JP2007316983A - Image processor, image processing method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor removing noises while reducing edge blur caused by coring processing. <P>SOLUTION: The image processor comprises an image pick-up element 101 which takes an image; a wavelet transform part 103 which performs wavelet transform of the image taken by the image pickup element 101; an edge detection part 104 which detects edges contained in the image inputted by the image pick-up element 101; a coring threshold determination part 105 which determines, for each unit area of coring processing, a threshold of coring according to the detection result of edges by the edge detection part 104; a coring processing part 106 which cores a signal transformed by the wavelet transform part 103 by use of the threshold determined by the coring threshold determination part; and a wavelet inverse transform part 107 performing wavelet inverse transform of the signal cored by the coring processing part 106. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and method for removing noise by performing image processing on an image including noise, and a program thereof.

  2. Description of the Related Art Conventionally, as a method for removing noise in image processing, a method of taking an average of neighboring pixels is known. In this method, a high effect can be obtained when the correlation between the pixels is high. However, in the case of a pixel that does not have a high correlation, blurring of the edge or the like may occur, and the contour portion may become unclear.

Conventionally, a noise reduction method using wavelet transform is known (Patent Document 1). Wavelet transform is one of multi-resolution transforms. In this noise reduction method, first, an image signal is separated into subbands of a low frequency band and a high frequency band based on wavelet transform. Next, the obtained subband signal is further separated into a low frequency band and a high frequency band. Thus, the image signal can be separated into subbands for each frequency band. Next, based on the premise that the wavelet transform coefficient having a small value with respect to the high frequency component obtained by the wavelet transform is a noise component, the transform coefficient whose absolute value is smaller than a predetermined threshold is replaced with 0. Perform the coring process. Then, the noise is reduced by reconstructing the signal by performing wavelet inverse transformation on the high frequency component and the low frequency component subjected to the coring process. This method uses a method called wavelet degeneration proposed by Donoho et al. In 1994. Wavelet degeneration is introduced in Non-Patent Document 1.
JP-A-9-212623 Hironobu Nakano, Ikuo Yamamoto, Ikuo Yoshida “Signal Processing and Image Processing with Wavelets” Kyoritsu Shuppan, p101

  In the method of averaging with surrounding pixels, the noise component is averaged with the surrounding pixels and smoothed to remove the noise. In some cases, the contour is smoothed and blurred.

  Further, even in the method using the wavelet degeneration described in Patent Document 1, since it is impossible to distinguish between noise and a signal including an edge, noise removal is uniformly performed up to an edge portion including a high-frequency component. For this reason, the edge portion may be blurred simultaneously with noise removal.

  In view of the above background, an object of the present invention is to provide an image processing apparatus capable of removing noise while reducing edge blurring caused by coring processing.

  An image processing apparatus according to the present invention is a device that performs noise reduction processing using wavelet degeneration on an input image, and is an edge detection unit that detects edges included in an image and a unit of coring processing. A coring threshold value determining unit that determines a coring threshold value according to the presence or absence of an edge detected by the edge detecting unit is provided for each region.

  With this configuration, since the coring threshold is determined according to the presence or absence of an edge, noise can be removed while suppressing blurring of the edge portion, and a clear image can be obtained.

  An image processing apparatus according to another aspect of the present invention includes an image input unit that inputs an image, a wavelet transform unit that performs wavelet transform on the image input by the image input unit, and an image that is input by the image input unit. An edge detection unit that detects an edge included in the coring process, and a coring threshold value determination unit that determines a coring threshold value according to the presence / absence of an edge detected by the edge detection unit for each region that is a unit of coring processing, A coring processing unit that performs coring using the threshold value determined by the coring threshold value determination unit with respect to the signal converted by the wavelet conversion unit, and a signal that is cored by the coring processing unit And a wavelet inverse transform unit for inversely transforming the wavelet.

  With this configuration, the coring threshold is determined according to the presence or absence of an edge, and the coring processing is performed using the determined threshold, so that noise is removed while suppressing blurring of the edge portion, and a clear image is obtained. Can do.

  In the image processing apparatus of the present invention, the coring threshold value determining unit sets a smaller threshold value when an edge is detected by the edge detecting unit than when no edge is detected.

