JP2008199448A - Imaging apparatus, image processor and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an imaging apparatus capable of suppressing noise components according to the degree of noise amplification caused when performing a gradation correction. <P>SOLUTION: This imaging apparatus 1000 performs the gradation correction and noise reduction using wavelet transformation to a picked up image. The imaging apparatus 1000 includes: a gradation correction processing part 103 which performs the gradation correction for correcting contrast of an image signal based on a gradation correction curve representing input/output characteristics of the gradation correction; a wavelet transformation part 106 which divides the image signal into a plurality of frequency components; a coring processing part 109 which performs coring for noise reduction to the image signal divided into the plurality of frequency components by the wavelet transformation part 106 and a threshold setting part 108 which sets a coring threshold based on the gradation correction curve. The coring processing part 109 performs the coring based on the coring threshold set by the coring threshold setting part 108. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an imaging apparatus, an image processing apparatus, and an image processing method that perform gradation correction and noise reduction using wavelet transform on an image. Imaging devices are mainly represented by digital still cameras, digital video cameras, and the like.

  In general imaging devices such as digital cameras, tone correction is performed to enhance or weaken the contrast of an image by enhancing the contrast of the captured image using a tone correction curve (tone curve). ing. The gradation correction curve associates the pixel value of the input image with the pixel value of the output image. In the imaging apparatus, the gradation correction of the image is performed by performing conversion processing that correlates the brightness of the image and amplifies or attenuates the signal such as the luminance signal based on the input / output characteristics represented by the gradation correction curve. The

  Here, if tone correction is performed on an image with added noise using the above-described tone correction curve, a noise component that is not originally emphasized depends on the input / output characteristics represented by the tone correction curve. It was sometimes emphasized. For example, if the tone correction curve is set to amplify the input signal and output the output signal, the noise component originally contained in the input signal will be amplified at the same ratio in the output signal, which is essentially emphasized The noise component that I didn't want to have was emphasized.

  Conventionally, as a method for reducing noise in an image, for example, by calculating an average of pixel values of a target pixel and surrounding pixels, a smoothing filter called a low-pass filter is operated to remove a high-frequency component regarded as noise. There is known a method of removing noise from an image by suppressing the noise.

  As another method for reducing image noise, a noise reduction method using wavelet transform, which is one of multi-resolution conversions, is known. As an example of a noise reduction method using wavelet transform, for example, a method using wavelet shrinkage proposed by Donoho et al. In 1994 is known (see, for example, Non-Patent Document 1). . In this noise reduction method using wavelet degeneration, image noise is reduced as follows.

  First, using wavelet transform, the image signal is separated into a low frequency band and a high frequency band subband, and the same separation process is repeated for the low frequency band image signal. As a result, the image signal can be separated into subbands for each frequency component. Next, the coring process is performed by regarding the wavelet transform coefficient having a small absolute value as a noise component with respect to the high frequency component of the image obtained by the wavelet transform. In the coring process, when the absolute value of the wavelet transform coefficient obtained by the wavelet transform is smaller than a preset threshold value, a process of replacing the wavelet transform coefficient with 0 is performed. Then, after coring processing, the signal is reconstructed by performing wavelet inverse transform on the high frequency component and low frequency component of the image. Thereby, an image with reduced noise is obtained.

An example of a noise reduction method using wavelet transform is also disclosed in Patent Document 1, for example.
Hironobu Nakano, Ikuo Yamamoto, Ikuo Yoshida, "Signal processing and image processing using wavelets", first edition, Japan, Kyoritsu Shuppan, April 10, 2004, p.101 JP 2004-173078 A

  However, in the conventional imaging device, when noise removal processing after gradation correction is performed by a low-pass filter, the strength of noise removal is uniformly determined for one image by the characteristics of the selected low-pass filter. . When using wavelet transform, the same coring threshold is applied to one image. On the other hand, in the tone correction performed before noise reduction, the degree of noise amplification differs for each luminance due to the input / output characteristics of the tone correction curve.

  Thus, conventionally, the degree of noise amplification that occurs during tone correction is not taken into account, and the filter coefficient and coring threshold are uniformly determined regardless of changes in the degree of noise amplification. However, there is a problem that noise components that are not originally desired to be emphasized cannot be suppressed.

  The present invention has been made to solve the above-described conventional problems, and an imaging apparatus, an image processing apparatus, and an image that can suppress noise components in accordance with the degree of noise amplification that occurs when tone correction is performed. An object is to provide a processing method.

  An imaging apparatus according to the present invention is an imaging apparatus configured to perform gradation correction and noise reduction using wavelet transform on a captured image, and represents input / output characteristics of gradation correction. Based on the gradation correction curve, a gradation correction unit that performs gradation correction for correcting the contrast of the image signal, a wavelet conversion unit that separates the image signal into a plurality of frequency components, and the wavelet conversion unit A coring processing unit that performs coring for noise reduction on the image signal separated into frequency components, and a coring threshold value for the coring processing unit to perform the coring, the gradation correction curve A coring threshold value setting unit configured to set based on the coring threshold value setting unit, the coring processing unit based on the coring threshold value set by the coring threshold value setting unit Has a configuration for performing the coring.

  With this configuration, a coring threshold value for performing coring is set corresponding to the input / output characteristics represented by the gradation correction curve. The degree of noise amplification by gradation correction varies according to the input / output characteristics of gradation correction. Therefore, by setting the coring threshold based on the gradation correction curve, it is possible to suppress noise components according to the degree of noise amplification that occurs when performing gradation correction.

  In the imaging apparatus according to the aspect of the invention, the coring threshold value setting unit may determine the coring threshold value based on the gradation correction curve according to the input / output ratio of the gradation correction corresponding to the luminance of the target region of the wavelet transform. It has the structure which adjusts.

  With this configuration, the coring threshold value for performing coring can be set according to the input / output ratio represented by the tone correction curve, so that noise can be set according to the degree of noise amplification that occurs when tone correction is performed. Ingredients can be suppressed.

  In addition, the imaging apparatus of the present invention includes a wavelet inverse transform unit that performs wavelet inverse transform on the image signal that has been subjected to the coring by the coring processing unit, and reconstructs the image signal.

