WO2012073865A1 - Image processing device, image processing method, image processing program, and display device - Google Patents

Abstract

Provided is an image processing device comprising: a scaler unit (2) that outputs up-converted image signals (UV) that have input image signals (Vin) that contain a noise component and that have been converted to a higher resolution in the horizontal direction in a frame image; and noise reduction sections (4-1, 4-2, 4-3) that take in the up-converted image signals (UV) and detect noise-containing pixels in a pixel row in the horizontal direction, and corrects pixel values in noise-containing pixels such that the noise component is reduced.

Description

Image processing apparatus, image processing method, image processing program, and display apparatus

The present invention relates to an image processing device, an image processing method, an image processing program, and a display device.
This application claims priority based on Japanese Patent Application No. 2010-267698 filed in Japan on November 30, 2010, the contents of which are incorporated herein by reference.

When imaging or transmitting a video signal, noise components such as snow noise, Gaussian noise, and shot noise may be mixed in the video signal. For example, in analog television broadcasting, if the electric field strength on the broadcast wave receiving side is weak, the S / N ratio of the received signal decreases. In digital television broadcasting, an analog video source may be digitized and rebroadcast, and the video signal broadcast in this way also includes noise components. There is known a noise reduction circuit that reduces a noise component superimposed on such a video signal (see, for example, Patent Document 1).

In recent years, television receivers compatible with HDTV (High Definition Television) have been used in many cases. Such a television receiver capable of displaying high-definition video displays an SDTV (Standard Definition Television) video, which is an analog video source received by broadcast, after up-converting it into high-definition video. Therefore, even when a standard-definition video is converted into a high-definition video and displayed, there is a high need for reducing noise components included in the video. In a conventional television receiver, a noise component included in an image is reduced by a noise reduction circuit, and an image whose S / N ratio is improved to some extent is up-converted.

Japanese Patent Laid-Open No. 7-250264

However, if the noise reduction process is performed before the up-conversion process, the fineness of the obtained image is lowered. On the other hand, if the noise reduction process is weakened to maintain a fine feeling, noise is conspicuous in the obtained image.
Therefore, the present invention has been made in view of the above circumstances, and provides an image processing device, an image processing method, an image processing program, and a display device that obtain an image with high definition and reduced noise components. With the goal.

[1] In order to solve the above problems, an image processing apparatus according to an aspect of the present invention provides an up-converted image signal obtained by increasing the resolution of an input image signal including a noise component in at least one direction of a frame image. A noise reduction unit that detects a noise-containing pixel including noise from the resolution changing unit to be generated and the up-converted image signal generated by the resolution changing unit and corrects the pixel value of the noise-containing pixel so that the noise component is reduced And a section.

[2] In the image processing device described above, according to one aspect of the present invention, the noise reduction unit includes a pixel value of a target pixel for noise detection included in the up-converted image signal and the up-converted image signal. The noise-containing pixel is detected based on at least the pixel value of the two comparison pixels sandwiching the target pixel in the direction in which the resolution changing unit increases the resolution.

[3] In the above-described image processing device, one aspect of the present invention is a ratio between the resolution in the predetermined direction in the frame image of the input image signal and the resolution in the predetermined direction in the frame image of the up-converted image signal. And a sampling control unit for determining an interval between at least one of the comparison pixels and the target pixel.

[4] In the image processing apparatus described above, one aspect of the present invention is characterized in that the ratio is a real number having a decimal part.

[5] In the image processing device described above, according to one aspect of the present invention, a plurality of the noise reduction units are connected in cascade, and an interval between the two pixels of each of the plurality of noise reduction units is output from an input side. It is characterized by being shortened toward the side.

[6] In the image processing apparatus described above, one aspect of the present invention is characterized in that the at least one direction is a horizontal direction or a vertical direction, or a horizontal direction and a vertical direction in the frame image.

[7] In the image processing device described above, one aspect of the present invention is characterized in that the noise reduction unit includes a 3-tap median filter.

[8] In the image processing device described above, one aspect of the present invention is characterized in that the noise reduction unit includes a bilateral filter.

[9] In order to solve the above problem, in the image processing method according to one aspect of the present invention, the resolution changing unit increases the resolution of the input image signal including the noise component in at least one direction of the frame image. A resolution changing step for outputting an up-converted image signal, and a noise reducing unit detects a noise-containing pixel containing noise from the up-converted image signal output by the resolution changing unit in the resolution changing step, and a noise component is reduced. And a noise reduction step of correcting the pixel value of the noise-containing pixel.

[10] In order to solve the above-described problem, an image processing program according to an aspect of the present invention is an up-conversion in which a computer increases the resolution of an input image signal including a noise component in at least one direction of a frame image. A resolution changing unit that generates an image signal, and detecting a noise-containing pixel containing noise from the up-converted image signal generated by the resolution changing unit, and correcting the pixel value of the noise-containing pixel so that the noise component is reduced And function as a noise reduction unit.

[11] In order to solve the above problem, a display device according to one embodiment of the present invention generates an up-converted image signal obtained by increasing the resolution of an input image signal including a noise component in at least one direction of a frame image. A resolution changing unit that detects noise-containing pixels containing noise from the up-converted image signal generated by the resolution changing unit, and corrects the pixel value of the noise-containing pixels so that the noise component is reduced And.

According to the present invention, it is possible to obtain an image with high definition and reduced noise components.

1 is a block diagram illustrating a schematic functional configuration of an image processing apparatus according to a first embodiment of the present invention. It is a block diagram showing the schematic function structure of the scaler part in the embodiment. It is a figure showing the part containing a noise component among one horizontal line of an image signal. It is a figure showing a mode that the noise component was reduced from the part shown to FIG. 3A. It is a block diagram showing the functional structure of the noise reduction part in the embodiment. It is a block diagram showing the functional structure of the representative value determination part in the embodiment. It is a block diagram showing the functional structure of the reduction calculating part in the embodiment. It is a figure showing the input image signal Vin. It is a figure showing the image signal and noise component which the up sampler 21 of the scaler part 2 obtained by up-converting the horizontal resolution of an original image to 4 times the resolution. 6 is a waveform diagram after an image signal after up-conversion processing is passed through the low-pass filter 22 of the scaler unit 2. FIG. It is a wave form diagram showing a mode that the noise reduction unit 3 reduced the noise component. It is a figure showing the input image signal Vin. It is a wave form diagram after passing an input image signal through a noise reduction unit. It is a figure showing the image signal obtained by the up-sampler up-converting the horizontal resolution of the original image to four times the resolution. FIG. 6 is a waveform diagram after an image signal after up-conversion processing is passed through a low-pass filter. 2 is a block diagram illustrating a functional configuration of a liquid crystal television receiver to which the image processing apparatus according to the embodiment is applied. FIG. FIG. 10 is a block diagram illustrating a functional configuration of a representative value determination unit included in a noise reduction unit in a first modification of the embodiment. It is a block diagram showing the functional composition of the sampling control part contained in the noise reduction part in the 2nd modification of the embodiment. It is a block diagram showing the functional structure of the representative value determination part in the modification. It is a block diagram showing the schematic function structure of the image processing apparatus which is 2nd Embodiment of this invention. In the same embodiment, it is the figure which represented typically the distribution of the spatial frequency versus luminance value of an up-conversion image signal. It is a block diagram showing the functional structure of the noise reduction part in the image processing apparatus which is 3rd Embodiment of this invention.

Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings.
[First Embodiment]
FIG. 1 is a block diagram showing a schematic functional configuration of the image processing apparatus according to the first embodiment of the present invention. As shown in the figure, the image processing apparatus 1 takes in an input image signal Vin, an original image resolution value Hx, and a converted image resolution value Hy supplied from the outside, and outputs an output image signal Vout. The input image signal Vin is an image signal of an image (original image) before the resolution is up-converted. It is assumed that the input image signal Vin includes a noise component (noise component). The original image resolution value Hx is a resolution value in the horizontal direction in the frame image of the original image. The converted image resolution value Hy is a resolution value in the horizontal direction of an image (converted image) after up-converting the resolution in the horizontal direction. The output image signal Vout is an image signal of an image obtained by reducing noise components from the upconverted image signal.

In the following description, reducing the noise component includes removing the noise component. Therefore, the meaning of “reducing noise components” means “removing or reducing noise components”.

The image processing apparatus 1 includes a scaler unit (resolution changing unit) 2 and a noise reduction unit 3. The scaler unit 2 takes in the input image signal Vin, the original image resolution value Hx, and the converted image resolution value Hy. Then, the scaler unit 2 generates an up-converted image signal UV obtained by up-converting the horizontal resolution in the frame image of the input image signal Vin based on the original image resolution value Hx and the converted image resolution value Hy.

The noise reduction unit 3 takes in the up-converted image signal UV, the original image resolution value Hx, and the converted image resolution value Hy, reduces the horizontal noise component contained in the frame image of the up-converted image signal UV, and outputs the output image signal. Vout is output. More specifically, the noise reduction unit 3 includes one or a plurality of noise reduction units 4. However, FIG. 1 is an example in which the noise reduction unit 3 includes three noise reduction units 4 (noise reduction units 4-1, 4-2, 4-3). The noise reduction unit 4 includes a 3-tap median filter and can set the tap interval based on the original image resolution value Hx and the converted image resolution value Hy. In other words, the noise reduction unit 4 can cause the 3-tap median filter to function with tap settings corresponding to the horizontal resolution of the frame image.

When the noise reduction unit 3 includes a plurality of noise reduction units 4, the noise reduction unit 3 causes the plurality of noise reduction units 4 to be cascade-connected (cascade connection), and the tap setting is different for each noise reduction unit 4. . For example, as shown in FIG. 1, the noise reduction units 4-1 4-2, and 4-3 are cascade-connected. Although not shown in the figure, taps are set so that the tap interval is sequentially narrowed in the direction from the first stage noise reduction unit 4-1 to the third stage noise reduction unit 4-3. .

Note that the interval between two pixels sandwiching the target pixel for noise detection in each of the plurality of noise reduction units can be shortened from the input side (for example, the first stage) toward the output side (for example, the third stage). That's fine. Here, the fact that the interval between the two pixels becomes shorter from the input side to the output side includes that the interval between the two pixels becomes shorter monotonically from the input side to the output side. In addition, when divided into sections from the input side to the output side, the fact that the interval between the two pixels decreases from the input side to the output side means that the average of the interval between the two pixels for each section is shortened. Is included.

With this configuration, the spatial frequency band for reducing the noise component is sequentially increased from the noise reduction unit 4-1 to the noise reduction unit 4-3. Therefore, the noise reduction unit 3 can increase the noise component in the spatial frequency and form the noise component in the vicinity of the Nyquist frequency of the up-converted image signal UV that is wider than the Nyquist frequency of the input image signal Vin.

Next, the functional configuration of the scaler unit 2 will be described. FIG. 2 is a block diagram illustrating a schematic functional configuration of the scaler unit 2. As shown in the figure, the scaler unit 2 includes an upsampler 21 and a low-pass filter 22. The upsampler 21 takes in the input image signal Vin, the original image resolution value Hx, and the converted image resolution value Hy. The upsampler 21 calculates a multiple of the horizontal resolution to be up-converted based on the original image resolution value Hx and the converted image resolution value Hy, that is, a multiple of the sampling frequency, and the input image signal Vin according to the multiple. The interpolation (interpolation) process is executed.

For example, when the original image resolution value Hx is “640” and the converted image resolution value Hy is “1920”, the upsampler 21 calculates Hy / Hx = 1920/640 = 3 as a multiple of upsampling. Then, the upsampler 21 executes an interpolation process by interpolating a “0” (zero) value for the input image signal Vin in accordance with “3” that is a multiple of the sampling frequency. The interpolation process here is an upsampling process, and the interpolation process is completed by the filter process by the low-pass filter 22 in the next stage. Therefore, if the sampling frequency of the original image having a horizontal resolution of 640 pixels is fs, the Nyquist frequency of the original image is fs / 2, the sampling frequency of the image after the upsampling process is 3 fs, and the Nyquist frequency is ( 3/2) fs.

The low-pass filter 22 is a low-pass filter that reduces aliasing components generated by the upsampling process of the upsampler 21. Specifically, the low-pass filter 22 mainly passes an image signal having a frequency band equal to or lower than the Nyquist frequency fs / 2 of the original image, while cutting an image signal having a frequency band exceeding the Nyquist frequency fs / 2 of the original image. Filter characteristics. That is, the low-pass filter 22 has a filter characteristic in which the cutoff frequency is the Nyquist frequency fs / 2 of the original image. From the image signal processed by the up-sampler 21, an image signal in a frequency band lower than the Nyquist frequency fs / 2 of the original image is extracted and output as an up-converted image signal UV. As the low-pass filter 22, for example, a known Lanchos filter is applied.

Next, before describing the functional configuration of the noise reduction unit 4, an algorithm of noise reduction processing executed by the noise reduction unit 4 will be described. FIG. 3 is a diagram schematically showing how noise components are reduced from one horizontal line of the image signal up-converted by the scaler unit 2.
3A is a diagram illustrating a portion including a noise component in one horizontal line of an image signal, and FIG. 3B is a diagram illustrating a state in which the noise component is reduced from the portion illustrated in FIG. 3A.
In both FIG. 3A and FIG. 3B, the horizontal axis is the pixel position in the horizontal direction of the image, and the vertical axis is the pixel value. The pixel value is, for example, any one of a luminance value, a color difference value, and a color value.

For example, in a pixel row of one horizontal line in a frame image of an image signal obtained by imaging, the pixel values of adjacent pixels are approximate, and the pixel value of the pixel row is likely to change gently. Therefore, in FIG. 3A, a pixel that suddenly protrudes like a pixel surrounded by a circle is highly likely to be a pixel including a noise component (noise-containing pixel). Therefore, in the present embodiment, the noise reduction unit 4 includes the target pixel for noise detection, the pixel that is n (n is an integer of 1 or more) in the horizontal direction from the target pixel, and the left side in the horizontal direction from the target pixel. The pixel values of three pixels are compared with pixels separated by p (p is an integer of 1 or more), and it is determined whether or not the target pixel is a pixel that is highly likely to contain a noise component.