  With this configuration, when an edge is included in the partial image to be processed, a small threshold is set, so that blurring of the edge due to the coring process can be suppressed.

  The image processing method of the present invention includes an image input step for inputting an image, a wavelet transform step for wavelet transforming the image input in the image input step, and an edge included in the image input in the image input step An edge detection step for detecting the coring threshold, a coring threshold value determining step for determining a coring threshold value according to the presence or absence of the edge detected in the edge detection step for each region serving as a unit of coring processing, and the wavelet transform The coring processing step for performing coring on the signal converted in the step using the threshold value determined in the coring threshold determination step, and the wavelet inverse transform on the signal cored in the coring processing step And a wavelet inverse transform step.

  With this configuration, the coring threshold is determined according to the presence or absence of an edge, and the coring processing is performed using the determined threshold, so that noise is removed while suppressing blurring of the edge portion, and a clear image is obtained. Can do.

  A program according to the present invention is a program for performing noise reduction processing of an input image, and includes an image input step for inputting an image to a computer, and a wavelet for wavelet transforming the image input in the image input step. Depending on the presence or absence of an edge detected in the edge detection step for each region that is a unit of coring processing, an edge detection step that detects an edge included in the image input in the image input step A coring threshold determining step for determining a coring threshold, and a coring processing step for performing coring on the signal converted in the wavelet transform step using the threshold determined in the coring threshold determining step. And the coring signal in the coring processing step. To perform a wavelet inverse transform step of over wavelet inverse transform.

  With this configuration, the coring threshold is determined according to the presence or absence of an edge, and the coring processing is performed using the determined threshold, so that noise is removed while suppressing blurring of the edge portion, and a clear image is obtained. Can do.

  According to the present invention, the coring threshold is determined according to the presence or absence of an edge, and the coring process is performed using the determined threshold, so that noise is removed while suppressing blurring of the edge portion, and a clear image is obtained. Obtainable.

Hereinafter, an image processing apparatus according to an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram illustrating an image processing apparatus according to an embodiment. The image processing apparatus includes an image sensor 101 that captures an image, an image memory 102 that stores an image obtained by the image sensor 101, and an image display unit 108 that displays the image.

  The image processing apparatus has a configuration for removing noise included in an image stored in the image memory 102. That is, the image processing apparatus includes a wavelet transform unit 103 that performs wavelet transform on an image read from the image memory 102, a coring processing unit 106 that performs coring processing on the signal after wavelet transform, and a post-coring process. A wavelet inverse transform unit 107 that reconstructs a signal from the signal and the low frequency component obtained from the wavelet transform unit 103 is provided.

  The coring processing unit 106 changes the threshold value used for the coring process according to whether or not an edge is included in the correction target image. The image processing apparatus includes an edge detection unit 104 that detects an edge based on the signal converted by the wavelet conversion unit 103, and a threshold determination unit 105 that determines a coring threshold from the edge detection result of the edge detection unit 104. ing. The coring processing unit 106 performs coring processing using the threshold value determined by the threshold value determination unit 105.

  Next, the wavelet transform unit 103 will be described in detail. The wavelet transform unit 103 performs level 3 or higher wavelet transform on the input image. The level means the number of subband decompositions. That is, the wavelet transform unit 103 performs wavelet transform by performing subband decomposition three times or more.

  FIG. 2 is a diagram for explaining the wavelet transform. In FIG. 2, a level 1 wavelet transform is taken as an example. In the example shown in FIG. 2, the wavelet transform, which is one of the multi-resolution transforms, is used to decompose the observation signal into a low-frequency component and a high-frequency component (subband decomposition) by LPF and HPF, respectively. Subband decomposition of the low-frequency and high-frequency components obtained by subband decomposition into a low-frequency band that includes a low-frequency signal component using a low-pass filter and a high-frequency band that includes a high-frequency signal component using a high-pass filter To do. Thereby, horizontal, vertical, and diagonal high-frequency signal components and low-frequency signal components can be obtained.