  With this configuration, an image in which the noise component is suppressed can be reconstructed from each frequency component that has been wavelet transformed and subjected to coring according to the degree of noise amplification that occurs when performing tone correction.

  An image correction apparatus according to the present invention is an image processing apparatus configured to perform gradation correction and noise reduction using wavelet transform on an image, and has input / output characteristics for gradation correction. A gradation correction unit that performs gradation correction for correcting contrast of an image signal based on a gradation correction curve that represents the wavelet conversion unit that separates the image signal into a plurality of frequency components; A coring processing unit that performs coring for noise reduction on the image signal separated into frequency components, and a coring threshold for the coring processing unit to perform the coring, the tone correction A coring threshold setting unit configured to set based on a curve, wherein the coring processing unit is based on the coring threshold set by the coring threshold setting unit. There are, it has a configuration for performing the coring.

  This configuration is not limited to the image processing technology of the imaging device. That is, the image processing device may or may not be provided in the imaging device. Then, the image processing of the present invention may be performed not only on an image captured by the imaging apparatus but also on a general image. This configuration also provides the advantages of the present invention described above.

  The image processing program of the present invention is an image processing program that causes a computer to perform gradation correction and noise reduction using wavelet transform on an image, and represents an input / output characteristic of gradation correction. Based on the tone correction curve, gradation correction for correcting the contrast of the image signal, separating the image signal into a plurality of frequency components, and separating the image signal into the plurality of frequency components. A step of performing coring for noise reduction on the separated image signal and a step of setting a coring threshold for performing the coring based on the gradation correction curve are executed in the computer. And in the step of performing the coring, the coralin set by the step of setting the coring threshold value. Based on the threshold value, and has a configuration of performing the coring.

  This configuration is not limited to the imaging apparatus. This configuration also provides the advantages of the present invention described above.

  Another aspect of the present invention is a computer-readable recording medium that records the above-described image processing program.

  The image processing method of the present invention is an image processing method for performing gradation correction and noise reduction using wavelet transform on an image, and represents a gradation representing input / output characteristics of gradation correction. Based on the correction curve, a gradation correction for correcting the contrast of the image signal, a step of separating the image signal into a plurality of frequency components, and a step of separating the image signal are separated into the plurality of frequency components. Performing coring for noise reduction on the processed image signal, and setting a coring threshold for performing the coring based on the gradation correction curve. In the step of performing, the coring is performed based on the coring threshold set in the step of setting the coring threshold.

  The application target of this image processing method is not limited to the imaging apparatus. Also by this image processing method, the above-described advantages of the present invention can be obtained.

  The present invention sets the coring threshold for noise reduction using wavelet transform based on the input / output characteristics represented by the tone correction curve, thereby adjusting the level of noise amplification that occurs when tone correction is performed. Thus, it is possible to provide an imaging apparatus, an image processing apparatus, and an image processing method having an effect of suppressing noise components.

Hereinafter, an imaging apparatus according to an embodiment of the present invention will be described with reference to the drawings.
A block diagram of an imaging apparatus according to an embodiment of the present invention is shown in FIG. As shown in FIG. 1, an imaging apparatus 1000 according to an embodiment of the present invention includes an imaging device 101, an image memory 102, a gradation correction processing unit 103, a gradation correction curve generation unit 104, and a luminance signal. A generation unit 105, a wavelet transformation unit 106, a coring threshold control curve generation unit 107, a coring threshold setting unit 108, a coring processing unit 109, a wavelet inverse transformation unit 110, and an image display unit 111 are provided. ing.

  The image sensor 101 is for capturing an image, and photoelectrically converts an input subject image. The image memory 102 stores an image obtained by the image sensor 101. The gradation correction processing unit 103 performs gradation correction for correcting the contrast of the image signal based on the gradation correction curve. Here, the gradation correction curve is also called a tone curve and represents input / output characteristics of gradation correction. The gradation correction curve is generated by the gradation correction curve generation unit 104. Note that the gradation correction processing unit 103 corresponds to a gradation correction unit of the present invention.

  An example of the gradation correction curve is shown in FIG. As shown in FIG. 2, in the gradation correction curve, the horizontal axis indicates the gradation of the original image (the luminance Ya of the input signal), and the darker part of the image goes to the left and the image goes to the right. The bright part of The vertical axis indicates the gradation (output Yb) of the image after correction. Thus, in the gradation correction curve, the luminance Ya and the output Yb of the input signal are related to each other.

  Accordingly, when one luminance Ya of the input signal is determined, one output Yb is determined based on the gradation correction curve. In FIG. 2, line A is a gradation correction curve that enhances the contrast of the image signal, line B is a gradation correction curve that does not change the contrast before and after conversion of the image signal, and line C is the contrast of the image signal. It is a gradation correction curve that weakens. The gradation correction curve shown in FIG. 2 is an example for explaining this embodiment.

  The gradation correction curve generation unit 104 generates a gradation correction curve for performing gradation correction on the image. When generating the gradation correction curve, the gradation correction curve generation unit 104 may register in advance the lines A, B, and C as shown in FIG. To create a unique tone correction curve, or apply an algorithm that automatically derives the tone correction curve so that the user can obtain the desired image. Any configuration may be applied as long as it is possible to read the tone correction curve that is the basis of the tone correction.

  The luminance signal generation unit 105 generates a luminance signal for the image stored in the image memory 102. A luminance signal is a signal representing the brightness (luminance) of an image and is also called a Y signal. For example, when the luminance signal is Y and the magnitudes of the three primary color image signals are R, G, and B, the luminance signal Y can be calculated using the calculation formula Y = (R + G + B) / 3. Note that the formula for calculating the luminance signal Y shown here is merely an example for explaining the present embodiment, and may be calculated using a calculation formula other than the one exemplified for the luminance signal.

  Here, FIG. 3 is a diagram for explaining the input / output ratio by the gradation correction curve. In FIG. 3, the line A shown in FIG. 2 is a gradation correction curve. Here, it is assumed that the luminance value Ya of the input signal is α and the output Yb is β. In this case, the tone correction input / output ratio R in the tone correction curve is R = β / α.