Specifically, the noise reduction unit 4 has the largest (peak shape) or the smallest (valley shape) pixel value of the target pixel located in the middle of these three pixels. In this case, it is determined that the target pixel is a pixel that is highly likely to include a noise component.
As illustrated in FIG. 3B, the noise reduction unit 4 corrects the pixel value of the target pixel in a direction in which the peak or valley of the target pixel determined to be a pixel that has a high possibility of including a noise component is reduced.

Next, the functional configuration of the noise reduction unit 4 will be described. The noise reduction unit 4 detects a noise-containing pixel containing noise from the up-converted image signal generated by the scaler unit (resolution changing unit) 2 and corrects the pixel value of the noise-containing pixel so that the noise component is reduced. . Here, the noise reduction unit 4 has a pixel value of a target pixel for noise detection included in the up-converted image signal and a pixel value included in the up-converted image signal, and the scaler unit (resolution changing unit) 2 increases the resolution. The noise-containing pixel is detected based at least on the pixel values of the two comparison pixels sandwiching the target pixel in the direction (in this embodiment, the horizontal direction as an example).

FIG. 4 is a block diagram illustrating a functional configuration of the noise reduction unit 4. As shown in the figure, the noise reduction unit 4 takes in the image signal X before noise reduction, the original image resolution value Hx, and the converted image resolution value Hy, and outputs the image signal Y after noise reduction. In the first stage noise reduction unit 4-1 of the noise reduction unit 3, the pre-noise reduction image signal X is the up-converted image signal UV. In the noise reduction unit 4-3 at the final stage (third stage in FIG. 1) of the noise reduction unit 3, the noise-reduced image signal Y is the output image signal Vout.
As shown in the figure, the noise reduction unit 4 includes a sampling control unit 5, a noise detection unit 6, a noise level detection unit 7, and a reduction calculation unit 8.

The sampling control unit 5 takes in the original image resolution value Hx and the converted image resolution value Hy, and calculates a horizontal resolution ratio (Hy / Hx> 1) between the converted image resolution value Hy and the original image resolution value Hx. Then, the sampling control unit 5 obtains a value obtained by multiplying the horizontal resolution ratio by a first constant value previously held, and specifies a tap value for designating p comparison pixels that are separated from the noise detection target pixel on the left side in the horizontal direction. This is supplied to the noise detection unit 6 as Tp. Further, the sampling control unit 5 obtains a value obtained by multiplying the horizontal resolution ratio by a second constant value that is previously held, and noise as a tap value Tn for designating a comparison pixel that is n times away from the target pixel in the horizontal direction. It supplies to the detection part 6.

That is, the sampling control unit 5 up-converts the interval between at least one of the comparison pixels and the noise detection target pixel with the resolution in a predetermined direction (for example, the horizontal direction) in the frame image of the input image signal. This is determined based on the ratio of the image signal frame image to the resolution in the predetermined direction (for example, the horizontal direction).

By configuring the sampling control unit 5 in this way, the tap value Tp and the tap value Tn change in proportion to the horizontal resolution ratio. Thereby, the noise reduction part 4 can perform an appropriate noise reduction process according to horizontal resolution.
The first and second constant values may be the same value or different values. The first and second constant values are determined according to the image quality of the noise reduction target image and the purpose of use of the image processing apparatus 1.

Also, as shown in FIG. 1, in the cascade-connected noise reduction units 4-1, 4-2, 4-3, the first-stage noise reduction unit 4-1 to the third-stage noise reduction unit 4 Each of the noise reduction units 4-1, 4-2, and 4-3 is arranged so that the tap interval is gradually narrowed in the direction toward −3, in other words, the first and second constant values are sequentially reduced. The first and second constant values are set.

The noise detection unit 6 takes in the tap value Tp and the tap value Tn supplied from the sampling control unit 5. Further, the noise detection unit 6 sequentially takes in pixel values from the image signal X before noise reduction.
The noise detection unit 6 sequentially outputs the pixel value Dout of the target pixel for noise detection and the noise determination result Cout for the target pixel.
The noise detection unit 6 includes a delay unit 61, a representative value determination unit 62, and a comparison unit 63 as its functional configuration.

The representative value determining unit 62 sequentially takes in the pixel values of the image signal X before noise reduction, and for each pixel, a pixel for noise detection and a pixel that is n pixels away from the target pixel in the horizontal direction on the right in the horizontal direction. Then, three pixels are selected, which are p pixels away from the target pixel on the left side in the horizontal direction according to the tap value Tp. That is, the interval (distance) between two pixels sandwiching the target pixel is proportional to the horizontal resolution ratio. That is, the interval between two pixels sandwiching the target pixel is the resolution in a predetermined direction (for example, horizontal direction) in the frame image of the input image signal and the resolution in the predetermined direction (for example, horizontal direction) in the frame image of the up-converted image signal. And is determined based on the ratio.
Then, the representative value determining unit 62 selects a pixel having an intermediate level value among the three pixels as a representative pixel, and outputs the pixel value Sout of this representative pixel. A specific method for selecting the representative pixel will be described later.

The delay unit 61 sequentially captures pixel values from the image signal X before noise reduction, delays them for a predetermined time, and supplies them to the comparison unit 63 and the reduction calculation unit 8 as the pixel value Dout of the target pixel for noise detection. The delay unit 61 matches the timing of outputting the pixel value Sout of the representative pixel corresponding to the pixel value after the representative value determining unit 62 takes in the pixel value of the pre-noise reduction image signal X and outputs the pixel value Sout of the target pixel for noise detection. The pixel value Dout is output.

The comparison unit 63 takes in the pixel value Dout of the target pixel for noise detection and the pixel value Sout of the representative pixel, and determines whether or not the target pixel for noise detection is a pixel that is likely to contain a noise component. The noise determination result Cout is output. A specific determination method will be described later.

The noise level detection unit 7 takes in the image signal X before noise reduction, detects a noise component in, for example, a vertical blanking period and / or a horizontal blanking period, and corrects the level value of the detected noise component as a correction value L is supplied to the reduction calculation unit 8.

The reduction calculation unit 8 takes in the pixel value Dout and the noise determination result Cout of the target pixel supplied from the noise detection unit 6 and the correction value L supplied from the noise level detection unit 7, respectively, and outputs a noise-reduced image signal Y is output. The reduction calculation unit 8 corrects the pixel value Dout of the target pixel with the correction value L according to the noise determination result Cout, and outputs the corrected pixel value. Details will be described later.

FIG. 5 is a block diagram showing a functional configuration of the representative value determining unit 62. As shown in the figure, the representative value determining unit 62 includes a delay circuit 621, a first selector 622n, a second selector 622p, and a representative value selecting unit 623.

