JP2011234316A - Image encoding apparatus, image decoding apparatus and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reproduce a high-quality image closest to a source image from an image obtained by downsampling the source image.SOLUTION: A downsampling section 10 performs wavelet decomposition on and downsamples a source image F to produce a low-resolution downsampling image CA, and an encoding section 12 encodes the downsampling image CAproduced by the downsampling section 10. A variance value determining section 11 determines variance values σ, σ, σof a Gaussian filter for producing an image closest to the source image in upsampling processing at a receiving side based on downsampling images CA, CAproduced by the downsampling section 10. The receiving side performs spatial frequency upsampling using the variance values σ, σ, σto produce a high-quality image closest to the source image F.

Description

  The present invention relates to an image encoding device, an image decoding device, and a program suitable for use in a device that transmits high-definition images.

  In recent years, high-definition digital images have been advanced in the field of image coding. Since a high-definition image has an enormous amount of information, it needs to be highly compressed and encoded in order to transmit in a limited band. However, there is a limit to processing and transmitting a high-definition image having an enormous amount of information with existing encoding. Therefore, a method of performing a downsampling process by linear LPF (Low Pass Filter) and thinning has been devised.

  In this method, the image encoding device on the transmission side encodes and transmits the downsampled image. The image decoding apparatus on the receiving side receives and decodes the encoded down-sampled image, and then performs up-sampling processing using a linear filter, a Laplacian pyramid, wavelet decomposition, nonlinear processing, and the like to reproduce the original original image. Thereby, the original high-definition image can be reproduced without encoding and transmitting a high-definition image having an enormous amount of information. As a spatial frequency upsampling method using a basic wavelet, for example, the one disclosed in Patent Document 1 is known.

Special table 2007-504523

  However, when the original image is reproduced by performing the spatial frequency down-sampling process on the spatial frequency down-sampled image, there is a problem that it is difficult to reproduce an image having a high quality comparable to the original image.

  Therefore, the present invention has been made to solve such a problem. The purpose of the present invention is to obtain a high-level image that is closest to the original image on the receiving side even if an image obtained by spatial frequency down-sampling the original image is transmitted. An object of the present invention is to provide an image encoding device, an image decoding device, and a program capable of reproducing an image with high image quality.

  In order to achieve the above object, an image encoding apparatus according to claim 1 is generated by a downsampling unit that performs a spatial frequency downsampling process on an original image and generates a spatial frequency downsampling image, and the downsampling unit. Based on the spatial frequency down-sampled image, filtering parameter determining means for determining a filtering parameter for filtering processing used when the original image is reproduced by spatial frequency up-sampling, and generated by the down-sampling means Coding means for coding at least the spatial frequency down-sampled image among the spatial frequency down-sampled image and the filtering parameter determined by the filtering parameter determination means; Characterized by comprising a.

  According to a second aspect of the present invention, in the image coding apparatus according to the first aspect, the down-sampling means performs an octave decomposition filter bank process on the original image, and a low frequency region component of decomposition rank n (n is an integer). Spatial frequency down-sampled image, and the filtering parameter determination means performs the octave decomposition filter bank processing on the low-frequency domain components of the decomposition rank n generated by the down-sampling means, and the horizontal, vertical and Diagonal high-frequency region components are obtained, and each of these horizontal, vertical, and oblique high-frequency region components is set as a spatial low-frequency region component, and the zero matrix is expanded as a spatial high-frequency region component and expanded by octave reconstruction filter bank processing. The horizontal, vertical, and diagonal high-frequency region components of the decomposition rank n obtained by Filtering processing is performed using ring parameters, and horizontal, vertical, and diagonal high-frequency region components of the decomposition rank n obtained by filtering processing are used as spatial high-frequency region components, and the spatial high-frequency region components and the low-frequency region of the decomposition rank n The component is expanded by octave reconstruction filter bank processing, and the low-frequency region component of decomposition rank n-1 obtained by the expansion, and the low-frequency region component of decomposition rank n-1 generated by the downsampling means And the filtering parameter is determined by changing the value of the filtering parameter so that the difference value is minimized.

  According to a third aspect of the present invention, in the image encoding device according to the second aspect, wavelet decomposition is used as the octave decomposition filter bank processing.

  According to a fourth aspect of the present invention, in the image encoding device according to the second or third aspect, wavelet reconstruction is used as the octave reconstruction filter bank processing.

  According to a fifth aspect of the present invention, in the image encoding device according to any one of the first to fourth aspects, a Gaussian filter process is used for the filtering process, and a variance value of a Gaussian filter is used for the filtering parameter. It is characterized by that.

