KR100922376B1 - Method for resizing image using interpolation filter selected in block transform domain based on characteristic of image - Google Patents

Abstract

The present invention relates to a method for changing the size of a video image. More specifically, when block transforming a video image of a spatial domain into a frequency component, the present invention is adaptive according to an image characteristic determined from the block-transformed video image. A method of changing the size of a video image by selecting an interpolation filter.

The method for changing the size of a video image according to the present invention uses a single fixed filter regardless of the video characteristics by reducing / enlarging the size of the video image using different interpolation filters according to the video characteristics of the video image. Compared to the conventional method of reducing / enlarging the size of a video image, a high quality video image is provided. In addition, when the integer DCT is used as an example of the adaptive filter used in the method for changing the size of the video image according to the present invention, it is applicable to the integer DCT used in the H.264 / AVC codec. It is compatible with AVC codec. In addition, when applied to integer DCT, it is possible to prevent a bit error caused by changing a real operation into an integer operation during encoding / decoding of a video image.

Block Transform, DCT, Integer DCT, Interpolation Filter

Description

Method for resizing image using interpolation filter selected in block transform domain based on characteristic of image}

The present invention relates to a method for changing the size of a video image. More specifically, when block transforming a video image of a spatial domain into a frequency component, the present invention is adaptive according to an image characteristic determined from the block-transformed video image. A method of changing the size of a video image by selecting an interpolation filter.

The current communication environment has different communication networks such as broadband convergence network (BcN), which is a core network that enables services such as wired / wireless interworking, convergence of broadcasting networks and communication networks, and IP convergence using IP networks. They are converging with each other, and this trend will continue to accelerate.

Recently, according to the trend of digital convergence in different network environments, a lot of researches on general-purpose media communication have been conducted. As the types of multimedia terminals used in various communication networks are becoming more and more diverse, various techniques for changing the size of a video image that can reasonably harmonize different networks and multimedia terminals are being developed.

Discrete Cosine Transform (DCT) is one of the block transform techniques for transforming the image of the spatial domain into frequency components to change the size of the image. The MPEG-1, MPEG-2, MPEG-4, H It is widely used in video codecs such as .261, H.263, and H.26L. The DCT technique is an orthogonal transformation encoding technique that represents N data arrays as the sum of N cosine functions. In other words, the DCT technique converts data of a video image in a spatial domain into frequency components and expresses the converted frequency components as a sum of a cosine function. Here, the coefficient of each cosine function is called a DCT coefficient, and each DCT coefficient represents a frequency component of data for an image image of a spatial domain.

The DCT technique described above has strong energy compaction and optimization when compared to other block transformation techniques such as Discrete Fourier Transform (DFT), Hadamard Transform, Discrete Wavelet Transform (DWT), and Karhunen Loeve Transform (KLT). It is easy.

Figure 1 shows a functional block diagram of a device for changing the size of a conventional video image.

Referring to FIG. 1, the image block converter 1 divides an input image of a spatial region into an N × N sized image block, and converts the divided image block into image block units. For example, the image block converter 1 divides an input image into a 4x4 or 8x8 sized image block, and divides each divided image block using a DCT technique or an integer DCT technique. To block conversion.

The interpolator 2 reduces or enlarges the size of the input video image by interpolating the image block of the spatial region converted by the image block converter 1. The interpolator 2 uses one fixed one interpolation filter to reduce or enlarge the size of the image block converted by the image block converter 1. Meanwhile, the encoder 3 quantizes the data of the reduced or enlarged video image and encodes the data of the quantized video image according to a method such as predictive coding, variable length coding, or arithmetic coding. To generate a bitstream of the video or video.

In the apparatus for changing the size of a conventional video image described with reference to FIG. 1, the interpolator 2 always reduces only one interpolation filter that is fixed regardless of the image characteristics of the video image to reduce or enlarge the size of the video image. To reduce or enlarge the size of the video image.

However, when the size of a video image is reduced or enlarged using an interpolation filter, the interpolation performance of the video signal may be improved by using an adaptive interpolation filter according to the image characteristics of the video image. For example, when an image block is an intra block, the image block is preferably interpolated using an interpolation filter capable of preserving as many low frequency components as possible. On the other hand, if the video block is an Inter block, the video block is preferably interpolated using a filter that preserves as many high-frequency components as possible. Here, an intra block refers to an image block that is not predicted well from a reference image and is encoded in a frame, and an inter block refers to an image block that can be predicted from a reference image and encoded between frames.