The delay circuit 621 cascades P (P is an integer of 1 or more) + N (N is an integer of 1 or more) data latches (data holding circuit), and sequentially shifts the pixels of the image signal X before noise reduction. To store pixel values for P + N pixels. The delay circuit 621 has a −N tap at the input terminal of the first stage data latch. Further, the delay circuit 621 has -N + 1 tap, -N + 2 tap,..., -2 tap, -1 tap, 0 tap, 1 tap, etc. at the output terminals of all data latches from the first stage to the last stage. 2 taps,..., P-1 tap, P tap are provided.
N taps from −N tap to −1 tap are connected to an input terminal of the first selector 622n. Further, P taps from 1 tap to P tap are connected to an input terminal of the second selector 622p. The 0 (zero) tap is connected to the representative value selection unit 623.
The pixel value X0 obtained from the 0 (zero) tap is the pixel value of the target pixel for noise detection.

The first selector 622n obtains a pixel value of a pixel that is n pixels away from the target pixel corresponding to the 0 (zero) tap to the right in the horizontal direction, from the taps from the -N tap to the -1 tap of the delay circuit 621. The tap is selected according to the tap value Tn supplied from the sampling control unit 5. By the tap selection based on the tap value Tn, the pixel value Xn is obtained from the first selector 622n.

The second selector 622p selects a tap from which the pixel value of a pixel separated from the target pixel corresponding to the 0 (zero) tap to the left in the horizontal direction is obtained from the taps from the 1 tap to the P tap of the delay circuit 621. The selection is made according to the tap value Tp supplied from the sampling control unit 5. The pixel value Xp is obtained from the second selector 622p by tap selection based on the tap value Tp.

The representative value selection unit 623 includes the pixel value X0 of the target pixel supplied from the 0 (zero) tap of the delay circuit 621 and the pixels supplied from the second selector 622p and separated by p pixels to the left in the horizontal direction. The pixel value Xp and the pixel value Xn of the pixel that is n pixels away from the target pixel in the horizontal direction are supplied. Then, the representative value selection unit 623 selects a pixel having an intermediate level value among these three pixels as a representative pixel, and outputs the pixel value Sout of this representative pixel.
Specifically, the representative value selection unit 623 obtains the pixel value Sout of the representative pixel by calculating the following equation (1).

Next, a method for determining whether or not the noise detection target pixel is a pixel having a high possibility of including a noise component by the comparison unit 63 will be described. The comparison unit 63 takes in the pixel value Dout of the target pixel supplied from the delay unit 61 and the pixel value Sout of the representative pixel supplied from the representative value determination unit 62, and calculates the target pixel by the following equation (2). Is a pixel that is highly likely to contain a noise component, and the noise determination result Cout is output.

According to Expression (2), the comparison unit 63 may cause the pixel value of the target pixel to be increased by a noise component when the pixel value Dout of the target pixel is larger than the pixel value Sout of the representative pixel. Is determined to be high, and the noise determination result Cout = “+ 1” is output. Further, when the pixel value Dout of the target pixel is the same as the pixel value Sout of the representative pixel, the comparison unit 63 determines that the target pixel does not include a noise component, and the noise determination result Cout = “0 (zero)”. Is output. Further, when the pixel value Dout of the target pixel is smaller than the pixel value Sout of the representative pixel, the comparison unit 63 determines that the target pixel is likely to have a small pixel value due to a noise component. The noise determination result Cout = “− 1” is output.

FIG. 6 is a block diagram illustrating a functional configuration of the reduction calculation unit 8. As shown in the figure, the reduction calculation unit 8 includes an adder 81, a subtracter 82, and a selector 83.
The adder 81 adds the pixel value Dout of the target pixel supplied from the delay unit 61 and the correction value L supplied from the noise level detection unit 7.
The subtracter 82 subtracts the correction value L from the pixel value Dout of the target pixel.
The selector 83 is based on the noise determination result Cout (any one of −1, 0, and +1) supplied from the comparison unit 63, and when the noise determination result Cout = “− 1”, the calculation result of the adder 81 When the noise determination result Cout = “0 (zero)”, the pixel value Dout of the target pixel is obtained. When the noise determination result Cout = “+ 1”, the calculation result of the subtractor 82 is obtained. Output as Y pixel.

That is, the reduction calculation unit 8 lowers the level by subtracting the correction value L for the pixel value Dout of the target pixel whose pixel value is likely to be large due to the noise component. Further, the correction value L is added to the pixel value Dout of the target pixel which is likely to have a small pixel value due to the noise component, and the level is increased. In addition to these two cases, the pixel value Dout of the target pixel is not corrected.

The noise reduction effect of the image processing apparatus 1 according to the present embodiment will be described. 7A to 7D are waveform diagrams schematically showing how noise components of an image signal up-converted by the image processing apparatus 1 are reduced. In each waveform diagram from FIG. 7A to FIG. 7D, the horizontal axis is the spatial frequency, and the vertical axis is the luminance value.
FIG. 7A is a diagram illustrating the input image signal Vin. In FIG. 7A, the luminance component 301 and the noise component 302 of the image are distributed in a spatial frequency region lower than the Nyquist frequency fo / 2.

FIG. 7B is a diagram illustrating an image signal and a noise component obtained by the up-sampler 21 of the scaler unit 2 up-converting the horizontal resolution of the original image to four times the resolution. In FIG. 7B, the luminance component (solid line) and the noise component (broken line) of the image are distributed to a spatial frequency region lower than the Nyquist frequency fu / 2 (= 2fo) of the image signal after the up-conversion processing.

FIG. 7C is a waveform diagram after the image signal after the up-conversion processing is passed through the low-pass filter 22 of the scaler unit 2. When the image signal and the noise component after the up-conversion process are passed through the low-pass filter 22 having the filter characteristic 303 whose cutoff frequency is the Nyquist frequency fo / 2 of the original image, the result is as shown in FIG. 7C. At this stage, the noise component is reduced in a frequency region higher than the Nyquist frequency fo / 2, but the noise component remains in the vicinity of the Nyquist frequency fo / 2 and in a lower frequency region.

FIG. 7D is a waveform diagram illustrating a state in which the noise reduction unit 3 reduces the noise component. As shown in FIG. 7D, when the noise reduction unit 3 executes the noise reduction process, the noise component distributed in the low frequency region becomes a noise component 304.
Since the noise component 304 is formed in the vicinity of the Nyquist frequency fu / 2 of the image after the up-conversion processing, it is difficult for the observer to perceive the noise component 304. Further, since the edge component of the image is formed in the vicinity of the Nyquist frequency fu / 2, it is possible to obtain the output image signal Vout having a high definition feeling by enhancing the edge of the image.

8A to 8D are diagrams schematically showing waveforms of an image signal and a noise component in the case of a conventional method in which up-conversion processing is performed after performing noise reduction processing.
FIG. 8A is a diagram illustrating the input image signal Vin. In FIG. 8A, the luminance component 301 and the noise component 302 of the image are distributed in the spatial frequency region lower than the Nyquist frequency fo / 2.
FIG. 8B is a waveform diagram after the input image signal is passed through the noise reduction unit. As shown in FIG. 8B, the noise component becomes a noise component 305 by increasing the frequency in the vicinity of the Nyquist frequency fo / 2.