  Fig.3 (a)-FIG.3 (c) are figures which show the example of subband decomposition | disassembly. 3A is an original image, and FIG. 3B is a diagram showing each frequency component subjected to subband decomposition. As shown in FIG. 3C, the low-frequency signal component obtained by the subband decomposition (see FIG. 3B) is further subband decomposed to perform higher-level wavelet transform. As a wavelet transform method, for example, a specific method called Haar's base composed of only two filter coefficients, “1” and “−1”, may be used.

  FIG. 4A and FIG. 4B are diagrams showing a specific calculation method for each component subjected to subband decomposition. FIG. 4A is a diagram showing two-dimensional horizontal and vertical two pixels, and FIG. 4B is a diagram showing a specific calculation method. The low-frequency component for two-dimensional horizontal and vertical two pixels [A, B, C, D] can be calculated as LL = A + B + C + D, and each high-frequency component is also HL = A−B + C−D, LH = A + B−C−D, It can be calculated as HH = A−B−C + D. Also, each higher level component can be calculated by calculating the lower level low frequency component LL in the same manner.

  Next, the edge detection unit 104 will be described in detail. The edge detection unit 104 can use a generally known edge detection filter. For example, Prewitt, Solvel, Laplacian, or the like can be used as the edge detection filter.

  Next, the coring threshold determination unit 105 will be described. The coring threshold determination unit 105 has a function of adaptively changing the coring threshold depending on whether or not the target pixel reconstructed by the wavelet inverse transformation includes an edge. The target pixel is an area that is a unit of coring processing.

  FIG. 5 is a diagram illustrating the operation of the coring threshold determination unit 105. The coring threshold value determination unit 105 performs edge detection (S10) and determines whether the target pixel includes an edge (S12). When it is determined that the target pixel does not include an edge (NO in S12), the coring threshold value determination unit 105 applies the standard coring threshold value Thn (S14). When it is determined that the target pixel includes an edge, the coring threshold value determination unit 105 applies a coring threshold value Ths that is smaller than the standard coring threshold value Thn (S16).

  Next, the coring processing unit 106 will be described. The coring processing unit 106 performs coring processing on the high frequency component signal input from the wavelet transform unit 103. Specifically, when the absolute value of the high frequency component signal is smaller than the coring thresholds Thn and Ths determined by the coring threshold determination unit 105, the high frequency component signal is replaced with 0. As a result, the coring processing unit 105 removes noise.

  FIG. 6A is a diagram showing the coring characteristic based on the standard coring threshold Thn, and FIG. 6B is a diagram showing the coring characteristic based on the coring threshold Ths smaller than the standard. When the signal reconstructed by the subsequent wavelet inverse transform unit 107 includes an edge, the coring processing unit 106 applies the small coring threshold Ths to perform the small threshold coring processing (see FIG. 6B). ). Conversely, when it is determined that the signal reconstructed by the wavelet inverse transform unit 107 at the subsequent stage does not include an edge, the standard coring threshold Thn is applied to perform standard threshold coring processing (see FIG. 6A). ).

  Next, the wavelet inverse transform unit 107 will be described. The wavelet inverse transform unit 107 has a function of reconstructing a signal by performing wavelet inverse transform on the high-frequency component subjected to coring processing by the coring processing unit 106 and the low-frequency component obtained by the wavelet transform unit 103.

  FIG. 7 is a diagram for explaining the wavelet inverse transform. In the wavelet inverse transformation, the reconstruction [LPF ′] and [HPF ′] generated from the basis functions used in the subband decomposition process from the LL, LH, HL, and HH components obtained by the band division are used. Reconstruct the signal.

  FIG. 8A and FIG. 8B are diagrams illustrating examples of signal reconstruction. FIG. 8A shows a signal subjected to subband decomposition, and FIG. 8B shows a reconstructed image. A reconstructed image shown in FIG. 8B is obtained by performing wavelet inverse transform on the subband-divided signal as shown in FIG. 8A.

  FIGS. 9A and 9B are diagrams showing a specific calculation method for reconstructing two-dimensional horizontal and vertical two pixels [A, B, C, D] from each component subjected to subband decomposition. It is. FIG. 9A is a diagram showing two-dimensional horizontal and vertical two pixels, and FIG. 9B is a diagram showing a specific calculation method. In the wavelet inverse transform, a level 1 wavelet transform is performed using a Haar basis. As shown in FIG. 9B, A = (LL + HL + LH + HH) / 4, B = (LL-HL + LH-HH) / 4, C = (LL + HL-LH-HH) / 4, D = (LL-HL-LH + HH) / 4 can be calculated.