Accordingly, an object of the present invention is to provide a method for changing the size of an image image using an interpolation filter adaptively selected according to the image characteristics of the image image in the block transform region.

In order to achieve the above object of the present invention, the method for changing the size of a video image according to the present invention comprises the steps of converting a video image of an input spatial domain into a frequency component, and converting the video image into a frequency component. And an interpolation filter for reducing or enlarging the size of the image image based on the image characteristic of the analyzed image image.

Preferably, the input image of the spatial domain is block-converted by the DCT technique or the integer DCT technique and is block-converted by the unit of image blocks having a predetermined size.

Preferably, the video characteristic of the video image is analyzed based on the coding gain or the block mode of the video block.

The method for changing the size of a video image according to the present invention uses a single fixed filter regardless of the video characteristics by reducing / enlarging the size of the video image using different interpolation filters according to the video characteristics of the video image. Compared to the conventional method of reducing / enlarging the size of a video image, it provides a high quality video image while reducing complexity.

In addition, when the integer DCT is used as an example of the adaptive filter used in the method of reducing / enlarging the size of the video image according to the present invention, the present invention can be applied to the integer DCT used in the H.264 / AVC codec. It is compatible with 264 / AVC codec. In addition, when applied to integer DCT, it is possible to prevent a bit error caused by changing a real operation into an integer operation during encoding / decoding of a video image.

Hereinafter, an apparatus for changing the size of an image image according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

2 is a functional block diagram of an apparatus for changing the size of a video image with adaptive filter selection in accordance with the present invention.

Referring to FIG. 2, the image block converter 10 converts an input image of a spatial region into a frequency component by using a block transformation technique. Preferably, the image block converter 10 divides an image image of an input spatial region into an image block having a predetermined size and converts the divided image block into a frequency component using a block transformation technique. Preferably, the DCT technique or the integer DCT technique is used as an example of the block transform technique.

The image characteristic analyzer 11 analyzes an image characteristic of an input image image from the converted frequency component. As the video characteristic of the video image, the block mode or the coding gain of the video block is used. The image characteristic analyzer 11 analyzes the block mode of the image block by determining whether the image block converted into the frequency component is an inter block predicted in the reference image or an intra block that is poorly predicted in the reference image, The coding energy of the image block is analyzed by determining the density of coefficient energy of the image block transformed into frequency components.

The block mode of the picture block is determined using the bit-distortion cost for each macroblock 16x16. The inter mode is determined by determining a motion vector and a reference picture before calculating the bit-distortion cost, and then using a minimum value of an equation using a sum absolute distortion (SAD) between the reconstructed picture and the original picture according to the motion vector and the reference picture. In the intra mode, the intra prediction direction is determined and the optimal mode is determined for the macroblock, and then the minimum value of the equation using the SSD (Sum Square Distortion) between the original image and the reconstructed image is determined. The block mode of each macroblock is determined as the mode having the smaller minimum value between the minimum value of the inter mode and the minimum value of the intra mode. For example, when the minimum value of the inter mode of the macroblock is smaller than the minimum value of the intra mode, the macro block is determined to be an inter mode, and the image block constituting the macro block is determined to be an inter block.

The interpolation filter selector 12 adaptively selects an interpolation filter applied to each image block according to the characteristics of the image block analyzed by the image characteristic analyzer 11. That is, the interpolation filter selector 12 selects different types of interpolation filters for reducing or enlarging the image block according to the block mode and the coding gain of the image block.

On the other hand, the interpolator 13 reduces or enlarges the size of the video image converted into the frequency component by the image block converter 10 using the interpolation filter selected by the interpolation filter selector 12. The encoder 14 compresses the data of the reduced or enlarged video image through quantization and compresses the data of the compressed video image in a predetermined manner, for example, predictive coding, variable length coding, Coded according to arithmetic coding or the like to generate a bitstream of the video or video.

3 is a flowchart illustrating a method of changing a size of a video image according to an embodiment of the present invention.

Referring to FIG. 3, when an image image of a spatial domain is input, the input image image is divided into image blocks having a predetermined size (step 1), and the divided image blocks are converted into frequency components using a block conversion technique ( Step 2). As an example of the block transform scheme, DCT or integer DCT may be used.

An image characteristic of each image block converted into frequency components is analyzed using block mode or encoding gain (step 3), and an interpolation filter for reducing or enlarging each image block using the image characteristic of each analyzed image block is performed. Select adaptively (step 4).

A method of selecting an interpolation filter according to an image characteristic of each image block analyzed by dividing the image image into a case of reducing and enlarging the image is described below.