FIG. 8C is a diagram illustrating an image signal obtained by the up-sampler up-converting the horizontal resolution of the original image to four times the resolution. In FIG. 8C, the luminance component and noise component of the image are distributed to a spatial frequency region lower than the Nyquist frequency fu / 2 (= 2fo) of the image signal after the up-conversion processing.
FIG. 8D is a waveform diagram after the image signal after the up-conversion processing is passed through the low-pass filter. When the image signal and the noise component after the up-conversion processing are passed through a low-pass filter having a filter characteristic whose cutoff frequency is the Nyquist frequency fo / 2 of the original image, the result is as shown in FIG. 8D. Thus, a noise component remains in the vicinity of the Nyquist frequency fo / 2, and the image is perceived as an image with a noticeable granularity to the observer's eyes.

Next, an example in which the image processing apparatus 1 according to this embodiment is applied to a liquid crystal television receiver (display device) will be described.
FIG. 9 is a block diagram illustrating a functional configuration of a liquid crystal television receiver to which the image processing apparatus 1 according to the present embodiment is applied.
As shown in the figure, a liquid crystal television receiver 10 includes a high-definition multimedia interface (HDMI) receiver 100, a disk drive 102, a tuner 103, an IP (Internet Protocol) broadcast tuner 104, and a satellite broadcast tuner 105. An OSD (On Screen Display) generation unit 106, a video selector 107, a video processing circuit 108, an LCD (Liquid Crystal Display) controller 109, an LCD 110, an audio selector 111, and an audio processing unit 112. An amplifier 113, a speaker 114, a network interface (I / F) 115, a ROM 116, a RAM 117, a CPU 118, and the infrared light receiving unit 1. 19, HDMI input terminals 11a, 11b, and 11c, an external video input terminal 101a, and an external audio input terminal 101b.

The CPU 118 reads out and expands the control program stored in the ROM 116 to the RAM 117, and controls each part of the liquid crystal television receiver 10 by executing each step of the program code.
The HDMI receiver 100 takes in an HDMI signal supplied from an external device via any one of the HDMI input terminals 11a, 11b, and 11c, and mainly separates the HDMI signal into a video signal and an audio signal.
The disc drive 102 reproduces content recorded on an optical disc such as a DVD (Digital Versatile Disc) or BD (Blu-ray Disc) or a magnetic hard disk.

The tuner 103 extracts a desired television broadcast program content from the broadcast wave of the television broadcast received by the receiving antenna.
The IP broadcast tuner 104 receives an IP broadcast transmitted by an Internet service provider or the like via the network I / F 115.
The satellite broadcast tuner 105 extracts a desired satellite broadcast program content from the satellite broadcast radio wave received by the receiving antenna.

The video selector 107 receives video signals supplied from the HDMI receiver 100, the disk drive 102, the tuner 103, the IP broadcast tuner 104, the satellite broadcast tuner 105, and an external device connected to the external video input terminal 11a. One of the video signals is fetched and supplied to the video processing circuit 108 in accordance with the selection signal notified from the CPU 118.

The audio selector 111 receives audio signals supplied from the HDMI receiver 100, the disk drive 102, the tuner 103, the IP broadcast tuner 104, the satellite broadcast tuner 105, and an external device connected to the external audio input terminal 101b. One of the audio signals is fetched and supplied to the audio processing unit 112 according to the selection signal notified from the CPU 118.

The OSD generation unit 106 receives and supplies data such as date and time information and system setting menu information from the CPU 118, and generates an on-screen information signal.

The video processing circuit 108 takes in the video signal supplied from the video selector 107 and executes image processing based on the control data supplied from the CPU 118.
In the present embodiment, the image processing apparatus 1 is incorporated in the video processing circuit 108. Specifically, for example, the video processing circuit 108 takes in the video signal supplied from the video selector 107 and sequentially scans the frame image of the video signal based on the information related to scanning supplied from the CPU 118. The video processing circuit 108 receives and takes in the original image resolution value Hx and the converted image resolution value Hy from the CPU 118, and obtains the horizontal line data, the original image resolution value Hx, and the converted image resolution value obtained in accordance with the sequential scanning. Hy is supplied to the image processing apparatus 1 to perform the up-conversion process and the noise reduction process as described above.

The on-screen information signal generated by the OSD generation unit 106 is superimposed on the image signal after the image processing output from the image processing circuit 108, and the image signal on which the on-screen information signal is superimposed is supplied to the LCD controller 109. .
The LCD controller 109 causes the LCD 110 to display a video signal on which the on-screen information signal is superimposed.

Further, the audio processing unit 112 takes in the audio signal supplied from the audio selector 111 and executes audio processing based on the control data supplied from the CPU 118.
The amplifier 113 amplifies the sound signal processed by the sound processing unit 112 and outputs it as sound from the speaker 114.
The infrared light receiver 119 receives an optical remote control signal emitted from a remote control device (not shown) for remotely operating the liquid crystal television receiver 10, decodes the optical remote control signal, extracts key data, and notifies the CPU 118 of the key data.

According to the first embodiment, the image processing apparatus 1 includes the horizontal component included in the up-converted image signal UV at the subsequent stage of the scaler unit 2 that up-converts the horizontal resolution of the input image signal Vin and outputs the up-converted image signal UV. The noise reduction unit 3 for reducing the direction noise component is provided. With this configuration, the noise reduction unit 3 can form a horizontal noise component included in the up-converted image signal in the vicinity of the Nyquist frequency fu / 2. Therefore, according to the first embodiment, since noise components are collected in a spatial frequency band that is difficult to be perceived by the observer's eyes, the image processing apparatus 1 obtains an output image signal Vout with reduced horizontal noise components. Can do.
Further, since the edge component of the image is formed in the vicinity of the Nyquist frequency fu / 2, the image processing apparatus 1 can obtain the output image signal Vout having a high definition feeling by enhancing the edge of the image.

[First Modification of First Embodiment]
In the first embodiment described above, the image processing apparatus 1 that performs the up-conversion processing of the horizontal resolution of the frame image and the noise reduction processing in the pixel row in the horizontal direction has been described. In the first modified example of the first embodiment, an image processing apparatus that performs up-conversion processing of a vertical resolution of a frame image and noise reduction processing in a vertical pixel row will be described.
In this modification, for the configuration that needs to be differentiated from the configuration in the first embodiment, a new code is added or “′ (single quotation mark)” is added to the code of the configuration in the first embodiment. To distinguish. In this modification, the description of the same configuration as that in the first embodiment is omitted.

In this modification, instead of the original image resolution value Hx and the converted image resolution value Hy in the first embodiment, the original image resolution value Vx that is the vertical resolution value of the original image and the vertical resolution are up-converted. A converted image resolution value Vy that is a resolution value in the vertical direction of the subsequent image is used.

The scaler unit 2 ′ included in the image processing apparatus 1 ′ has the same configuration as the scaler unit 2 in the first embodiment, except that the up-conversion direction is the vertical direction in the frame image. Therefore, the specific description here is omitted.