  According to a sixth aspect of the present invention, there is provided an image decoding apparatus comprising: decoding means for decoding a spatial frequency down-sampled image encoded by the image encoding apparatus according to any one of claims 2 to 5; The low frequency region component of decomposition rank n which is the spatial frequency down-sampled image decoded by the means is subjected to octave decomposition filter bank processing, and each of the horizontal, vertical and diagonal high frequency region components of decomposition rank n + 1 obtained by this processing Is the spatial low-frequency region component, the zero matrix is the spatial high-frequency region component, and is expanded by octave reconstruction filter bank processing, and each of the expanded horizontal, vertical, and diagonal high-frequency region components of the decomposition rank n is Obtained by performing a filtering process using the filtering parameter determined by the converting apparatus, and a filtering process. Upsampling means for performing octave reconstruction filter bank processing using the horizontal, vertical and diagonal high frequency region components of the solution rank n and the low frequency region components of the decomposition rank n before the spatial frequency upsampling, and performing spatial frequency upsampling; It is provided with.

  The invention of claim 7 is an image encoding program for causing a computer to function as the image encoding apparatus according to any one of claims 1 to 5.

  The invention according to claim 8 is an image decoding program for causing a computer to function as the image decoding apparatus according to claim 6.

  According to the present invention, the original image is subjected to octave decomposition filter bank processing, and a spatial frequency down-sampled image is generated and encoded. The filtering parameter that makes it possible to reproduce close images was determined. As a result, even if the spatial frequency downsampled image obtained by downsampling the original image is transmitted to the receiving side, the receiving side performs higher spatial sampling by using the filtering parameter, thereby increasing the frequency. High quality images can be played back.

It is a block diagram which shows schematic structure of the image coding apparatus by the Example of this invention. It is a functional block diagram of an image coding apparatus. It is a figure which shows the mode of the downsampling process in an image coding apparatus. It is a figure which shows the outline | summary of the dispersion value determination process in an image coding apparatus. It is a flowchart which shows the outline | summary of the dispersion value determination process in an image coding apparatus. It is a flowchart which shows the encoding and transmission process in an image coding apparatus. It is a figure which shows the outline | summary of the modification of the variance value determination process in an image coding apparatus. It is a block diagram which shows schematic structure of the image decoding apparatus by the Example of this invention. It is a functional block diagram of an image decoding apparatus. It is a flowchart which shows the reception and decoding process in an image decoding apparatus. It is a figure which shows the mode of the upsampling process in an image decoding apparatus. It is a flowchart which shows the outline | summary of the upsampling process in an image decoding apparatus.

  The best mode for carrying out the present invention will be described below with reference to the drawings. In the embodiment of the present invention described below, an original image is subjected to spatial frequency downsampling (hereinafter referred to as “downsampling”) using wavelet decomposition, and a spatial frequency downsampled image (hereinafter referred to as “downsampling”) obtained thereby. Assume an image encoding apparatus that encodes and transmits a “sampled image”. In the embodiment, wavelet decomposition is used for downsampling, but a linear filter or the like can also be used. That is, any device that can perform octave decomposition filter bank processing can be used.

(Configuration of image encoding device)
FIG. 1 is a block diagram showing a schematic configuration of an image encoding device 1 according to an embodiment of the present invention. The image encoding apparatus 1 includes a downsampling unit (downsampling means) 10 that performs wavelet decomposition on an original image F, downsamples, and outputs a low-resolution downsampled image CA ⁽ⁿ⁾ (n: decomposition rank). Based on the downsampled images CA ⁽ⁿ⁾ and CA ⁽ⁿ⁻¹⁾ obtained by wavelet decomposition in the downsampling unit 10, variance values σ _CH ⁽ⁿ⁾ and σ _{CV of a} Gaussian filter used in the upsampling process described later. ^(N) , σ _CD ⁽ⁿ⁾ (subscripts CH, CV, CD indicate high-frequency components in the horizontal, vertical, and diagonal directions), a dispersion value determining unit (filtering parameter determining means) 11, and downsampling encoding for encoding the downsampled picture CA ⁽ⁿ⁾ obtained by wavelet decomposition in parts 10 And (encoding means) 12, the coded downsampled image EnCA the encoding unit 12 ⁽ⁿ⁾ and the variance of the Gaussian filter determined by the variance determining section ^{_{11 σ CH (n), σ}} CV (n) , Σ _CD ⁽ⁿ⁾ as a broadcast wave. Note that the variance value of the Gaussian filter is not encoded in this embodiment because the amount of information is small, but may be encoded.

  FIG. 2 is a functional block diagram of the image encoding device 1. The spatial wavelet decomposition process is a process performed by the downsampling unit 10 described above, and a Gaussian filter application process, a spatial wavelet enlargement process, and an image comparison process for each n-th spatial high-frequency component are performed by the variance value determination unit 11. It is processing. Hereinafter, processing in the downsampling unit 10, the variance value determining unit 11, the encoding unit 12, and the transmission unit 13 will be described in detail.