  Further, by performing inverse wavelet transform on the LL component reconstructed from the upper level and the high-frequency component of the lower level sequentially by a similar calculation method, an image signal with reduced edge blur from which noise has been removed is reconstructed.

  Next, the operation of the image processing apparatus according to the present embodiment will be described. The image processing apparatus stores an image captured by the image sensor 101 in the image memory 102. Next, the wavelet transform unit 103 reads an image stored in the image memory 102 and performs wavelet transform. The wavelet transform unit 103 inputs a signal obtained by the wavelet transform to the edge detection unit 104, the coring processing unit 106, and the wavelet inverse transform unit 107.

  The edge detection unit 104 of the image processing apparatus detects an edge included in the target pixel based on the signal input from the wavelet transform unit 103, and inputs the detection result to the threshold value determination unit 105. The threshold determination unit 105 determines a coring threshold based on the detection result input from the edge detection unit 104. Specifically, the threshold value determination unit 105 applies the standard coring threshold value Thn when an edge is not included in the target pixel, and when the edge is included, the threshold value determination unit 105 applies the standard coring threshold value Thn. Apply a small coring threshold. The threshold value determination unit 105 inputs the determined coring threshold value to the coring processing unit 106.

  The coring processing unit 106 performs a coring process on the signal input from the wavelet transform unit 103 using the coring threshold value input from the threshold value determination unit 105. The coring processing unit 106 inputs the coring-processed signal to the wavelet inverse transform unit 107.

The wavelet inverse transform unit 107 reconstructs a signal by performing wavelet inverse transform on the high-frequency component subjected to coring processing by the coring processing unit 106 and the low-frequency component obtained by the wavelet transform unit 103. The wavelet inverse transform unit 107 inputs the reconstructed image to the image display unit 108. The image display unit 108 displays the image input from the wavelet inverse transform unit 107.
The image processing apparatus and the image processing method of the present embodiment have been described above.

  The image processing apparatus according to the present embodiment determines the coring threshold used in the coring processing unit 106 based on whether or not an edge is included in the target pixel. Therefore, the image processing apparatus according to the present embodiment can remove noise while reducing blurring of edges caused by coring processing.

  For example, when an edge is included in the target pixel, if the high frequency coefficient is replaced with 0 with a standard coring threshold, an adverse effect such as blurring of the edge occurs. In the present embodiment, when an edge is detected by the edge detection unit 104, the coring processing unit 106 performs the coring process using the small coring threshold Ths, and thus blurring of the edge can be suppressed. Conversely, when no edge is detected by the edge detection unit 104, the coring process is performed using the standard coring threshold value Thn, so that noise can be reduced.

  FIG. 10A to FIG. 10C are diagrams showing simulation results of image processing using a one-dimensional signal including edges by the image processing apparatus according to the present embodiment. FIG. 10 (a) shows the original signal, FIG. 10 (b) shows the result of performing the reconfiguration by performing the coring process with the standard coring threshold, and FIG. 10 (c) shows the small coring threshold (for example, The result of reconfiguring by performing coring processing at a value half the standard coring threshold) is shown. As can be seen from FIG. 10B and FIG. 10C, the blur at the edge portion is reduced.

  FIG. 11A to FIG. 11C are diagrams illustrating simulation results of image processing using a one-dimensional signal including noise by the image processing apparatus according to the present embodiment. FIG. 11 (a) shows the original signal, FIG. 11 (b) shows the result of performing the reconfiguration by performing the coring process with the standard coring threshold, and FIG. 11 (c) shows the small coring threshold (for example, The result of reconfiguring by performing coring processing at a value half the standard coring threshold) is shown. Comparing FIG. 11B and FIG. 11C, it can be seen that the noise component is reduced by performing the standard coring process rather than performing the small coring process.

