KR101522260B1 - Method for scaling a resolution and an apparatus thereof - Google Patents

Abstract

According to an embodiment of the present invention, a method for scaling resolution of an image signal comprises the steps of: receiving the decoded image signal based on the encoded beat stream; selecting a scaling method based on the decoded image signal; operating scaling by corresponding to a first direction and a second direction of the image signal according to the selected scaling method; and outputting an image in which resolution is converted according to scaling.

Description

TECHNICAL FIELD [0001] The present invention relates to a resolution conversion method and apparatus,

The present invention relates to a resolution conversion method and apparatus, and more particularly, to a resolution conversion method and apparatus for improving resolution conversion performance by changing a screen resolution using a cubic convolutional scaler.

Recently, as the demand for high resolution and high definition video has increased, there has been a need for a highly efficient video compression technology for the next generation video service. JVT, which was jointly organized by MPEG and VCEG, started standardization in the late 1990s and completed H.264 / AVC in 2004 in response to market demand. H.264 / AVC encoders and decoders are similar in structure to previous standard codecs, but with the addition of sub-modules with improved performance inside the codec. In January 2010, MPEG and VCEG formed a Joint Collaborative Team on Video Coding (JCT-VC). In January 2013, JCT-VC established a next-generation video standard technology called HEVC (High Efficiency Video Coding) Respectively. The HEVC has a compression efficiency of about 50% or more when compared from the viewpoint of subjective image quality as compared with the H.264 / AVC High profile, which is conventionally known as having the highest compression efficiency.

Meanwhile, the resolution conversion device for converting the screen resolution may scale the input image to enlarge or reduce the resolution. Various techniques such as sample and hold, Bilinear, Cubic B-spline, and Cubic convolution have been proposed as various scalers applied to the resolution converter. A 4 tap Cubic convolution scaler technique consisting of four filter taps can be exemplified as a typical technique widely used among these techniques. The 4 tap Scaler can be applied to the horizontal and vertical directions of the image. And, 4 tap Scaler can use the method of acquiring the resolution converted image by applying the basis functions of the 4 tap Cubic convolution kernel by associating the width and the vertical axis width with four digital samples.

* Such a 4 tap Cubic convolution scaler has been used in many research and development. However, according to the tendency that the characteristic and resolution of the video signal are greatly improved, the performance is limited only by the 4 tap scalar, and as a result, the image quality of the resolution converted image is lowered.

An embodiment of the present invention is directed to improving the image quality and performance of a resolution conversion module using an improved kernel.

Another aspect of the present invention is to improve the image quality and performance of the resolution conversion module according to adaptive kernel application using pixel values of an image.

Another aspect of the present invention is to improve image quality and performance of a resolution conversion module by applying an optional scaler according to image analysis.

Another aspect of the present invention is to improve the image quality and performance of a resolution conversion module by applying an optional scaler according to a scaling direction.

The technical problem to be solved by the embodiments of the present invention is not limited to the above technical problems, and other technical problems may exist.

According to an aspect of the present invention, there is provided a method of scaling a resolution of an image signal, the method comprising: receiving a decoded image signal based on an encoded bitstream; Selecting a scaling method based on the decoded video signal; Performing adaptively scaling corresponding to a first direction and a second direction of the video signal according to the selected scaling scheme; And outputting a resolution-converted image according to the scaling.

According to another aspect of the present invention, there is provided an apparatus for converting a resolution of a video signal, the apparatus comprising: an input unit for receiving a video signal decoded based on an encoded bitstream; At least one scaler for selecting a scaling method based on the decoded video signal and performing adaptive scaling corresponding to a first direction and a second direction of the video signal according to the selected scaling method; And an output unit for outputting a resolution-converted image according to the scaling.

An embodiment of the present invention can improve the image quality and performance of a resolution conversion module by using more samples for resolution conversion using an improved 6 tap kernel.

In addition, according to another embodiment of the present invention, image quality and performance of the resolution conversion module can be improved by adapting an appropriate kernel using a pixel value of an image.

In addition, according to another embodiment of the present invention, the image quality and performance of the resolution conversion module can be improved by selectively applying the most optimized scaler according to the image analysis.

Meanwhile, another embodiment of the present invention can improve the image quality and performance of the resolution conversion module by applying different scalers for the horizontal and vertical according to the scaling direction.

The present invention has been made in view of the above problems, and it is an object of the present invention to provide a method and apparatus for scaling a resolution of a video signal having a resolution higher than that of a current image, , The effect of this patented technology in such an environment will appear very efficiently.

1 is a block diagram illustrating a decoder including a resolution conversion apparatus according to an embodiment of the present invention.
2 is a diagram for explaining a configuration of a resolution conversion apparatus according to an embodiment of the present invention.
FIGS. 3 and 4 are views for explaining the scaling method according to the first embodiment of the present invention in more detail.
5 and 6 are views for explaining an adaptive scaling method according to an image pixel value according to a second embodiment of the present invention.
7 is a flowchart for explaining a selective scaling method according to a third embodiment of the present invention.
8 is a block diagram illustrating a configuration of a resolution conversion apparatus according to the fourth embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. It should be understood, however, that the present invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In the drawings, the same reference numbers are used throughout the specification to refer to the same or like parts.