(Downsampling processing in the downsampling unit 10)
In this embodiment, wavelet decomposition is used for downsampling processing. The wavelet decomposition rank is determined by the size to be reduced. When the size to be reduced is 1/2 ⁿ in both the vertical and horizontal directions, the decomposition rank is n + 1. For example, the size to be reduced is both vertically and horizontally when 1/2 ^1, decomposition rank is 2. The wavelet used for the downsampling process is wavelet_n, the horizontal low frequency / vertical low frequency components of decomposition rank n are CA ⁽ⁿ⁾ (χ, Ψ), and the horizontal high frequency / vertical low frequency components are CH ^{(n )} (Χ, ψ), horizontal low frequency / vertical high frequency components are CV ⁽ⁿ⁾ (χ, ψ), and horizontal high frequency / vertical high frequency components are CD ⁽ⁿ⁾ (χ, ψ). The horizontal and vertical (x, y) is displayed as (χ, Ψ) because the sample is halved both vertically and horizontally when the resolution is converted, so (x, y) is not displayed. χ, Ψ).

The downsampling unit 10 uses the n-th order discrete wavelet decomposition DWT ⁽ⁿ⁾ (F (x, y, t), wavelet_n) for the original image F (x, y, t) as shown in Expression (1). Disassembled into t indicates time.
[Formula 1]
[CA ⁽ⁿ⁾ , CH ⁽ⁿ⁾ , CV ⁽ⁿ⁾ , CD ⁽ⁿ⁾ ] = DWT ⁽ⁿ⁾ (F (x, y, t), wavelet_n) (1)

FIG. 3 is a diagram illustrating a state of the downsampling process. By using the first-order discrete wavelet decomposition DWT ⁽¹⁾ , horizontal low frequency / vertical low frequency frequency component CA ⁽¹⁾ , horizontal high frequency / vertical low frequency component CH ⁽¹⁾ , horizontal low frequency / vertical high frequency component CV from the original image F are obtained. ⁽¹⁾ Horizontal high frequency / vertical high frequency component CD ⁽¹⁾ is obtained. Then, from CA ⁽¹⁾ to horizontal low frequency / vertical low frequency component CA ⁽²⁾ , horizontal high frequency / vertical low frequency component CH ⁽²⁾ , horizontal low frequency / vertical high frequency component CV ⁽²⁾ , horizontal high frequency / vertical. A high frequency component CD ⁽²⁾ is obtained. The finally obtained down-sampled image is an n-th order discrete wavelet decomposition component CA ⁽ⁿ⁾ .

(Determination processing of variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter in the variance value determination unit 11)
FIG. 4 is a diagram showing an outline of processing of the variance value determination unit 11. First, the dispersion value determining unit 11 performs first-order discrete wavelet decomposition on the low-frequency domain component CA ⁽ⁿ⁾ of the decomposition rank n, and each high-frequency domain component CH ^{(n + 1)} , horizontal, vertical, and diagonal directions in the decomposition rank n + 1. CV ^{(n + 1)} and CD ^{(n + 1)} are obtained (step S1). Each of the high-frequency region components CH ^{(n + 1)} , CV ^{(n + 1)} , and CD ^{(n + 1) in} the horizontal, vertical, and diagonal directions is used as the spatial low-frequency region component, and the zero matrix of the same size is used as the spatial high-frequency region component. ExCH ⁽ⁿ⁾ , ExCV ⁽ⁿ⁾ , and ExCD ⁽ⁿ⁾ are obtained by enlarging the horizontal and vertical directions twice by the discrete wavelet reconstruction (step S2).

Equation (2) shows an equation obtained by reconstructing the first-order discrete wavelet using CH ^{(n + 1)} as a spatial low frequency region component and a zero matrix of the same size as a spatial high frequency region component. 0 indicates a zero matrix.
[Formula 2]
ExCH ⁽ⁿ⁾ = IDWT ⁽¹⁾ (CH ^{(n + 1)} , 0, 0, 0, wavelet_n) (2)

After obtaining ExCH ⁽ⁿ⁾ , ExCV ⁽ⁿ⁾ , and ExCD ⁽ⁿ⁾ expanded twice in the horizontal and vertical directions, the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ^{(n) of the} Gaussian filter are obtained. And GExCH ⁽ⁿ⁾ , GExCV ⁽ⁿ⁾ , GExCD ⁽ⁿ⁾ are obtained (step S3). In this case, arbitrary values or empirical values are used as initial values for the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , and σ _CD ⁽ⁿ⁾ of the Gaussian filter. For example, if the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , and σ _CD ⁽ⁿ⁾ of the Gaussian filter are changed in increments of 0.1 to 0.1, the initial value is, for example, σ _CH ^{( n)} = 0.1, σ _CV ⁽ⁿ⁾ = 0.1, and σ _CD ⁽ⁿ⁾ = 0.1.