The representative value determination unit included in the noise reduction unit 4 ′ will be described. FIG. 10 is a block diagram illustrating a functional configuration of a representative value determination unit included in the noise reduction unit 4 ′. As shown in the figure, the representative value determining unit 62a has a configuration in which the delay circuit 621 is changed to the delay circuit 621a from the representative value determining unit 62 in the first embodiment.
The delay circuit 621a is obtained by changing P + N data latches included in the delay circuit 621 to P + N delay lines. Each delay line has a delay amount corresponding to one horizontal line in the frame image of the up-converted image signal. This delay line may be configured by cascading data latches corresponding to the number of pixels of one horizontal line.

The delay unit 61 ′ included in the noise reduction unit 4 ′ receives the pixel value Sout of the representative pixel corresponding to the pixel value after the representative value determination unit 62a captures the pixel value of the image signal X before noise reduction. In addition, the pixel value Dout of the target pixel for noise detection is output.

According to the first modification of the first embodiment, the image processing apparatus 1 performs the up-conversion image signal in the subsequent stage of the scaler unit 2 that up-converts the vertical resolution of the input image signal Vin and outputs the up-conversion image signal UV. The noise reduction unit 3 for reducing the noise component in the vertical direction included in the UV is provided. With this configuration, the noise reduction unit 3 can form a noise component in the vertical direction included in the up-converted image signal in the vicinity of the Nyquist frequency fu / 2. Therefore, according to the first modification of the first embodiment, since the noise components are collected in a spatial frequency band that is difficult to be perceived by the observer's eyes, the image processing apparatus 1 outputs an output image in which the noise components in the vertical direction are reduced. A signal Vout can be obtained.
Further, since the edge component of the image is formed in the vicinity of the Nyquist frequency fu / 2, the image processing apparatus 1 can obtain the output image signal Vout having a high definition feeling by enhancing the edge of the image.

[Second Modification of First Embodiment]
The first embodiment described above is an example in which the horizontal resolution ratio (Hy / Hx> 1) between the converted image resolution value Hy and the original image resolution value Hx is an integer. In the second modification of the first embodiment, a case where the horizontal resolution ratio is a real number having a fractional part (a case of a non-integer) will be described.
In this modification, for the configuration that needs to be differentiated from the configuration in the first embodiment, a new code is added, or “” (double quotation mark) is added to the code of the configuration in the first embodiment. In this modification, the description of the same configuration as that in the first embodiment is omitted.

In the present modification, when the original image resolution value Hx and the converted image resolution value Hy with which the horizontal resolution ratio is a non-integer are supplied to the noise reduction unit 4 ″, the noise reduction unit 4 ″ uses the first horizontal resolution ratio. Two pixels that are p and p + 1 away from the noise detection target pixel in the horizontal direction are selected using the closest integer values that are close to each other across the tap value Tp obtained by multiplying the constant values of. Then, the noise reduction unit 4 ″ performs weighting according to the decimal part of the tap value Tp to each of the pixel value of the p-th pixel and the pixel value of the p + 1-th pixel, and adds the two values to each other. Xp is obtained.

Similarly, the noise reduction unit 4 ″ performs horizontal detection from the target pixel for noise detection by using, as tap values, integer values closest to each other across the tap value Tn obtained by multiplying the horizontal resolution ratio by the second constant value. Two pixels that are separated from each other by n and n + 1 are selected on the right side of the direction. The noise reduction unit 4 ″ selects a tap value for each of the pixel value of the nth pixel and the pixel value of the n + 1th pixel. Weighting is performed according to the fractional part of Tn, and both are added to obtain a pixel value Xn.

FIG. 11 is a block diagram showing the functional configuration of the sampling control unit included in the noise reduction unit 4 ″. Here, as an example of specific numerical values, the original image resolution value Hx = “600”, the converted image resolution The configuration and operation of the sampling control unit 5a shown in the figure will be described using an example in which the value Hy = "1920", the first constant value Cp = "2", and the second constant value Cn = "3". .

In the case of the above example, the sampling control unit 5a calculates the horizontal resolution ratio Hy / Hx = 1920/600 = 3.2, and sets the tap value Tp = Cp × Hy / Hx = 2 × 3.2 = 6.4. calculate.
Next, the sampling control unit 5a calculates the nearest integer value Tpi = 6 smaller than the tap value Tp = 6.4 and the nearest integer value Tpi + 1 = 7 larger than the tap value Tp = 6.4. To do.
Further, the sampling control unit 5a calculates the fractional part Tpf = Tp−Tpi = 6.4-6 = 0.4 of the tap value Tp, and 1−Tpf = Tpi + 1−Tp = 7−6.4 = 0.6. Calculate Tpf represents the distance from the tap value Tp to Tpi, and 1-Tpf represents the distance from the tap value Tp to Tpi + 1.
The sampling control unit 5a supplies the calculated Tpi, Tpi + 1, Tpf, and 1-Tpf to the representative value determination unit 62b described later.

Similarly, the sampling control unit 5a calculates a tap value Tn = Cn × Hy / Hx = 3 × 3.2 = 9.6.
Next, the sampling control unit 5a calculates the nearest integer value Tni = 9 smaller than the tap value Tn = 9.6 and the nearest integer value Tni + 1 = 10 larger than the tap value Tn = 9.6. To do.
Further, the sampling control unit 5a calculates the fractional part Tnf = Tn−Tni = 9.6−9 = 0.4 of the tap value Tn, and 1−Tnf = Tni + 1−Tn = 10−9.6 = 0.4. Calculate Tnf represents the distance from the tap value Tn to Tni, and 1-Tnf represents the distance from the tap value Tn to Tni + 1.
The sampling control unit 5a supplies the calculated Tni, Tni + 1, Tnf, and 1-Tnf to the representative value determination unit 62b.

FIG. 12 is a block diagram showing a functional configuration of the representative value determining unit 62b. As shown in the figure, the representative value determining unit 62b has a configuration in which the first selector 622n and the second selector 622p are changed to the first selector 622nb and the second selector 622pb in the representative value determining unit 62 in the first embodiment. Have.

The first selector 622nb takes in Tni, Tni + 1, Tnf, and 1-Tnf supplied from the sampling control unit 5a. Then, the first selector 622 nb is 2 2 apart from the target pixel corresponding to 0 (zero) tap to the right in the horizontal direction from the taps from −N tap to −1 tap of the delay circuit 621. Two taps from which the pixel value of the pixel is obtained are selected according to Tni and Tni + 1.
The first selector 622nb performs weighting by multiplying the pixel value obtained from the delay circuit 621 by tap selection based on Tni by 1-Tnf, and sets Tnf to the pixel value obtained from the delay circuit 621 by tap selection based on Tni + 1. Multiply weights.
Then, the first selector 622nb adds the weighted pixel values and supplies the pixel value Xn to the representative value selection unit 623.