  Next, FIG. 4 shows an example of an input / output signal when the line A shown in FIG. 2 is a gradation correction curve, together with a noise amplification degree associated with the gradation correction. FIG. 4A is a diagram illustrating the relationship between the luminance Ya of the input signal and the output Yb in the gradation correction curve, and is a diagram for explaining that gradation correction is performed for each luminance of the input signal. FIG. 4B is a diagram showing the relationship between the luminance Ya1, Ya2, Ya3 of the input signal and the outputs Yb1, Yb2, Yb3.

  As shown in FIG. 4A, the luminances Ya1, Ya2, and Ya3 of the input signal are converted into outputs Yb1, Yb2, and Yb3 based on the line A that is the gradation correction curve. The gradation correction curve represents the input / output characteristics of gradation correction as described above. For example, the luminance Ya1 of the input signal is amplified up to the point where the line of Ya = Ya1 and the line A intersect, and converted to the output Yb1. Similarly, the luminances Ya2 and Ya2 of the input signal are also converted into outputs Yb2 and Yb3. In this way, tone correction using the tone correction curve is performed for each luminance of the input signal, and an output signal is output.

  Here, as shown in FIG. 4A, the luminance Ya of the input signal includes a fixed noise component Za, and the luminances Ya1, Ya2, and Ya3 of each input signal also include the same noise component Za. Yes.

  4B, from left to right in the figure, the relationship between the luminance Ya1 of the input signal and the output Yb1, the relationship between the luminance Ya2 of the input signal and the output Yb2, the luminance Ya3 of the input signal and the output Yb3 The relationship is shown.

  As shown in FIG. 4B, the noise component Za included in the luminance Ya of the input signal is amplified according to the input / output ratio R for gradation correction. That is, by converting the luminance Ya1 of the input signal into the output Yb1, the noise component Za is amplified to the noise component Zb1 according to the input / output ratio R = Yb1 / Ya1 corresponding to the luminance Ya1 of the input signal. . Similarly, by converting the luminance Ya2 of the input signal into the output Yb2, the noise component Za is amplified to the noise component Zb2 according to the input / output ratio R = Yb2 / Ya2 corresponding to the luminance Ya2 of the input signal. Yes. Similarly, the luminance Ya3 of the input signal is converted into the output Yb3, so that the noise component Za becomes a noise component in accordance with the tone correction input / output ratio R = Yb3 / Ya3 corresponding to the luminance Ya3 of the input signal. Amplified to Zb3. Thus, when the luminance of the input signal is increased or decreased according to the input / output ratio of the gradation correction curve, the noise component included in the input signal is also increased or decreased according to the input / output ratio of the gradation correction curve. Since the input / output ratio varies depending on the luminance, the noise amplification level also varies depending on the luminance. As shown in the example in the figure, when the tone correction curve is a line A that increases the contrast of an image, a signal with lower luminance has a larger noise amplification degree and a larger noise component after conversion.

  The wavelet transform unit 106 performs wavelet transform on the image whose tone has been corrected by the tone correction processing unit 103. The wavelet transform unit 106 performs wavelet transform on the tone-corrected image by separating the image signal into a plurality of frequency components as follows. An example of wavelet transformation by the wavelet transformation unit 106 will be described with reference to the drawings. FIG. 4 shows a level-1 wavelet transform process diagram. The wavelet transform is a kind of subband coding and is one of multi-resolution transforms. The level indicates the number of subband decompositions. Level-1 performs one subband decomposition, and level-2 performs two subband decompositions.

  As shown in FIG. 5, in the level-1 wavelet transform, first, an image signal is separated into a low-frequency component and a high-frequency component using a low-pass filter (LPF) and a high-pass filter (HPF). By subband decomposition. Further, the low-frequency component and the high-frequency component of the image signal after subband decomposition are subjected to subband decomposition using LPF and HPF.

  In this manner, the image signal is sequentially subband decomposed into a low frequency band including a low frequency signal component by the LPF and a high frequency band including a high frequency signal component by the HPF, so that the low frequency signal component (referred to as LL) , Vertical and diagonal high-frequency signal components (respectively, HL, LH, and HH) can be obtained. At this time, in the subband decomposition using the first-stage LPF and HPF, horizontal division is performed on the image, and in the subband decomposition using the second-stage LPF and HPF, vertical decomposition is performed on the image. The Further, by performing subband decomposition similar to the above on the low-frequency signal components one after another, higher-level wavelet transform can be performed.

  Next, a calculation method for wavelet transform will be described. Here, the wavelet transform calculation method will be described based on FIG. 6 and FIG. The Haar basis wavelet transform has a low accuracy of band separation, but is known as one of simple conversion methods in that the filter coefficient is composed of only two of “1” and “−1”. .

  6A and 6B are diagrams showing the process before and after the wavelet transform. FIG. 6A shows the original image, and FIG. 6B shows the separation state of each frequency component after the level-1 wavelet transform. FIG. 6C is a diagram illustrating a separation state of each frequency component after the level-2 wavelet transform. As shown in FIG. 6A and FIG. 6B, by performing wavelet transform on the image signal of the original image, the frequency components of 1LL, 1LH, 1HL, and 1HH are separated. Here, the numbers given in front of LL, LH, HL, and HH indicate the wavelet transform level. In the level-2 wavelet transform, wavelet transform is further performed on the low-frequency component 1LL of the image signal after the level-1 wavelet transform. As a result, as shown in FIG. 6C, the low-frequency component 1LL of the image signal after the level-1 wavelet transform is separated into frequency components of 2LL, 2HL, 2LH, and 2HH.

  A description will be given of a calculation formula for each frequency component of a pixel signal after wavelet transformation when the target region of wavelet transformation has a configuration of two horizontal stages and two vertical stages. FIG. 7A shows the respective components of the two-dimensional horizontal and vertical two pixels in the case where the target area of the wavelet transform has a configuration of two horizontal stages and two vertical stages. FIG. 7B shows a calculation formula for calculating the high-frequency signal component and the low-frequency signal component after wavelet transform.