After performing filtering using a Gaussian filter to obtain GExCH ⁽ⁿ⁾ , GExCV ⁽ⁿ⁾ , GExCD ⁽ⁿ⁾ , the low frequency region component CA ⁽ⁿ⁾ is used as the spatial low frequency region component, and GExCH ⁽ⁿ⁾ , GExCV ^{( n)} , GExCD ⁽ⁿ⁾ is used as a spatial high-frequency region component, and doubled horizontally and vertically by first-order discrete wavelet reconstruction to obtain GExCA ^(n-1) (step S4).

After obtaining GExCA ^(n-1) , a difference value from CA ^(n-1) is calculated. If the current difference value is smaller than the previous difference value, the current difference value and the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter used this time are stored. The variance value determining unit 11 has a memory (not shown), and in this memory, the current difference value and the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ^{(n) of} the Gaussian filter used this time. Remember. Since there is no previous difference value at the first time, the difference value calculated this time and the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter used this time are stored as they are. Further, as an image evaluation method, there is a mean square error (MSE), and this evaluation method may be used.

The processes in steps S1 to S5 are repeated while changing the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter, and GExCA ⁽ⁿ⁻¹⁾ and CA ⁽ⁿ⁻ The variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , and σ _CD ⁽ⁿ⁾ of the Gaussian filter when the difference value from ¹⁾ is the smallest are determined. The determined variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ and σ _CD ⁽ⁿ⁾ of the Gaussian filter are transmitted to the receiving side (image decoding apparatus 2 described later ⁾ together with the down-sampled image EnCA ⁽ⁿ⁾ . On the receiving side, by receiving and using the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter, an image closest to the original image from the down-sampled image EnCA ⁽ⁿ⁾ is obtained. Can be played.

FIG. 5 is a flowchart showing an outline of processing for determining the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , and σ _CD ⁽ⁿ⁾ of the Gaussian filter. The image encoding device 1 performs spatial wavelet decomposition on the original image F (x, y, t), and uses the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of each nth-order Gaussian filter. Determine and output with CA ⁽ⁿ⁾ .

When a wavelet other than the wavelet such as Haar having linear (linear) phase characteristics is used, it is necessary to pay attention to the phases of ExCH ⁽ⁿ⁾ , ExCV ⁽ⁿ⁾ , and ExCD ⁽ⁿ⁾ . ExCH ⁽ⁿ⁾ , ExCV ⁽ⁿ⁾ , ExCD ⁽ⁿ⁾ are shifted by horizontal and vertical phases (φ _μ , φ _ν ) ExCH ⁽ⁿ⁾ (χ + φ _μ , Ψ + φ _ν ), ExCV ⁽ⁿ⁾ (χ + φ _{μ , Ψ + φ ν), ExCD} (n) (χ + φ μ, Ψ + φ ν) GExCH (n) ^{for, GExCV (n),} to obtain GExCD ^(n), the variance value of the finally Gaussian filter while changing the phase sigma _CH ^(N) , σ _CV ⁽ⁿ⁾ , and σ _CD ⁽ⁿ⁾ may be determined. In this embodiment, however, CA ⁽ⁿ⁻¹⁾ and GExCA ⁽ⁿ⁻¹⁾ are compared while changing the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , and σ _CD ⁽ⁿ⁾ of the Gaussian filter. In this part, deterioration due to phase shift is absorbed to some extent. For this reason, a phase matching process that requires a large amount of processing and a large-scale circuit process is not necessarily required.

(Details of determination processing of variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter in the variance value determination unit 11)
The variance value determining unit 11 enlarges the diagonal high-frequency region component CD ^{(n + 1)} in the layer n + 1 twice in the horizontal and vertical directions (step S2 in FIG. 4), and then filters using a Gaussian filter (in FIG. 4). Step S3). Equation (3) shows a Gaussian filter.
[Formula 3]
Gauss (x, y, σ) = exp (− (x ² + y ² ) / 2σ ² ) (3)

Σ ^{2 in the} formula (3) represents dispersion. After the diagonal high-frequency region component CD ^{(n + 1)} is doubled horizontally and vertically to obtain ExCD ^{(n) (} step S2 in FIG. 4), the equation (4 ⁾ is obtained using the Gaussian filter of equation (3). (Step S3 in FIG. 4 ⁾ to obtain GExCD ⁽ⁿ⁾ . Note that σ in Expression (3) is σ _CD ⁽ⁿ⁾ because this processing is within the processing for filtering ExCD ⁽ⁿ⁾ .