Specifically, the first selector 622nb takes in, for example, Tni = 9, Tni + 1 = 10, Tnf = 0.6, and 1-Tnf = 0.4 supplied from the sampling control unit 5a. The first selector 622nb can obtain two pixel values of 9 pixels and 10 pixels separated from the target pixel corresponding to the 0 (zero) tap on the right side in the horizontal direction from the taps of the delay circuit 621. Taps are selected according to Tni = 9 and Tni + 1 = 10.
The first selector 622nb performs weighting by multiplying the pixel value obtained from the delay circuit 621 by tap selection based on Tni = 9 by 1-Tnf = 0.4, and from the delay circuit 621 by tap selection based on Tni + 1. Weighting is performed by multiplying the obtained pixel value by Tnf = 0.6.
The first selector 622nb adds the weighted pixel values (3.6 + 6.0), and supplies the pixel value Xp = 9.4 to the representative value selection unit 623.

Further, the second selector 622pb takes in Tpi, Tpi + 1, Tpf, and 1-Tpf supplied from the sampling control unit 5a. The second selector 622pb includes two pixels that are p pixels and p + 1 pixels horizontally apart from the target pixel corresponding to the 0 (zero) tap among the taps from the 1 tap to the N tap of the delay circuit 621. Two taps from which pixel values are obtained are selected according to Tpi and Tpi + 1.
The second selector 622pb performs weighting by multiplying the pixel value obtained from the delay circuit 621 by tap selection based on Tpi by 1-Tpf, and sets Tpf to the pixel value obtained from the delay circuit 621 by tap selection based on Tpi + 1. Multiply weights.
Then, the second selector 622pb adds the weighted pixel values and supplies the pixel value Xp to the representative value selection unit 623.

Specifically, the second selector 622pb takes in, for example, Tpi = 6, Tpi + 1 = 7, Tpf = 0.4, and 1-Tpf = 0.6 supplied from the sampling control unit 5a. The second selector 622pb can obtain two pixel values of two pixels that are 6 pixels and 7 pixels horizontally apart from the target pixel corresponding to the 0 (zero) tap among the taps of the delay circuit 621. Taps are selected according to Tpi = 6 and Tpi + 1 = 7.

The second selector 622pb performs weighting by multiplying the pixel value obtained from the delay circuit 621 by the tap selection based on Tpi = 6 by 1-Tpf = 0.6, and from the delay circuit 621 by the tap selection based on Tpi + 1. Weighting is performed by multiplying the obtained pixel value by Tpf = 0.4.
The second selector 622pb adds the weighted pixel values (3.6 + 2.8) and supplies the pixel value Xn = 6.4 to the representative value selection unit 623.

[Second Embodiment]
In the first embodiment, the noise reduction unit 3 is an example in which the original image resolution value Hx and the converted image resolution value Hy are received from the outside (for example, the CPU 118) and fetched to determine the tap value Tp and the tap value Tn. It was.

However, for example, when the disk drive 102 in the liquid crystal television receiver 10 shown in FIG. 9 outputs video content obtained by increasing the resolution of the original video content having a low resolution, the noise reduction unit 3 of the first embodiment is Since the resolution value of the original original video content is not given, an appropriate tap value Tp and tap value Tn cannot be determined for the original image.

Therefore, in the second embodiment of the present invention, an example will be described in which the noise reduction unit determines an appropriate tap value Tp and tap value Tn for the original image in the supplied up-convert image signal UV.

FIG. 13 is a block diagram illustrating a schematic functional configuration of an image processing apparatus according to the second embodiment of the present invention. In the figure, the same components as those in the first embodiment are denoted by the same reference numerals, and the description thereof is omitted.
As shown in the figure, the image processing device 1a has a configuration in which the noise reduction unit 3 is changed to a noise reduction unit 3a with respect to the image processing device 1 in the first embodiment.

The noise reduction unit 3a takes in the up-converted image signal UV supplied from the scaler unit 2. The noise reduction unit 3a then analyzes the frequency component of the up-converted image signal UV to estimate the original original image resolution value H (hat) x and the converted image resolution value H (hat) y. A horizontal noise component included in the up-converted image signal UV is reduced using the original original image resolution value H (hat) x and the converted image resolution value H (hat) y, and an output image signal Vout is output.

Specifically, the noise reduction unit 3a has a configuration further including a resolution estimation unit 9 with respect to the noise reduction unit 3 in the first embodiment.
The resolution estimator 9 captures the up-converted image signal UV, analyzes the spatial frequency versus pixel value of the up-converted image signal UV, and obtains the original original image resolution value H (hat) x and the converted image resolution value H ( Hat) y is estimated.

Specifically, for example, the resolution estimation unit 9 detects the sampling frequency fs of the up-converted image signal UV and calculates the Nyquist frequency fn = fs / 2. Next, the resolution estimation unit 9 detects the upper limit value of the spatial frequency band in which the luminance value is distributed, for example, from the distribution of the spatial frequency versus the luminance value of the up-converted image signal UV. For example, when the spatial frequency versus luminance value of the up-converted image signal UV has the distribution shown in FIG. 14, the resolution estimation unit 9 detects the upper limit value of the spatial frequency band in which the luminance value is distributed as fn / 3. Next, the resolution estimation unit 9 calculates the estimated horizontal resolution ratio H (hat) y / H (hat) x as fn / (fn / 3) = 3.

The resolution estimating unit 9 detects a horizontal line from the up-converted image signal UV and obtains a converted image resolution value H (hat) y. Next, the resolution estimation unit 9 calculates the original original image resolution value H (hat) x based on the calculated estimated horizontal resolution ratio H (hat) y / H (hat) x. In the example shown in FIG. 14, the original original image resolution value H (hat) x = H (hat) y / 3 is calculated. Next, the resolution estimation unit 9 supplies the original original image resolution value H (hat) x and the converted image resolution value H (hat) y to the noise reduction unit 4.

According to the second embodiment, in addition to the effect achieved by the first embodiment, the noise reduction unit 3a can select an appropriate tap value for the original original image in the up-converted image signal UV supplied from the scaler unit 2. Tp and tap value Tn can be determined.

[Third Embodiment]
The first embodiment and the second embodiment are examples in which the noise reduction unit 4 detects and corrects a pixel that is likely to contain a noise component by a 3-tap median filter. The third embodiment of the present invention is another example of a filter circuit that performs noise reduction processing.
The present embodiment is an example in which a bilateral filter is applied as a filter circuit that performs noise reduction processing.

FIG. 15 is a block diagram illustrating a functional configuration of a noise reduction unit in the image processing apparatus according to the third embodiment of the present invention. In the figure, the same components as those in the first embodiment are denoted by the same reference numerals, and the description thereof is omitted.
When the bilateral filter is limited to one-dimensional three pixels, it is expressed as the following equation (3). However, σ ₁ and σ ₂ are preset parameters that determine the range of pixels to be processed. As σ ₁ and σ ₂ increase, the range of the pixel to be processed increases.

The bilateral filter is a non-linear filter. By disposing the bilateral filter in the subsequent stage of the scaler unit 2, the noise component can be increased in the same manner as in the first embodiment.