  As shown in FIG. 7A, the two-dimensional horizontal and vertical two-pixel components can be expressed as [A, B, C, D]. In this case, the low-frequency component LL of the image signal after the level-1 wavelet transform can be calculated by a calculation formula of LL = A + B + C + D as shown in FIG. The horizontal, vertical, and diagonal high-frequency signal components HL, LH, and HH are calculated using the following formulas: HL = A−B + C−D, LH = A + B−C−D, and HH = A−B−C + D. Can do. LL, HL, LH, and HH are also called wavelet coefficients and wavelet transform coefficients. In addition, in each calculation formula shown by FIG.7 (b), the display of the level of the wavelet transformation which should be attached | subjected before LL, LH, HL, and HH is abbreviate | omitted.

  The low-frequency component 2LL and high-frequency components 2HL, 2LH, 2HH of the image signal after the level-2 wavelet transform shown in FIG. 6C are also calculated according to the calculation formula shown in FIG. be able to.

  As described above, the wavelet transform unit 106 separates the image signal into a plurality of frequency components LL, HL, LH, and HH. At this time, the wavelet transform unit 106 calculates each frequency component [LL, HL, LH, HH] of the image signal from each component [A, B, C, D] of two-dimensional horizontal and vertical two pixels.

  The coring threshold control curve generation unit 107 generates a characteristic curve (referred to as a coring threshold control curve) that serves as a basis for control of a coring processing threshold (referred to as a coring threshold) performed by the coring processing unit 109. The coring threshold control curve is used when the coring threshold setting unit 108 sets the coring threshold. The setting of the coring threshold and the coring process will be described in detail in the description of the coring threshold setting unit 108 and the coring processing unit 109.

  Here, the coring threshold control curve is for variably controlling the coring threshold according to the following principle of the present invention. As described above with reference to FIG. 4B, after the tone correction processing unit 103 performs tone correction based on the tone correction curve, the fixed noise component Za is also based on the tone correction curve. Amplification is performed in accordance with the input / output ratio R = β / α for gradation correction corresponding to the luminance of the input signal. Therefore, there are cases where the noise component after conversion increases as the luminance decreases. Therefore, when setting the coring threshold value for performing coring for noise reduction, the coring threshold value is set based on the characteristics of the tone correction curve so that the conversion result by tone correction is reflected. As a result, adaptive noise reduction can be performed in accordance with the degree of noise amplification that occurs when tone correction is performed. In the example shown in FIG. 4B, it can be seen that the lower the luminance, the larger the noise component after the conversion, so that the coring threshold for noise reduction needs to be increased on the low luminance side.

  FIG. 8 shows a diagram specifically showing the coring threshold control curve in relation to the gradation correction curve. FIG. 8A is a diagram showing an example of a coring threshold control curve in the case of using a gradation correction curve M that enhances contrast, and the gradation correction curve M here is the line A shown in FIG. It corresponds to. FIG. 8B is a diagram showing an example of the coring threshold control curve in the case where the gradation correction curve N that weakens the contrast is used. The gradation correction curve N is a line C shown in FIG. It corresponds to.

  In FIG. 8A and FIG. 8B, R = β / α (dotted line) is an input / output ratio curve obtained from the gradation correction curves M and N. The coring threshold control curve Rs is set along the input / output ratio curve R so that the coring threshold control curve Rs matches the input / output ratio curve R. However, as shown in the figure, in the region where the luminance of the input signal is low, the coring threshold control curve Rs is adjusted, so that the coring threshold control curve Rs is separated from the curve R of the input / output ratio. In the following description, the value of a point on the coring control threshold control curve Rs is also referred to as a coring threshold control value Rs using the same reference numeral.

  The adjustment of the coring threshold control curve Rs in the low luminance region will be described. As shown in FIG. 8A, when the luminance value of the input signal is low, the value of the input / output ratio R in the gradation correction curve M in the gradation correction curve M may become extremely large. Further, as shown in FIG. 8B, when the luminance value of the input signal is low, the value of the input / output ratio R in the gradation correction curve N may become extremely small. On the other hand, when the signal component is extremely small and the luminance value is low, the change in the converted noise component may be small depending on the gradation correction curve. Therefore, as illustrated, Rs when the luminance value is 0 is set to an appropriate value (Rs = 1 in the example in the figure). Then, this point (luminance = 0, Rs = 1) is connected to the curve of the input / output ratio R by a straight line to generate a coring threshold control curve Rs.

  Specifically, the coring threshold control curve Rs is generated as follows. As described above, Rs when the luminance value is 0 is 1 (i1 in FIG. 8A, j1 in FIG. 8B). In addition, a threshold value for setting the input / output ratio R to be effective at a predetermined luminance value or more is set. This threshold is referred to as the R validation threshold. The activation means that R = Rs, and the R activation threshold is a luminance value at the start end of the range of R = Rs. Rs when the luminance value is the R validation threshold corresponds to i2 in FIG. 8A and j2 in FIG. 8B. Further, a parameter for determining the sampling interval with the input / output ratio R as Rs is set. The R enable threshold is used as a starting point, and the input / output ratio R is sequentially calculated as Rs at this sampling interval (i3, i4, i5... In FIG. 8A, FIG. 8B). j3, j4, j5... These points are connected to generate a coring threshold control curve.

  The R validation threshold is a starting point at which the input / output ratio curve R and the coring threshold control curve Rs become the same as described above. The R validation threshold and the sampling interval can be freely determined. Depending on the R activation threshold, it is determined in which range on the low luminance side the coring threshold control curve Rs is separated from the input / output ratio R. Further, the fineness of the setting of the coring threshold Rs varies depending on the sampling interval. Therefore, the R validation threshold and the sampling interval are set according to the required processing accuracy.