Throughout this specification, when a part is referred to as being "connected" to another part, it is not limited to a case where it is "directly connected" but also includes the case where it is "electrically connected" do.

Throughout this specification, when a member is located on another member, this includes not only when a member is in contact with another member but also when another member is present between the two members.

Throughout this specification, when an element is referred to as "including " an element, it is understood that the element may include other elements as well, without departing from the other elements unless specifically stated otherwise. The terms "about "," substantially ", etc. used to the extent that they are used throughout the specification are intended to be taken to mean the approximation of the manufacturing and material tolerances inherent in the stated sense, Accurate or absolute numbers are used to help prevent unauthorized exploitation by unauthorized intruders of the referenced disclosure. The word " step (or step) "or" step "used to the extent that it is used throughout the specification does not mean" step for.

Hereinafter, a resolution conversion method and apparatus according to an embodiment of the present invention will be described.

FIG. 1 is a block diagram of an image decoding apparatus including a resolution converting apparatus according to an embodiment of the present invention. FIG. 1 shows an image decoding apparatus preceded by H.264 decoding. The same structure can be applied to decoding for other codecs.

Referring to FIG. 1, a unit for processing data in a decoding apparatus may be a macroblock having a size of 16 x 16 pixels, and may be a coding unit of a predetermined unit. The decoding apparatus receives the bitstream and decodes the bitstream into an intra mode or an inter mode to output a reconstructed image.

In the intra mode, the switch is switched to the intra mode, and in the inter mode, the switch is switched to the inter mode. The main flow of the decoding process is to generate a reconstructed block by adding a block and a prediction block as a result of decoding a bitstream after generating a prediction block.

First, the generation of the prediction block is performed according to the intra mode and the inter mode. In case of the intra mode, spatial prediction is performed using the already encoded neighboring pixel values of the current block in the intra prediction process to generate a prediction block. In the inter mode, the motion vector is stored in the reference image buffer A prediction block is generated by searching a region in a reference image and performing motion compensation.

The entropy decoding unit performs entropy decoding of the input bitstream according to a probability distribution and outputs a quantized coefficient. The decoding apparatus generates a reconstructed block by performing a dequantization process and an inverse transform on a quantized coefficient to generate a reconstructed block through a prediction image and an adder, and then filters the block through at least one of a deblocking filter or a sample adaptive offset filter, And stores it in the reference image buffer at the same time.

Thereafter, the reconstructed and decoded image is input to the resolution conversion apparatus 100.

The resolution conversion apparatus 100 receives the decoded image as an input image, and outputs the enlarged or reduced image by converting (scaling) the resolution. In particular, according to the first embodiment of the present invention, the scaler used in the resolution conversion apparatus can be a scaler using a 6-tap based cubic convolution kernel. Accordingly, resolution conversion efficiency and conversion image quality can be improved by performing screen resolution conversion and interpolation, which will be described later.

In addition, according to the second embodiment of the present invention, the scaler used in the resolution conversion apparatus 100 can use a kernel determined adaptively according to the image pixel value. For example, the scaler may assign different tuning parameter values according to the variance values for a portion of the image pixels, and use an adaptive cubic convolution kernel based thereon.

In addition, according to the third embodiment of the present invention, the scaler used in the resolution conversion apparatus 100 can calculate the efficiency for each blocked image area and selectively determine any one of 4-tap and 6-tap kernels . For example, the scaler may calculate a cost value for each area, and selectively apply a kernel of either a 4-tap or a 6-tap cubic convolution kernel based on the cost value.

Meanwhile, according to the fourth embodiment of the present invention, the scaler used in the resolution conversion apparatus 100 uses a 6-tap kernel for scaling an image in a first direction and uses a different kernel for scaling in a second direction . For example, a scaler can apply a 6-tap kernel in the horizontal direction, and a 4-tap kernel in the vertical direction can increase the computation efficiency.

Hereinafter, the resolution conversion process of the resolution conversion apparatus 100 according to the embodiment of the present invention will be described in more detail with reference to FIG. 2 to FIG.

Referring to FIG. 2, the resolution conversion apparatus 100 may perform resolution conversion in the following steps.

The resolution conversion apparatus 100 may include a horizontal direction scaler 101 and a vertical direction scaler 102.

According to the first embodiment of the present invention, the horizontal direction scaler 101 and the vertical direction scaler 102 can perform resolution conversion using a six-tap cubic convolution scaling method, respectively.

More specifically, in order to apply the 6-tap cubic convolutional scaler according to the first embodiment of the present invention, each of the scalers 101 and 102 extracts digital samples with six filter taps and outputs digital samples for the horizontal and vertical directions Scaling the samples and interpolating the pixels located therebetween.