As shown in FIG. 15, the noise reduction unit 4 a includes a bilateral filter represented by Expression (3). In the noise reduction unit 4a, the first selector 622n outputs the pixel value Xn obtained from the delay circuit 621 by tap selection based on the tap value Tn as f (i + 1). Further, the delay circuit 621 outputs the pixel value X0 as f (i + 0) from the 0 (zero) tap. Further, the second selector 622p outputs the pixel value Xp obtained from the delay circuit 621 by tap selection based on the tap value Tp as f (i−1).

The noise reduction unit 4a executes a calculation process based on Expression (3) based on the pixel values f (i−1), f (i), and f (i + 1), and outputs a noise-reduced image signal Y.

According to the third embodiment, the noise reduction unit 3 including the noise reduction unit 4a can form a noise component included in the up-converted image signal in a harmonic band that is difficult to be perceived by the observer. Therefore, according to the third embodiment, similarly to the first embodiment, the image processing apparatus 1 can obtain the output image signal Vout in which the noise component in the horizontal direction is reduced.
Further, since the edge component of the image is formed in the harmonic band, the image processing apparatus 1 can obtain the output image signal Vout having a high-definition feeling by enhancing the edge of the image, as in the first embodiment. it can.

[Other variations]
The image processing apparatuses 1 and 1a may be configured to execute both the horizontal resolution up-conversion process and noise reduction process, and the vertical resolution up-conversion process and noise reduction process.
That is, the scaler unit (resolution changing unit) 2 generates an up-converted image signal obtained by increasing the resolution of an input image signal including a noise component in at least one direction of a frame image. Here, it is assumed that at least one direction is determined in advance by a designer, and is a horizontal direction or a vertical direction, or a horizontal direction and a vertical direction in the frame image.

Moreover, in each embodiment mentioned above, the scaler part 2 was taken as the example which performs an interpolation process. The scaler unit 2 may execute up-conversion processing of at least one of the horizontal resolution and the vertical resolution by other methods. For example, the scaler unit 2 may execute a resolution up-conversion process by applying a known pixel interpolation technique such as a linear interpolation method or a cubic convolution interpolation method.

In each embodiment, the noise reduction units 4, 4 ″, 4 a are provided with a delay unit 61, and the noise reduction unit 4 ′ is provided with a delay unit 61 ′. For example, in the case of the noise reduction unit 4, the pixel value Dout of the noise detection target pixel may be output from the 0 (zero) tap of the representative value determination unit 62 without including 61 and 61 ′.

In addition, although the representative value determination units 62, 62a, and 62b are configured to select the median value as the representative value, other than this, for example, an average value of pixel values of three pixels may be obtained as the representative value.

In addition, a part of the functions of the image processing apparatus in the first embodiment and each modification described above, and in the second embodiment and the third embodiment may be realized by a computer. In this case, the image processing program for realizing the control function is recorded on a computer-readable recording medium, and the image processing program recorded on the recording medium is read by the computer system and executed. Also good. Here, the “computer system” includes an OS (Operating System) and hardware of peripheral devices. The “computer-readable recording medium” refers to a portable recording medium such as a flexible disk, a magneto-optical disk, an optical disk, and a memory card, and a storage device such as a magnetic hard disk built in the computer system. Furthermore, the “computer-readable recording medium” dynamically holds a program for a short time like a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line. In this case, it may include a device that holds a program for a certain period of time, such as a volatile memory inside a computer system serving as a server device or a client. Further, the above program may be for realizing a part of the functions described above, or may be realized by a combination with the program already recorded in the computer system. .

The embodiment of the present invention has been described in detail above with reference to the drawings. However, the specific configuration is not limited to the embodiment, and includes design and the like within the scope not departing from the gist of the present invention.

1 Image processing device 2 Scaler unit (resolution change unit)
3,3a Noise reduction unit 4,4-1,4-2,4-3,4a Noise reduction unit 5,5a Sampling control unit 6 Noise detection unit 7 Noise level detection unit 8 Reduction calculation unit 9 Resolution estimation unit 10 Liquid crystal television John Receiver 21 Upsampler 22 Low-pass filter 61 Delay unit 62, 62a, 62b Representative value determining unit 63 Comparison unit 81 Adder 82 Subtractor 83 Selector 108 Video processing circuit 621, 621a Delay circuit 622n, 622nb First selector 622p, 622pb Second selector 623 representative value selection unit

Claims (11)

1. A resolution changing unit that generates an up-converted image signal obtained by increasing the resolution of an input image signal including a noise component in at least one direction of a frame image;
   Detecting a noise-containing pixel containing noise from the up-converted image signal generated by the resolution changing unit, and correcting a pixel value of the noise-containing pixel so as to reduce a noise component;
   An image processing apparatus comprising:
2. The noise reduction unit includes a pixel value of a target pixel for noise detection included in the up-converted image signal and a pixel value included in the up-converted image signal in the direction in which the resolution changing unit increases the resolution. The image processing apparatus according to claim 1, wherein the noise-containing pixel is detected based at least on a pixel value of two comparison pixels sandwiching the pixel.
3. Based on the ratio between the resolution in the predetermined direction in the frame image of the input image signal and the resolution in the predetermined direction in the frame image of the up-converted image signal, at least one comparison pixel of the comparison pixels and the target The image processing apparatus according to claim 2, further comprising a sampling control unit that determines an interval between the pixels.
4. 4. The image processing apparatus according to claim 3, wherein the ratio is a real number having a decimal part.
5. A plurality of the noise reduction units are connected in cascade,
   5. The image processing apparatus according to claim 2, wherein an interval between the two pixels of each of the plurality of noise reduction units is shortened from the input side toward the output side.
6. The image processing apparatus according to any one of claims 1 to 5, wherein the at least one direction is a horizontal direction or a vertical direction, or a horizontal direction and a vertical direction in the frame image.
7. The image processing apparatus according to any one of claims 1 to 6, wherein the noise reduction unit includes a 3-tap median filter.
8. The image processing apparatus according to claim 1, wherein the noise reduction unit includes a bilateral filter.
9. A resolution changing step for generating an up-converted image signal obtained by increasing the resolution of the input image signal including the noise component in at least one direction in the frame image;
   A noise reduction unit detects a noise-containing pixel containing noise from the up-converted image signal generated by the resolution change unit in the resolution changing step, and corrects the pixel value of the noise-containing pixel so that the noise component is reduced Noise reduction step to
   An image processing method comprising:
10. Computer
    A resolution changing unit that generates an up-converted image signal obtained by increasing the resolution of an input image signal including a noise component in at least one direction of a frame image;
    Detecting a noise-containing pixel containing noise from the up-converted image signal generated by the resolution changing unit, and correcting a pixel value of the noise-containing pixel so as to reduce a noise component;
    Image processing program to function as
11. A resolution changing unit that generates an up-converted image signal obtained by increasing the resolution of an input image signal including a noise component in at least one direction of a frame image;
    Detecting a noise-containing pixel containing noise from the up-converted image signal generated by the resolution changing unit, and correcting a pixel value of the noise-containing pixel so as to reduce a noise component;
    A display device comprising:

Images