  FIG. 9 shows a process for generating a coring threshold control curve. As shown in FIG. 9, first, the coring threshold control curve generation unit 107 calculates the input / output ratio R = β / α of the gradation correction curve based on the input / output characteristics of the gradation correction curve. Values are sequentially plotted to create an R curve (step (hereinafter referred to as S) 901). Next, based on the R curve, Rs that is actually used as a basis for controlling the coring threshold is created (S902). Here, the coring threshold control curve generation unit 107 sets Rs = 1 when the luminance value is 0, and sets the sampling start luminance threshold (the above-described R validation threshold) and the sampling interval set in advance. , Rs is sampled based on this setting, and the result is connected to obtain a coring threshold control curve Rs.

  In the example shown in FIGS. 8A and 8B, a coring threshold control curve Rs is shown in which the effective luminance value (R enabling threshold) is 40 and the sampling interval is 20. In the example shown in FIG. 8A, the coring threshold is obtained by connecting the point i1 corresponding to the luminance value of the input signal 0, the point i2 corresponding to the effective luminance value, and the points i3 to i13 obtained as a result of sampling. A control curve Rs is generated. In the example shown in FIG. 8B, the coring threshold value is obtained by connecting the point j1 corresponding to the luminance value of the input signal j1, the point j2 corresponding to the effective luminance value, and the points j3 to j13 obtained as a result of sampling. A control curve Rs is generated.

  The coring threshold setting unit 108 generates a coring threshold Thr for the coring processing unit 109 to perform coring based on the coring threshold control curve Rs. At this time, the coring threshold value setting unit 108 calculates the average luminance value Yave of the target region to be wavelet transformed by the wavelet transformation unit 106 from the luminance signal generated by the luminance signal generation unit 105, and the coring threshold value. The coring threshold control curve Rs generated by the control curve generation unit 107 is read out. Then, the coring threshold value setting unit 108 sets the coring threshold value Thr from the Rs value corresponding to the average luminance value Yave of the target region of the wavelet transform. As described above, the coring threshold Thr is a threshold used when the coring processing unit 109 performs coring.

  The coring threshold value Thr is calculated from the Rs value as follows. The default coring threshold for the target image is Thdef. Thdef may be preset and stored. Let Rs ′ be a value obtained from the coring threshold control curve Rs and the average luminance value Yave of the wavelet transform target area. Rs ′ is a value on the coring threshold control curve Rs corresponding to Yave. Then, the coring threshold value Thr actually used for coring is calculated by a calculation formula of Thr = Thdef × Rs ′ × t. Here, t is a coefficient for fine adjustment.

  An example of the coring threshold Thr generated by the coring threshold setting unit is shown in FIG. FIG. 10A is a diagram illustrating an example of a coring threshold when a gradation correction curve that increases contrast is used, and FIG. 10B is a coring threshold when a gradation correction curve that decreases contrast is used. It is a figure which shows an example. In FIG. 10A, the coring threshold Thr calculated when Thdef = 5 and t = 0.5 based on the coring threshold control curve Rs shown in FIG. It is shown by a curve with the brightness. In FIG. 10B, the coring threshold Thr calculated when Thdef = 5 and t = 0.5 based on the coring threshold control curve Rs shown in FIG. It is shown by a curve with the brightness.

  As shown in FIGS. 10A and 10B, the coring threshold value Thr is represented by a curve that changes according to the luminance, and is set based on the coring threshold value control curve Rs. Further, since the coring threshold control curve Rs is set based on the gradation correction input / output ratio R = β / α in the gradation correction curve, the coring threshold Thr is the gradation correction in the gradation correction curve. Is set based on the input / output ratio R = β / α. Therefore, the coring threshold value setting unit 108 can adjust the coring threshold value Thr according to the tone correction input / output ratio R = β / α in the tone correction curve.

  Here, as described with reference to FIGS. 4A and 4B, the noise component included in the luminance of the input signal is the input / output ratio R of the gradation correction in the gradation correction curve. = Amplified according to β / α. Accordingly, when the coring threshold value setting unit 108 sets the coring threshold value Thr, the coring threshold value setting unit 108 reflects the tone correction input / output ratio R = β / α in the tone correction curve. A coring threshold Thr can be set according to the degree of noise amplification that occurs. In other words, when the noise component increases in accordance with the degree of noise amplification that occurs when performing tone correction, the noise component is acted in the direction of increasing the coring threshold Thr that determines the strength of noise removal or reduction. When becomes small, it can be made to act in the direction which makes coring threshold value Thr small. As a result, adaptive noise reduction can be performed according to the degree of noise amplification that occurs when tone correction is performed.

  FIG. 11 shows a process for setting the coring threshold. As shown in FIG. 11, the coring threshold value setting unit 108 first obtains the average luminance value Yave of the pixels in the target region of the wavelet transform (S1101). Next, the coring threshold setting unit 108 reads the coring threshold control curve Rs generated by the coring threshold control curve generating unit 107, and obtains a coring threshold control curve value Rs' corresponding to Yave (S1102). Next, the coring threshold value setting unit 108 reads the coring threshold value Thdef and the fine adjustment coefficient t set as defaults in the main image (S1103). Then, the coring threshold setting unit 108 calculates the coring threshold Thr from the calculation formula Thr = Thdef × Rs ′ × t (S1104). As described above, the coring threshold setting unit 108 sets the coring threshold Thr with a curve that changes in accordance with the luminance value.

  Here, the coring threshold value Thr that is actually used for coring by the coring processing unit 109 is set for each average luminance value Yave of the wavelet transform target region as follows. FIG. 12 is a diagram for specifically explaining that the coring threshold Thr is set for each average luminance value Yave of the target region of the wavelet transform. FIG. 12 shows the coring threshold Thr when the average luminance value Yave of the wavelet transform target area is Yave = 50, 150 based on the result of FIG.

  As shown in FIG. 12, for example, when the average luminance value Yave of the target region of the wavelet transform is Yave = 50, the coring threshold Thr is set to Thr = Tha. Similarly, for example, when the average luminance value Yave of the target region of the wavelet transform is Yave = 150, the coring threshold Thr is set to Thr = Thb. Thus, the coring threshold value Thr that is actually used for coring by the coring processing unit 109 is set by the coring threshold value setting unit 108 for each average luminance value Yave of the target region of the wavelet transform.