<How to select interpolation filter to reduce video image>

Converting an image block into a frequency component using a block transform technique and preserving a lot of low frequency components to reduce the size of the image block when the block mode of the image block converted to the frequency component is an intra mode during video encoding. Select an interpolation filter. On the other hand, when the block mode of an image block converted into a frequency component using the block transformation technique and the frequency component is an inter mode in video encoding, the block mode is an intra mode to reduce the size of the image block. The interpolation filter which preserves a relatively high frequency component is selected rather than the case of.

On the other hand, if the coding gain of the image block calculated from the transformed frequency component is converted to a frequency component using a block transform technique and is smaller than the first threshold value, the low frequency component is preserved to reduce the size of the image block. Select an interpolation filter. On the other hand, if the encoding gain of the image block is converted to a frequency component using a block transform technique and the image block calculated from the transformed frequency component is larger than the first threshold value, the encoding gain of the image block is reduced to reduce the size of the image block. An interpolation filter is selected that preserves more high frequency components than is smaller than the first threshold.

<How to select an interpolation filter to enlarge the video image>

Interpolation that preserves a lot of low-frequency components in order to enlarge the size of the image block when the block mode of the image block is converted to frequency component using the block transform technique and the image block is converted to intra mode during video encoding. Select a filter. On the other hand, when the block mode of an image block converted into a frequency component using the block transform technique and the frequency component is an inter mode in video encoding, the block mode is an intra mode to enlarge the size of the image block. The interpolation filter which preserves a relatively high frequency component is selected rather than the case of.

On the other hand, if the coding gain of the image block calculated from the transformed frequency component by converting the image block into a frequency component using a block transform technique is smaller than the first threshold value, a large number of low frequency components are preserved to enlarge the size of the image block. Select an interpolation filter. On the other hand, when the encoding gain of the image block calculated from the transformed frequency component by converting the image block into the frequency component using the block transform technique is larger than the first threshold value, the encoding gain of the image block is reduced to increase the size of the image block. Choose an interpolation filter that preserves more high frequency components than if it is less than one threshold.

A detailed example of selecting an interpolation filter according to the image characteristics of each image block analyzed by dividing the image image into the case of reducing and enlarging the image will be described below.

<Example 1 of interpolation filter for reducing video image>

For example, when the block mode of an image block converted to a frequency component using a 4 × 4 integer DCT technique and the frequency block is intra mode is the following mode that preserves a lot of low frequency components to reduce the size of the image block. The interpolation filter of equation (1) is selected. On the other hand, when the block mode of the image block converted to the frequency component by the 4x4 integer DCT technique and the frequency block is converted to the frequency component is the inter mode, the high frequency than the block mode is the intra mode to reduce the size of the image block. An interpolation filter of Equation (2) below, which preserves a relatively large amount of components, is selected.

[Equation 1]

[Equation 2]

On the other hand, when the coding gain of the image block calculated from the transformed frequency component by converting the image block to the frequency component using a 4x4 integer DCT technique is smaller than the first threshold value, a large number of low frequency components are preserved to reduce the size of the image block. An interpolation filter of Equation (1) is selected. On the other hand, when the encoding gain of the image block is calculated from the transformed frequency component by the 4 × 4 integer DCT technique and calculated from the transformed frequency component is larger than the first threshold value, the coding gain of the image block to reduce the size of the image block An interpolation filter of Equation (2) above is selected, which preserves a relatively high frequency component than when it is smaller than the first threshold.

<Example 1 of interpolation filter to enlarge video image>

For example, when a block of an image block converted to a frequency component using a 4x4 integer DCT technique and the block mode of the image block converted to a frequency component is an intra mode, the following low frequency components are preserved in order to enlarge the size of the image block. The interpolation filter of equation (3) is selected. On the other hand, when the block mode of an image block converted into a frequency component using a 4 × 4 integer DCT technique and converted into a frequency component is an inter mode, a high frequency component is used to increase the size of the image block than when the block mode is an intra mode. The interpolation filter of Equation (4) below, which preserves a relatively large number, is selected.

[Equation 3]

[Equation 4]

On the other hand, if the coding gain of the image block calculated from the transformed frequency component by transforming the image block into a frequency component using a 4x4 integer DCT technique is less than the first threshold value, a large number of low frequency components are preserved to enlarge the size of the image block. An interpolation filter of Equation (3) is selected. On the other hand, when the encoding gain of the image block is calculated from the transformed frequency component by the 4 × 4 integer DCT technique and calculated from the transformed frequency component is larger than the first threshold value, the coding gain of the image block to reduce the size of the image block An interpolation filter of Equation (4) is selected which preserves a relatively high frequency component more than the case where it is smaller than this first threshold.