[Formula 4]

Further, the variance value determining unit 11 enlarges the horizontal high-frequency region component CH ^{(n + 1)} and the vertical high-frequency region component CV ^{(n + 1)} in the layer ^{n + 1} twice in the horizontal and vertical directions, and then expands the directionality. Use to enlarge. Equation (5) shows a Gaussian filter expanded in directionality.
[Formula 5]
ExGauss (x, y, α, β, σ) = exp (− (αx ² + βy ² ) / 2σ ² )
... (5)

In the equation (5), α and β are directional expansion coefficients, α = 1 and β = 0 in the case of the horizontal high-frequency region component CH ⁽ⁿ⁾ , and α = in the case of the vertical high-frequency region component CV ^(n). 0, β = 1. Hereinafter, only the case of determining σ _CH ⁽ⁿ⁾ will be described, but σ _CD ⁽ⁿ⁾ can be determined in the same manner.

After the horizontal high-frequency region component CH ^{(n + 1)} in the hierarchy n + 1 is expanded twice in the horizontal and vertical directions to obtain ExCH ⁽ⁿ⁾ , using the Gaussian filter expanded in the direction of Expression (5), Expression (6) ) To obtain GExCH ⁽ⁿ⁾ . Note that σ in Expression (5) is σ CH ^{(n) because} the present process is within the process of filtering _CH ⁽ⁿ⁾ .
[Formula 6]

Through the above processing, the variance value determination unit 11 calculates the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter, and then CA ⁽ⁿ⁾ , GExCH ⁽ⁿ⁾ , GExCV. ^(N) , GExCD ⁽ⁿ⁾ is used to perform first-order discrete wavelet reconstruction (step S4 in FIG. 4 ⁾ to obtain GExCA ^(n-1) . Finally, as shown in the equation (7), the mean square error (RMS) in the area of CA ^{(n-1) and} the area of GExCA ^(n-1 ) is calculated, and RMS (CA ^(n-1) ) And RMS (GExCA ^(n-1) ) are obtained and the difference value is calculated (step S5 in FIG. 4). The above processing is performed while changing the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter, and Diff_rms (σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ^(N) The variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter that minimizes) are obtained.

[Formula 7]
Diff_rms (σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ ) =
RMS (CA ^(n-1) (σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ ))-RMS (GExCA ^(n-1) (σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ )
σ _CH ⁽ⁿ⁾ = argmin (Diff_rms (σ _CH ⁽ⁿ⁾ ))
σ _CV ⁽ⁿ⁾ = argmin (Diff_rms (σ _CV ⁽ⁿ⁾ )) (7)
σ _CD ⁽ⁿ⁾ = argmin (Diff_rms (σ _CD ⁽ⁿ⁾ ))

(Encoding process in the encoding unit 12)
The encoding unit 12 encodes CA ⁽ⁿ⁾ obtained by Expression (1). In the present embodiment, since the downsampling process and the upsampling process are performed in units of one frame, the encoding unit 12 performs JPEG, JPEG200, MPEG-2, H.264, and so on. Various encoding methods such as H.264 | MPEG-4 AVC can be used.

(Transmission processing in the transmission unit 13)
The transmission unit 13 transmits EnCA ⁽ⁿ⁾ encoded by the encoding unit 12 as a broadcast wave. At this time, for example, the variance value σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter determined by the variance value determination unit 11 is accommodated in the header of MPEG-2 TS and transmitted. . FIG. 6 is a flowchart showing the encoding and transmission processing. The image encoding device 1, ^CA a ⁽ⁿ⁾ to give enca by encoding ^(n), enca variance of the Gaussian filter with _{^{(n) σ CH (n)}} , σ CV (n), σ CD (n) Send.

As described above, the image encoding device 1 performs wavelet decomposition on the original image F, downsamples, and obtains and encodes a low-resolution downsampled image CA ⁽ⁿ⁾ . Further, the image encoding device 1 is based on the down-sampled images CA ⁽ⁿ⁾ and CA ^(n−1), and the variance value σ _CH ^{(n of the} Gaussian filter that can generate an image closest to the original image by the up-sampling process. ⁾ , Σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ are determined and encoded downsampled image EnCA ⁽ⁿ⁾ and variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ^{(n) of the} Gaussian filter. And send.

The variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ and σ _CD ⁽ⁿ⁾ of the Gaussian filter are determined as follows. That is, wavelet decomposition is performed on the low frequency region component CA ⁽ⁿ⁾ , and the horizontal high frequency region component CH ^{(n + 1)} , the vertical high frequency region component CV ^{(n + 1)} , and the oblique high frequency region component CD ^{(n + 1)} obtained thereby. ExCH ⁽ⁿ⁾ , ExCV ⁽ⁿ⁾ , ExCD ^{(n) are} expanded horizontally and vertically by the first-order discrete wavelet reconstruction, each of which is a spatial low frequency region component and a zero matrix of the same size as a spatial high frequency region component. Get. Then, they are filtered by a Gaussian filter using arbitrary dispersion values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ , and GExCH ⁽ⁿ⁾ , GExCV ⁽ⁿ⁾ , GExCD ^{(n ) )} Next, CA ⁽ⁿ⁾ is used as a spatial low frequency region component, and GExCH ⁽ⁿ⁾ , GExCV ⁽ⁿ⁾ , and GExCD ⁽ⁿ⁾ are used as spatial high frequency region components, and the horizontal and vertical directions are doubled by first-order discrete wavelet reconstruction. To obtain GExCA ^(n-1) . Then, the resulting ^{GExCA (n-1)} compared same as decomposition rank of ^{CA (n-1)} and the dispersion value sigma _CH of Gaussian filter the difference value becomes smallest ^{_{(n), σ CV (n}} ), Determine σ _CD ⁽ⁿ⁾ .