  In FIG. 12, more specifically, the tone correction input / output ratio R = β / α differs for each average luminance value Yave of the wavelet transform target region, and the coring threshold is set from the input / output ratio R. The Therefore, the coring threshold Thr can be set according to the input / output ratio R of the gradation correction with respect to the luminance for each target area of the wavelet transform.

  A coring processing unit 109 performs a noise reduction core on the image signal separated into a plurality of frequency components by the wavelet transform unit 106 based on the coring threshold Thr set by the coring threshold setting unit 108. Do the ring. Here, the coring processor 109 performs a coring process on the high frequency components of the image obtained by the wavelet transform. In the coring process, when the absolute value of the wavelet transform coefficient obtained by the wavelet transform is smaller than the coring threshold Thr, a process of replacing the wavelet transform coefficient with 0 is performed.

  Specifically, the coring processing unit 109 uses the high-frequency component signal of the image separated by the wavelet transform unit 106 and the coring threshold Thr set by the coring threshold setting unit 108 to increase the image height. A coring process is performed on the band component. The coring processing unit 109 performs coring processing by replacing a coefficient whose absolute value of the coefficient of the high frequency component (wavelet coefficient, wavelet transform coefficient) of the image is smaller than the coring threshold by 0, and is included in the image Reduce noise. Note that the high frequency components of the image are, for example, 1HL, 1LH, and 1HH in FIG. 6B, and 1HL, 1LH, 1HH, 2HL, 2LH, and 2HH in FIG. 6C.

  An example of the coring input / output characteristic in the coring process based on the coring threshold is shown in FIG. FIG. 13 shows coring input / output characteristics when coring is performed based on the coring thresholds Tha and Thb set based on the result of FIG. FIG. 13A shows coring input / output characteristics based on the coring threshold Tha, and FIG. 13B shows coring input / output characteristics based on the coring threshold Thb. In FIGS. 13A and 13B, the horizontal axis is input (in) and the vertical axis is output (out).

  As shown in FIGS. 13A and 13B, in the coring processing by the coring processing unit 109, the values of the input coefficients (wavelet coefficients, wavelet transform coefficients) are the coring thresholds Tha and Thb. If the value is less than the absolute value, the value of the input coefficient is replaced with 0. On the other hand, when the value of the input coefficient (wavelet coefficient, wavelet transform coefficient) is larger than the absolute value of the coring thresholds Tha and Thb, the output coefficient is a value obtained by subtracting the coring thresholds Tha and Thb from the input coefficient. (When the input coefficient is negative, the coring thresholds Tha and Thb are added). Such a coring process is called soft coring.

  The wavelet inverse transform unit 110 performs wavelet inverse transform on the image signal that has been subjected to coring by the coring processing unit 109 to reconstruct the image signal. Specifically, the wavelet inverse transform unit 109 performs wavelet inverse transform using the high-frequency component cored by the coring processing unit 109 and the low-frequency component obtained by the wavelet transform unit 106, and performs image processing. Reconstruct the signal.

  An example of the wavelet inverse transform by the wavelet inverse transform unit 110 will be described with reference to the drawings. FIG. 14 shows a level-1 wavelet inverse transform process diagram. As shown in FIG. 14, a reconstruction function generated from a basis function used in a subband division process at the time of wavelet transform from each frequency component LL, LH, HL, HH of an image signal obtained by band division. The image signal can be reconstructed using LPF ′ and HPF ′. First, a low-frequency component LL that has passed through LPF ′ and a vertical high-frequency component LH that has passed through HPF ′ are combined, and a horizontal high-frequency component HL that has passed through LPF ′ and an oblique high-frequency component HH that has passed through HPF ′. And are synthesized. The combined signal of the low frequency component LL and the vertical high frequency component LH passes through LPF ′, and the combined signal of the horizontal high frequency component HL and the diagonal high frequency component HH passes through HPF ′ and passes through LPF ′ and HPF ′. When each synthesized signal is synthesized, an image signal is reconstructed as a reconstructed signal.

  Next, an example of an actual calculation method of wavelet inverse transformation will be described. Here, the Haar base described with reference to FIGS. 5 and 6 is used. Then, two-dimensional horizontal / vertical two-pixel components [A, B, C, D] from each frequency component [LL, LH, HL, HH], which are the results obtained by performing the level-1 wavelet transform. A calculation method for reconfiguring the data will be described with reference to FIGS. 15 and 16.

  FIG. 15 is a diagram illustrating a process before and after the wavelet inverse transformation, FIG. 15A is a diagram illustrating a separation state of each frequency component after the level-1 wavelet transform, and FIG. 15B is a reconstruction. It is a figure which shows the original image after. As shown in FIG. 15A and FIG. 15B, by performing wavelet inverse transform on each frequency component 1LL, 1LH, 1HL, 1HH of the image signal that has been wavelet transformed, the original image is obtained. Reconfigured. Here, the numbers given in front of LL, LH, HL, and HH indicate the wavelet transform level.

The calculation formula of the component of the pixel after the wavelet inverse transformation will be described in the case where the target area of the wavelet transformation has a configuration of two horizontal stages and two vertical stages. FIG. 16A shows a reconstructed image in the case where the target area of the wavelet transform has a configuration of two horizontal stages and two vertical stages.
As shown in FIG. 16A, the two-dimensional horizontal and vertical two-pixel components [A, B, C, D] are the same as in FIG. 7A. FIG. 16B shows a calculation formula for the pixel components [A, B, C, D] at this time.

  As shown in FIG. 16B, the components of each pixel of [A, B, C, D] are A = (LL + HL + LH + HH) / 4, B = (LL−HL + LH−HH) / 4, C = ( LL + HL−LH−HH) / 4 and D = (LL−HL−LH + HH) / 4. As described above, the wavelet inverse transform unit 110 reconstructs an image signal by performing wavelet inverse transform on the image signal subjected to coring by the coring processing unit 109. At this time, the wavelet inverse transform unit 110 calculates two-dimensional horizontal and vertical two-pixel components [A, B, C, D] from each frequency component [LL, LH, HL, HH] of the image signal.