Preferably, the interpolation filters of Equations (1), (2), (3) and (4) can be used as integers.

<Example 2 of interpolation filter for reducing video image>

For example, when the image block is transformed into a frequency component using an 8 × 8 DCT technique and the block mode of the image block converted into the frequency component is an intra mode, the following mathematics of preserving a lot of low frequency components to reduce the size of the image block is shown below. The interpolation filter of equation (5) is selected. On the other hand, when the block mode of an image block converted to a frequency component using the 8 × 8 DCT technique and the frequency component is inter mode is used to reduce the size of the image block, a higher frequency component is used than the intra mode. An interpolation filter of Equation (6) below, which is relatively saved, is selected.

[Equation 5]

[Equation 6]

On the other hand, the image block is converted into a frequency component using an 8 × 8 DCT technique, and when the coding gain of the image block calculated from the converted frequency component is smaller than the first threshold value, the low frequency component is preserved to reduce the size of the image block. The interpolation filter of Equation (5) is selected. On the other hand, if the encoding gain of the image block calculated from the transformed frequency component by the 8 × 8 DCT technique and calculated from the transformed frequency component is larger than the first threshold value, the encoding gain of the image block is reduced to reduce the size of the image block. An interpolation filter of Equation (6), which preserves a relatively high frequency component more than the case where it is smaller than the first threshold value, is selected.

<Example 2 of interpolation filter for enlarging video image>

For example, when the image block is transformed into a frequency component using an 8 × 8 DCT technique and the block mode of the image block converted into the frequency component is an intra mode, the following mathematical equation preserves a lot of low frequency components to enlarge the size of the image block. The interpolation filter of equation (7) is selected. On the other hand, when the block mode of an image block converted to a frequency component using the 8 × 8 DCT technique and the frequency component is an inter mode, a high frequency component is used to increase the size of the image block than when the block mode is an intra mode. The interpolation filter of Equation (8) below which is relatively saved is selected.

[Equation 7]

[Equation 8]

On the other hand, the image block is converted into a frequency component using an 8x8 DCT technique, and when the coding gain of the image block calculated from the transformed frequency component is smaller than the first threshold value, the low frequency component is preserved in order to enlarge the size of the image block. The interpolation filter of Equation (7) is selected. On the other hand, if the encoding gain of the image block calculated from the transformed frequency component by the 8 × 8 DCT technique and calculated from the transformed frequency component is larger than the first threshold value, the encoding gain of the image block is reduced to reduce the size of the image block. An interpolation filter of Equation (8) is selected, which preserves a relatively high frequency component than when it is smaller than the first threshold.

The image block converted to the frequency component is reduced or enlarged using an interpolation filter selected according to the image characteristic (step 5). The image data of the image block reduced or enlarged by the selected interpolation filter is compressed through quantization, and the data of the compressed image image is predicted, variable length coding, arithmetic coding, or the like. It is encoded to generate a bitstream of an image or a video (step 6).

Meanwhile, the above-described embodiments of the present invention can be written as a program that can be executed in a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium. The computer-readable recording medium may be a magnetic storage medium (for example, a ROM, a floppy disk, a hard disk, etc.), an optical reading medium (for example, a CD-ROM, DVD, etc.) and a carrier wave (for example, the Internet). Storage medium).

Although the present invention has been described with reference to the embodiments shown in the drawings, this is merely exemplary, and it will be understood by those skilled in the art that various modifications and equivalent other embodiments are possible. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

Figure 1 shows a functional block diagram of a device for changing the size of a conventional video image.

2 is a functional block diagram of an apparatus for changing the size of a video image with adaptive filter selection in accordance with the present invention.

3 is a flowchart illustrating a method of changing a size of a video image according to an embodiment of the present invention.

Claims (12)

Block converting the input image image of the spatial domain into a frequency component; Analyzing image characteristics of the image image converted to the frequency component; And Selecting an interpolation filter for reducing or enlarging the size of the image image based on the image characteristic of the analyzed image image, Selecting the interpolation filter, When the block mode of the video image is an intra mode, an interpolation filter for preserving low frequency components is selected to reduce or enlarge the size of the video image, When the block mode of the video image is an inter mode, an interpolation filter for further preserving high frequency components is selected as compared to a filter for preserving the low frequency components in order to reduce or enlarge the size of the video image. How to change the size of. delete delete delete delete delete delete delete delete delete delete delete

Images