  Thereby, even if a spatial frequency down-sampled image obtained by down-sampling the original image F is transmitted, a high-quality image closest to the original image F can be reproduced on the receiving side.

In the present embodiment, by comparing GExCA ⁽ⁿ⁻¹⁾ and CA ⁽ⁿ⁻¹⁾ , the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter are obtained. The dispersion value of the Gaussian filter is determined by comparing each of CH ⁽ⁿ⁾ and GExCH ⁽ⁿ⁾ , CV ⁽ⁿ⁾ and GExCV ⁽ⁿ⁾ , and CD ⁽ⁿ⁾ and GExCD ^(n). σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , and σ _CD ⁽ⁿ⁾ may be determined.

FIG. 7 is a diagram illustrating an outline of a modified example of the dispersion value determination process in the image encoding device. The variance value determination unit 11 obtains a variance value σ _CH ⁽ⁿ⁾ that minimizes the mean square error (MSE) in the comparison between CH ⁽ⁿ⁾ and GExCH ⁽ⁿ⁾ . Also, a variance value σ _CV ⁽ⁿ⁾ that minimizes the mean square error in the comparison between CV ⁽ⁿ⁾ and GExCV ⁽ⁿ⁾ is obtained. Further, a variance value σ _CD ⁽ⁿ⁾ that minimizes the mean square error in the comparison between CD ⁽ⁿ⁾ and GExCD ⁽ⁿ⁾ is obtained.

  As a hardware configuration of the image encoding device 1, a normal computer can be used. That is, it can be configured by a computer having a volatile storage medium such as a CPU (Central Processing Unit) and a RAM (Random Access Memory), a nonvolatile storage medium such as a ROM (Read Only Memory), an interface, and the like. The functions of the downsampling unit 10, the variance value determination unit 11, the encoding unit 12, and the transmission unit 13 included in the image encoding device 1 can be realized by causing the CPU to execute a program describing these functions. These programs can be stored and distributed in a storage medium such as a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, or the like.

(Configuration of image decoding device)
Next, an image decoding apparatus according to an embodiment of the present invention will be described. FIG. 8 is a block diagram illustrating a schematic configuration of the image decoding apparatus. The image decoding device 2 receives the down-sampled image EnCA ⁽ⁿ⁾ and the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter transmitted from the image coding device 1 described above. Receiving section 20, a decoding section (decoding means) 21 for decoding the down-sampled image EnCA ⁽ⁿ⁾ received by the receiving section 20, and a Gaussian filter variance value σ _CH ⁽ⁿ⁾ , received by the receiving section 20 an upsampling unit (upsampling means) 22 for upsampling the downsampled image EnDeCA ⁽ⁿ⁾ decoded by the decoding unit 21 using σ _CV ⁽ⁿ⁾ and σ _CD ^(n). The

FIG. 10 is a flowchart showing processing of the receiving unit 20 and the decoding unit 21. The receiving unit 20 receives the encoded downsampled image EnCA ⁽ⁿ⁾ and the variance values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the Gaussian filter, and the decoding unit 21 receives the downsampling. The image EnCA ⁽ⁿ⁾ is decoded to obtain EnDeCA ⁽ⁿ⁾ .

  FIG. 9 is a functional block diagram of the image decoding device 2. Spatial wavelet decomposition processing, Gaussian filter application processing to each n-th spatial high-frequency component, and spatial wavelet expansion processing are processing performed by the upsampling unit 22. Hereinafter, processing in the upsampling unit 22 will be described.

