filter.

FIG. 13 is a block diagram of another conventional image processing apparatus.

[Description of Reference Signs] 1, 2, 3, 4 LPF (Low Pass Filter) 5, 6, 7, 8 Difference Circuit 11 AD Converter 12 Image Signal Preprocessor 16 Image Signal Postprocessor 21, 22, 23 Noise Reduction Circuit 24 Noise control unit 25, 26, 27, 28, 31, 32, 33, 34 Noise subtraction circuit 29 Addition circuit 30 Noise subtraction control unit

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) H04N 9/64 H04N 1/40 101C 5C077 F term (reference) 5B057 CA08 CA12 CA16 CB01 CB08 CB12 CB16 CC01 CE02 CH08 5C021 PA34 PA52 PA66 PA85 PA92 RB07 RB08 YA02 5C024 CX03 DX01 GY01 HX05 HX23 HX28 HX29 5C065 BB22 CC01 DD02 GG02 GG22 5C066 AA01 CA07 DD07 EC12 GA01 HA03 KC02 PP02QC07 PP08Q12 KF77 PP02 5C47 PP02 QC47 PP02

Claims (9)

[Claims] 1. An image processing apparatus for processing an output signal from a solid-state image pickup device to generate an image signal, wherein N + 1 (N is an integer of 2 or more) subband noise signals included in the image signal according to an operating condition. A noise control unit that calculates and outputs for each, a filter group that decomposes an image signal including noise into N + 1 subband components, and a subband component signal that is an output signal of the filter group that is output from the noise control unit A noise subtraction circuit group that appropriately subtracts the noise signal for each subband, and an addition circuit that adds the output signals of the noise subtraction circuit group to generate and output a noise-reduced image signal. Image processing device. 2. The image processing apparatus according to claim 1, further comprising a noise subtraction control unit that controls the operation of the noise subtraction circuit group by comparing the subband component signals. 3. The image processing apparatus according to claim 1, wherein the image signal is an R, G, B signal. 4. The image processing apparatus according to claim 1, wherein the image signal is a luminance signal Y and color difference signals Cr and Cb. 5. A solid-state image sensor, and claims 1 to 4.
A solid-state imaging device, comprising: the image processing device according to claim 1. 6. An image processing method for reducing noise by processing image signal data containing noise, wherein a noise signal included in the image signal from the image signal data is N + 1 (N is 2).
Noise signal generation step of calculating for each of the above (integer) subbands, a filter processing step of decomposing an image signal containing noise into N + 1 subband components, and a subband component signal obtained in the filter processing step A noise subtraction step of appropriately subtracting the noise signal for each sub-band obtained in the noise signal generation step from the above, and the signal for each sub-band obtained in the noise subtraction step is added to generate an image signal with reduced noise An image processing method, comprising: 7. An image processing method for reducing noise by processing image signal data containing noise, wherein a noise signal contained in the image signal from the image signal data is N + 1 (N is 2).
Noise signal generation step of calculating for each of the above (integer) subbands, a filter processing step of decomposing an image signal containing noise into N + 1 subband components, and a subband component signal obtained in the filter processing step A noise subtraction control signal generation step of generating a noise subtraction control signal from the noise subtraction control signal An image processing method comprising: a step of adding a signal for each subband obtained in the noise subtraction step to generate an image signal with reduced noise. 8. The image processing method according to claim 6, wherein the image signal is an R, G, B signal. 9. The image processing method according to claim 6, wherein the image signal is a luminance signal Y and color difference signals Cr and Cb.