  Here, in each calculation formula shown in FIG. 16B, the display of the level of the wavelet transform to be added before LL, LH, HL, and HH is omitted. In addition, when each frequency component of the image signal obtained by the high frequency division is obtained by wavelet transform of level-2 or higher, the LL component reconstructed from the upper level and the high frequency component of the lower level are shown in FIG. Similarly, an image can be reconstructed by sequentially performing inverse wavelet transform according to the calculation formula shown in FIG.

  The image display unit 111 displays the image that has been subjected to the wavelet inverse transform by the wavelet inverse transform unit 110. The image display unit 111 is configured by, for example, a liquid crystal display device or an organic EL display device.

  Next, the operation of the image pickup apparatus according to the embodiment of the present invention will be described. FIG. 17 shows an operation flow of the imaging apparatus according to the embodiment of the present invention. As shown in FIG. 17, first, the image sensor 101 captures a subject image (S1701). An image captured by the image sensor 101 is stored in the image memory 102. Next, the gradation correction processing unit 103 performs gradation correction on the image stored in the image memory 102 to correct the contrast of the image signal (S1702). At this time, the gradation correction processing unit 103 uses the luminance signal generated by the luminance signal generation unit 105 based on the gradation correction curve generated by the gradation correction curve generation unit 104 as an output after gradation correction. Convert. Here, as described with reference to FIGS. 4A and 4B, the luminance of the input signal is increased or decreased in accordance with the tone correction input / output ratio R in the tone correction curve. The noise component included in the input signal is also increased or decreased according to the gradation correction input / output ratio R in the gradation correction curve.

  Next, the wavelet transform unit 106 performs wavelet transform that separates the image signal whose tone has been corrected by the tone correction processing unit 103 into a plurality of frequency components (S1703). Here, the wavelet transform unit 106 performs subband decomposition by separating the image signal into a low frequency component and a high frequency component. For example, when the wavelet transform unit 106 performs level-1 wavelet transform, as shown in FIGS. 5 and 6B, the low-frequency signal component (LL) and the horizontal, vertical, and diagonal high-frequency signal components ( HL, LH, HH). In addition, as described with reference to FIG. 7, the wavelet transform unit 106 performs wavelet transform on the image signal that has been subjected to gradation correction by the gradation correction processing unit 103, so that each frequency component LL, HL, LH, HH. Is calculated.

  Next, the coring threshold setting unit 108 sets the coring threshold (S1704), and the coring processing unit 109 uses the wavelet transform unit 106 based on the coring threshold Thr set by the coring threshold setting unit 108. The separated image signal is subjected to coring for noise reduction (S1705). As described above, the coring threshold value setting unit 108 calculates the average luminance value Yave of the target region to be wavelet transformed by the wavelet transformation unit 106 from the luminance signal generated by the luminance signal generation unit 105, and generates a coring threshold control curve. The coring threshold control curve Rs generated by the unit 107 is read, and the coring threshold Thr is set from the Rs value corresponding to the average luminance value Yave of the wavelet transform target region. When the coring processing unit 109 has a value of a coefficient (wavelet coefficient, wavelet transform coefficient) input as illustrated with reference to FIGS. 13A and 13B smaller than the absolute value of the coring threshold The input coefficient value is replaced with 0 and output. The coring processing by the coring processing unit 110 is performed for each frequency component of the image separated by the wavelet transform.

  Here, as described above, the coring threshold value setting unit 108 reflects the tone correction input / output ratio R = β / α in the tone correction curve when setting the coring threshold value Thr. A coring threshold Thr can be set according to the degree of noise amplification that occurs when correction is performed. In other words, when the noise component increases in accordance with the degree of noise amplification that occurs when performing tone correction, the noise component is acted in the direction of increasing the coring threshold Thr that determines the strength of noise removal or reduction. When becomes small, it can be made to act in the direction which makes coring threshold value Thr small. As a result, adaptive noise reduction can be performed according to the degree of noise amplification that occurs when tone correction is performed.

  Next, the wavelet inverse transform unit 110 performs the wavelet inverse transform on each frequency component of the image cored by the coring processing unit 109 to reconstruct the image (S1706). Here, the wavelet inverse transform unit 109 performs wavelet inverse transform using the high-frequency component cored by the coring processing unit 109 and the low-frequency component obtained by the wavelet transform unit 106, and performs image signal conversion. Reconfigure. Accordingly, as illustrated in FIG. 15 and FIG. 16, for example, from each frequency component [LL, LH, HL, HH], which is a result obtained by performing level-1 wavelet transform using a Haar basis, The original image is reconstructed by reconstructing the horizontal and vertical two-pixel components [A, B, C, D]. Thus, each frequency component of the luminance signal that has been wavelet transformed and subjected to coring can be synthesized, and an image in which the noise component is suppressed can be reconstructed according to the degree of noise amplification that occurs when performing tone correction.

  Next, the image reconstructed by the wavelet inverse transform of the wavelet inverse transform unit 110 is displayed by the image display unit 111 (S1707).

  As described above, according to the imaging apparatus 1000 according to the embodiment of the present invention, the coring threshold Thr for performing coring for noise reduction is adjusted by the coring threshold setting unit 108 with the tone correction curve. Is set based on The degree of noise amplification by gradation correction changes according to the input / output characteristics of the gradation correction curve. Therefore, by setting the coring threshold according to the input / output characteristics of the gradation correction, it is possible to suppress the noise component according to the degree of noise amplification that occurs when the gradation correction is performed. As a result, blurring due to insufficient noise removal and excessive removal of high-frequency components is reduced, and it is possible to obtain an image that reflects the effect of tone correction using the applied tone correction curve while suppressing noise enhancement. It becomes.

  Further, according to the imaging apparatus 1000 according to the embodiment of the present invention, the coring threshold value for performing coring for noise reduction is determined by the coring threshold value setting unit 108 based on the tone correction curve and wavelet transform. Is adjusted according to the gradation correction input / output ratio R corresponding to the luminance of the target area, so that the coring threshold Thr for performing coring is set to the gradation correction input / output represented by the gradation correction curve. It can be set according to the ratio R, and the noise component can be suppressed according to the degree of noise amplification that occurs when tone correction is performed.