(Upsampling process in the upsampling unit 22)
FIG. 11 is a diagram illustrating a state of the upsampling process. First, as shown in Expression (8), the upsampling unit 22 performs first-order discrete wavelet decomposition DWT ⁽¹⁾ on EnDeCA ⁽ⁿ⁾ obtained by receiving EnCA ⁽ⁿ⁾ and decoding EnEnCA ^{(n + 1). ), EnDeCH (n + 1)} , EnDeCV (n + 1), obtaining a EnDeCD ^{(n + 1) (step} S11).
[Formula 8]
[EnDeCA ^{(n + 1)} , EnDeCH ^{(n + 1)} , EnDeCV ^{(n + 1)} , EnDeCD ^{(n + 1)} ] = DWT ⁽¹⁾ (EnDeCA ⁽ⁿ⁾ , wavelet_n)
... (8)

Then, the horizontal, vertical and diagonal expansion components are calculated using EnDeCH ^{(n + 1)} , EnDeCV ^{(n + 1)} , EnDeCD ^{(n + 1)} and dispersion values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ To do. To EnDeCD ^{(n + 1),} EnDeCD the spatial low-frequency domain components ^{(n + 1),} twice the horizontal and the first floor discrete wavelet reconstruction at a zero matrix of the same size in the spatial frequency domain components, a vertical double enlargement component ExEnDeCD ^{( n)} is obtained (step S12). Similarly, the first-order discrete wavelet is reconfigured for EnDeCH ^{(n + 1)} and EnDeCV ^{(n + 1)} to obtain horizontally doubled and vertically doubled expanded components ExEnDeCH ⁽ⁿ⁾ and ExEnDeCV ⁽ⁿ⁾ (step S12).

Then, ExEnDeCH ⁽ⁿ⁾ , ExEnDeCV ⁽ⁿ⁾ , and ExEnDeCD ⁽ⁿ⁾ are subjected to a filtering process using a Gaussian filter with dispersion values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ , and GExEnDeCH ^{( n)} , GExEnDeCV ⁽ⁿ⁾ , GExEnDeCD ⁽ⁿ⁾ are obtained (step S13).

Then, as shown in Equation (9), first-order discrete wavelet reconstruction is performed using EnDeCA ⁽ⁿ⁾ , GExEnDeCH ⁽ⁿ⁾ , GExEnDeCV ⁽ⁿ⁾ , and GExEnDeCD ⁽ⁿ⁾ (step S14).
[Formula 9]
[EnDeCA ^(n-1) ] = IDWT ⁽¹⁾ (EnDeCA ⁽ⁿ⁾ , GExEnDeCH ⁽ⁿ⁾ , GExEnDeCV ⁽ⁿ⁾ , GExEnDeCD ⁽ⁿ⁾ , wavelet_n) (9)

  Thereafter, assuming that n = n−1, the processing in the upsampling method is repeated until n = 0 (step S15).

FIG. 12 is a flowchart showing an outline of the upsampling process in the upsampling unit 22. The upsampling unit 22 obtains EnDeCH ^{(n + 1)} , EnDeCV ^{(n + 1)} , EnDeCD ^{(n + 1)} by performing spatial wavelet decomposition on EnDeCA ⁽ⁿ⁾ obtained by decoding, and then reconstructs them to ExEnDeCH ^{( n)} , ExEnDeCV ⁽ⁿ⁾ , ExEnDeCD ⁽ⁿ⁾ are obtained. The variance _{^{σ CH (n), σ CV}} (n), performs filtering Gaussian filter _{^{σ CD (n), GExEnDeCH (}} n), GExEnDeCV (n), to obtain GExEnDeCD ^(n), these and EnDeCA Wavelet reconstruction is performed using ⁽ⁿ⁾ . The above processing is repeated with n = n−1 until n = 0.

As described above, according to the image decoding device 2 shown in FIG. 8, the received and decoded downsampled image EnDeCA ⁽ⁿ⁾ is subjected to wavelet decomposition, and the horizontal high-frequency region component EnDeCH ^{(n + 1)} , vertical obtained by this Each of the high-frequency region component EnDeCV ^{(n + 1)} and the oblique high-frequency region component EnDeCD ^{(n + 1)} is expanded by wavelet reconstruction using the spatial low-frequency region component and the zero matrix as the spatial high-frequency region component, and the resulting horizontal high-frequency region component ExEnDeCH ^(N) , vertical high frequency region component ExEnDeCV ⁽ⁿ⁾ , and oblique high frequency region component ExEnDeCD ⁽ⁿ⁾ using the received dispersion values σ _CH ⁽ⁿ⁾ , σ _CV ⁽ⁿ⁾ , σ _CD ⁽ⁿ⁾ of the received Gaussian filter The horizontal high frequency region component GEx obtained by filtering is obtained. EnDeCH ^(n), the vertical high-frequency domain components GExEnDeCV ^(n), since the wavelet reconstruction using oblique high-frequency domain components GExEnDeCD ⁽ⁿ⁾ and downsampling before EnDeCA ^(n), a closest quality to the original image F Images can be played back.

  As a hardware configuration of the image decoding device 2, a normal computer can be used. That is, it can be configured by a computer including a volatile storage medium such as a CPU and a RAM, a nonvolatile storage medium such as a ROM, an interface, and the like. Each function of the receiving unit 20, the decoding unit 21, and the upsampling unit 22 constituting the image decoding device 2 can be realized by causing the CPU to execute a program describing these functions. These programs can be stored and distributed in a storage medium such as a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, or the like.