While the image processing apparatus and the image processing method of the present invention have been described in detail with reference to the embodiments, the present invention is not limited to the above-described embodiments.
The present invention also includes a program for causing a computer to execute each step of the image processing method of the above-described embodiment.

  As described above, the present invention has an excellent effect that noise can be removed while suppressing blurring of an edge portion and a clear image can be obtained, and is useful as an image processing apparatus that removes noise. is there.

Block diagram of an image processing apparatus according to an embodiment Diagram for explaining wavelet transform (A) The figure which shows the original image before subband decomposition | disassembly (b) The figure which shows each frequency component by which subband decomposition | disassembly (c) The figure which shows the example by which the low frequency component was subband-decomposed (A) Diagram showing two-dimensional horizontal and vertical two pixels (b) Diagram showing a specific calculation method of each component subjected to subband decomposition The figure which shows the operation | movement of the coring threshold value determination in embodiment (A) The figure which shows the coring characteristic by standard coring threshold Thn (b) The figure which shows the coring characteristic by small coring threshold Ths Diagram for explaining inverse wavelet transform (A) A diagram showing a signal subjected to subband decomposition (b) A diagram showing a reconstructed image (A) Diagram showing two-dimensional horizontal and vertical two pixels (b) Diagram showing a specific calculation method for reconstructing two-dimensional horizontal / vertical two pixels from sub-band decomposed components (A) The figure which shows the original signal in the simulation of this Embodiment (b) The figure which shows the result of having performed the reconfiguration | reconstruction by performing a coring process with a standard coring threshold value (c) The coring process with a small coring threshold value Figure showing the result of reconfiguration (A) The figure which shows the original signal in the simulation of this Embodiment (b) The figure which shows the result of having performed the reconfiguration | reconstruction by performing a coring process with a standard coring threshold value (c) The coring process with a small coring threshold value Figure showing the result of reconfiguration

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Image pick-up element 102 Image memory 103 Wavelet transformation part 104 Edge detection part 105 Coring threshold value determination part 106 Coring process part 107 Wavelet inverse transformation part 108 Image display part

Claims (5)

An apparatus for performing noise reduction processing using wavelet degeneration on an input image,
An edge detector for detecting edges included in the image;
A coring threshold determination unit that determines a coring threshold according to the presence or absence of an edge detected by the edge detection unit for each area that is a unit of coring processing,
An image processing apparatus comprising: An image input unit for inputting an image;
A wavelet transform unit for wavelet transforming the image input in the image input unit;
An edge detection unit for detecting an edge included in the image input by the image input unit;
A coring threshold determination unit that determines a coring threshold according to the presence or absence of an edge detected by the edge detection unit for each area that is a unit of coring processing,
A coring processing unit that performs coring using the threshold value determined by the coring threshold value determination unit for the signal converted by the wavelet conversion unit,
A wavelet inverse transform unit for performing wavelet inverse transform on the signal cored by the coring processing unit;
An image processing apparatus comprising:   The image processing apparatus according to claim 1, wherein the coring threshold value determination unit sets a smaller threshold value when an edge is detected by the edge detection unit than when an edge is not detected. An image input step for inputting an image;
A wavelet transform step for wavelet transforming the image input in the image input step;
An edge detection step for detecting an edge included in the image input in the image input step;
A coring threshold value determining step for determining a coring threshold value according to the presence or absence of the edge detected in the edge detection step for each area that is a unit of the coring process;
A coring processing step for performing coring on the signal converted in the wavelet transform step using the threshold value determined in the coring threshold value determination step;
A wavelet inverse transform step for wavelet inversely transforming the signal that has been cored in the coring process step;
An image processing method comprising: A program for noise reduction processing of an input image.
An image input step for inputting an image;
A wavelet transform step for wavelet transforming the image input in the image input step;
An edge detection step for detecting an edge included in the image input in the image input step;
A coring threshold value determining step for determining a coring threshold value according to the presence or absence of the edge detected in the edge detection step for each area that is a unit of the coring process;
A coring processing step for performing coring on the signal converted in the wavelet transform step using the threshold value determined in the coring threshold value determination step;
A wavelet inverse transform step for wavelet inversely transforming the signal that has been cored in the coring process step;
A program that executes

Images