  The embodiments of the present invention have been described above by way of example, but the scope of the present invention is not limited to these embodiments, and can be changed or modified according to the purpose within the scope of the claims. is there.

  In the above embodiment, the imaging apparatus, the image processing apparatus, and the image processing method according to the present invention have been specifically described. However, the present invention is not limited to these. The present invention may be an image correction program that causes a computer to execute each step of the above-described image processing method, or may be a computer-readable recording medium that records this image processing program.

  As described above, according to the present invention, the coring threshold value for performing coring is set in accordance with the input / output characteristics represented by the gradation correction curve. Depending on the degree, the noise component can be suppressed, which is useful for an imaging device or the like.

The block diagram of the imaging device which concerns on embodiment of this invention Diagram showing an example of the tone correction curve Diagram for explaining the input / output ratio by gradation correction curve (A) It is a figure which shows the relationship between the brightness | luminance Ya of the input signal in the gradation correction curve, and the output Yb, Comprising: The figure explaining that gradation correction | amendment is performed according to the brightness | luminance of an input signal (b) The brightness | luminance of an input signal The figure which shows the relationship between Ya1, Ya2, Ya3 and output Yb1, Yb2, Yb3 Level-1 wavelet transform process diagram (A) The figure which shows an original image (b) The figure which shows the separation state of each frequency component after wavelet transformation of level-1 (c) The figure which shows the separation state of each frequency component after wavelet transformation of level-2 (A) The figure which shows each component of the two-dimensional horizontal and vertical 2 pixel in case the object area | region of wavelet transformation is the structure of 2 steps | paragraphs of horizontal and 2 steps | paragraphs. (B) The high frequency signal component and low frequency signal after wavelet transformation The figure which shows the calculation formula for calculating the component (A) A figure showing an example of a coring threshold control curve when a gradation correction curve for increasing contrast is used (b) A figure showing an example of a coring threshold control curve when using a gradation correction curve for reducing contrast The figure which shows the process which produces | generates a coring threshold value control curve (A) A figure showing an example of a coring threshold when a gradation correction curve for increasing contrast is used (b) A figure showing an example of a coring threshold when a gradation correction curve for reducing contrast is used The figure which shows the process which sets a coring threshold The figure for demonstrating concretely that the coring threshold value Thr is set for every average luminance value Yave of the object area | region of a wavelet transform. An example of coring input / output characteristics in coring processing based on coring threshold wave-1 inverse wavelet transform process diagram (A) The figure which shows the separation state of each frequency component after wavelet transformation of level-1 (b) The figure which shows the original image after a reconstruction (A) The figure which shows the reconstructed image in case the object area | region of a wavelet transformation is the structure of 2 steps | paragraphs of horizontal and 2 steps | paragraphs. (B) For calculating the component of the two-dimensional horizontal and vertical 2 pixel after wavelet inverse transformation Figure showing calculation formula Operation flow of imaging apparatus according to embodiment of the present invention

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Image pick-up element 102 Image memory 103 Tone correction process part 104 Tone correction curve generation part 105 Luminance signal generation part 106 Wavelet conversion part 107 Coring threshold control curve generation part 108 Coring threshold value setting part 109 Coring process part 110 Wavelet reverse part Conversion unit 111 Image display unit 1000 Imaging device

Claims (7)

An imaging apparatus configured to perform tone correction and noise reduction using wavelet transform on a captured image,
A gradation correction unit that performs gradation correction for correcting the contrast of an image signal based on a gradation correction curve that represents input / output characteristics of gradation correction;
A wavelet transform unit that separates the image signal into a plurality of frequency components;
A coring processing unit that performs coring for noise reduction on the image signal separated into the plurality of frequency components by the wavelet transform unit;
A coring threshold setting unit for setting a coring threshold for the coring processing unit to perform the coring based on the gradation correction curve;
The imaging apparatus, wherein the coring processing unit performs the coring based on the coring threshold set by the coring threshold setting unit.   The coring threshold setting unit adjusts the coring threshold based on the tone correction curve according to a tone correction input / output ratio corresponding to a luminance of a target region of wavelet transform based on the tone correction curve. Item 2. The imaging device according to Item 1.   3. The wavelet inverse transform unit configured to perform wavelet inverse transform on the image signal subjected to the coring by the coring processing unit to reconstruct the image signal. The imaging device described. An image processing apparatus configured to perform gradation correction and noise reduction using wavelet transform on an image,
A gradation correction unit that performs gradation correction for correcting the contrast of an image signal based on a gradation correction curve that represents input / output characteristics of gradation correction;
A wavelet transform unit that separates the image signal into a plurality of frequency components;
A coring processing unit that performs coring for noise reduction on the image signal separated into the plurality of frequency components by the wavelet transform unit;
A coring threshold setting unit for setting a coring threshold for the coring processing unit to perform the coring based on the gradation correction curve;
The image processing apparatus, wherein the coring processing unit performs the coring based on the coring threshold set by the coring threshold setting unit. An image processing program for causing a computer to perform tone correction and noise reduction using wavelet transform on an image,
Performing gradation correction for correcting contrast of an image signal based on a gradation correction curve representing input / output characteristics of gradation correction;
Separating the image signal into a plurality of frequency components;
Performing coring for noise reduction on the image signal separated into the plurality of frequency components by the step of separating the image signal;
Causing the computer to execute a step of setting a coring threshold for performing the coring based on the gradation correction curve,
In the coring step, the coring is performed based on the coring threshold set in the step of setting the coring threshold.   A computer-readable recording medium on which the image processing program according to claim 5 is recorded. An image processing method for performing tone correction and noise reduction using wavelet transform on an image,
Performing gradation correction for correcting contrast of an image signal based on a gradation correction curve representing input / output characteristics of gradation correction;
Separating the image signal into a plurality of frequency components;
Performing coring for noise reduction on the image signal separated into the plurality of frequency components by the step of separating the image signal;
Setting a coring threshold for performing the coring based on the gradation correction curve,
In the step of performing coring, the coring is performed based on the coring threshold set in the step of setting the coring threshold.