  The present invention has been described with reference to the embodiments. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the technical idea thereof. In the above-described embodiment, the image encoding device 1 transmits the downsampled image and the variance value of the Gaussian filter, and the image decoding device 2 upsamples the downsampled image using the variance value of the Gaussian filter. Although an image is reproduced, the present invention is applicable to all processes for performing spatial frequency downsampling and upsampling. For example, after the image encoding device 1 downsamples the spatial frequency, the image encoding device 1 transmits the compression encoded signal of the spatial frequency downsampled image and the additional information obtained at the time of the spatial frequency downsampling to the image decoding device 2. Then, the image decoding device 2 receives the compressed encoded signal and the additional information of the spatial frequency down-sampled image, and performs spatial frequency up-sampling after decoding the compressed encoded signal. This enables compression encoding with low image quality degradation and low transmission capacity. It can also be applied to processing such as compression recording on storage media.

  The present invention can also be used before or after the digital cinema compression / encoding process in which JPEG2000 is used for compression encoding in units of frames. H. H.264 | MPEG-4 AVC can be used before or after the super high-definition compression coding process that is being studied for compression coding. Further, it can be used in combination with a new compression encoding method in the future.

DESCRIPTION OF SYMBOLS 1 Image coding apparatus 2 Image decoding apparatus 10 Downsampling part 11 Dispersion value determination part 12 Encoding part 13 Transmission part 20 Reception part 21 Decoding part 22 Upsampling part

Claims (8)

Down-sampling means for performing spatial frequency down-sampling on the original image and generating a spatial frequency down-sampled image;
Filtering parameter determining means for determining a filtering parameter for filtering processing used when the original image is reproduced by spatial frequency upsampling based on the spatial frequency downsampled image generated by the downsampling means;
Encoding means for encoding at least the spatial frequency downsampled image among the spatial frequency downsampled image generated by the downsampling means and the filtering parameter determined by the filtering parameter determination means;
An image encoding apparatus comprising: The image encoding device according to claim 1,
The down-sampling means performs octave decomposition filter bank processing on the original image to generate a spatial frequency down-sampled image of a low-frequency region component of decomposition rank n (n is an integer),
The filtering parameter determination means performs the octave decomposition filter bank processing on the decomposition frequency n of the decomposition rank n generated by the downsampling means to obtain horizontal, vertical and diagonal high frequency area components in the decomposition rank n + 1, Each horizontal, vertical, and diagonal high-frequency domain component is a spatial low-frequency domain component, and the zero matrix is a spatial high-frequency domain component, and is expanded by octave reconstruction filter bank processing. The vertical and diagonal high-frequency region components are filtered using an arbitrary filtering parameter, and the horizontal, vertical and diagonal high-frequency region components of the decomposition rank n obtained by the filtering process are defined as the spatial high-frequency region components. Octave to the low frequency region component of the region component and decomposition rank n The low-frequency domain component of decomposition rank n-1 obtained by enlarging and expanding by reconstruction filter bank processing is compared with the low-frequency domain component of decomposition rank n-1 generated by the downsampling means, An image encoding apparatus, wherein the filtering parameter is determined by changing the value of the filtering parameter so that the difference value is minimized. The image encoding device according to claim 2,
An image coding apparatus using wavelet decomposition as the octave decomposition filter bank processing. In the image coding device according to claim 2 or 3,
An image coding apparatus using wavelet reconstruction as the octave reconstruction filter bank processing. In the image coding device according to any one of claims 1 to 4,
An image coding apparatus, wherein a Gaussian filter process is used for the filtering process, and a variance value of a Gaussian filter is used for the filtering parameter. Decoding means for decoding the spatial frequency downsampled image encoded by the image encoding device according to any one of claims 2 to 5,
The spatial frequency down-sampled image decoded by the decoding means is subjected to octave decomposition filter bank processing for the decomposition rank n low frequency region components, and the decomposition rank n + 1 horizontal, vertical and diagonal high frequency region components obtained by this processing Each of the spatial low-frequency region components and the zero matrix as the spatial high-frequency region component, and expanded by octave reconstruction filter bank processing, each of the expanded horizontal, vertical and diagonal high-frequency region components of the decomposition rank n, Filtering is performed using the filtering parameters determined by the image encoding apparatus, and the decomposition rank n obtained by the filtering process is reduced in the horizontal, vertical and diagonal high-frequency region components and the decomposition rank n before spatial frequency upsampling is reduced. Perform octave reconstruction filter bank processing using frequency domain components And up-sampling means for up-sampling,
An image decoding apparatus comprising:   An image encoding program for causing a computer to function as the image encoding apparatus according to any one of claims 1 to 5.   An image decoding program for causing a computer to function as the image decoding apparatus according to claim 6.

Images