을 적용할 수 있다.This may mean that each scaler 101,102 performs interpolation using six pixel information around the interpolation position for interpolation. In order to interpolate using the 6-tap cubic convolution method, each of the scalers 101 and 102 converts the filter kernel

Can be applied.

는

를 이용하여 얻어진 연속신호

의 연속성등을 이용해서 계산될 수 있으며 다음 수학식 2와 같이 표현될 수 있다.Unknown in equation (1)

The

Lt; RTI ID = 0.0 >

And can be expressed by the following equation (2). &Quot; (2) "

의 모양을 결정하는 튜닝 파라미터로써 α = 0.083으로 설정한

를 도식화하면 도 3에 도시된 바와 같다. 일반적으로 알파 값(α)은 선택 가능한 튜닝 파라미터일 수 있으며, 알파 값 선택에 따라 해상도 변환된 영상의 엣지 부분 표현 강도가 가변될 수 있다.The alpha (alpha) value of equation

0.0 > = 0.083 < / RTI >

As shown in FIG. 3. In general, the alpha value alpha may be a selectable tuning parameter, and the edge partial intensity of the resolution-converted image may be varied according to the alpha value selection.

를 이용하여 연속함수

를 계산하는 식은 다음의 수학식 3과 같다. 여기서 보간 위치는

와

사이인

로 가정될 수 있다.On the other hand,

And the continuous function

Is expressed by Equation (3). &Quot; (3) " Here, the interpolation position is

Wow

Saiin

. ≪ / RTI >

Accordingly, interpolation can be performed using the six pixels closest to each other at the position to be interpolated.

F (x_ (k-1)), f (x_k), f (x_ (k + 1)) and f ), and f (x_ (k + 3)). If the distance from the position x_ {k} + t to x_ {k} to be interpolated is t, 0? T? 1 can be obtained.

를 사용하는 6 탭 큐빅 컨볼루션 스케일러를 세로 방향으로 적용하여 해상도 스케일링을 수행할 수 있다. 그리고, 가로 방향 및 세로 방향 스케일링이 종료되면, 해상도 변환 장치(100)는 해상도 변환된 영상을 raw 데이터의 형태로 프레임 단위 또는 실시간으로 출력할 수 있다.In this way, the horizontal direction scaler 101 can apply 6-tap cubic convolution scaling in the horizontal direction to all pixels of a given image signal. Then, the vertical direction scaler 102, based on the data of the horizontally scaled signal,

A tap can be applied in the vertical direction to perform resolution scaling. When the horizontal and vertical scaling ends, the resolution conversion apparatus 100 can output the resolution-converted image in units of frames or in real time in the form of raw data.

FIG. 4 shows a process of obtaining an analog signal from a given digital signal using a 6-tap cubic convolutional scaler derived through the above process. In FIG. 4, a single-axis reference is shown for convenience. In contrast to the conventional 4-tap cubic convolution scaler, the resolution of the video signal is finely changed by using a 6-tap basis kernel, And the image quality of the resolution conversion image can be expected to be improved.

5 and 6 are diagrams for explaining an adaptive scaling method according to an image pixel value according to an embodiment of the present invention.

As described above, each tuning parameter value may exist in the 4-tap cubic convolutional scaler and the 6-tap cubic convolutional scaler. Generally, this value is defined as a fixed value by being defined as an alpha value. However, in the second embodiment of the present invention, a method of selecting an alpha value optimized for the local characteristic of a pixel value is proposed.

Accordingly, as shown in FIG. 5, the resolution conversion apparatus 100 according to the second embodiment of the present invention divides an image into respective M x M blocks, and in a cubic convolution for each divided area , Different tuning parameters alpha can be applied. That is, the scaler of the resolution conversion apparatus 100 may apply a kernel function based on different alpha values for each block. In this case, the shape of the block may be a square having the same length and the same length, and may be a rectangle having a different length and a different length.

At this time, there are various ways to consider the characteristics of each block. According to one embodiment, the alpha value can be selected using the variance value of the intra-block pixel values.

FIG. 6 is a flowchart for explaining the resolution conversion method according to the second embodiment of the present invention.

Referring to FIG. 6, the resolution conversion apparatus 100 receives an original image (S101).

The original image may include already decoded image data, as described above.

Then, the resolution conversion apparatus 100 blocks the image (S103).

The resolution conversion apparatus 100 can block an image by, for example, dividing each frame of the image into M x M blocks. In this case, the shape of the block may be a square having the same length and the same length, and may be a rectangle having a different length and a different length. Each blocked image can be individually subjected to resolution conversion in a different manner. When the image is blocked, the processing order corresponding to each block can be determined, and the resolution conversion can be sequentially performed according to the order.

Thereafter, the resolution conversion apparatus 100 calculates a variance value corresponding to the current block (S105).

In the embodiment of the present invention, the resolution conversion apparatus 100 may calculate the variance value of the pixel values using the pixel values of the pixels in the block corresponding to the current block.

Then, the resolution conversion apparatus 100 acquires an alpha value corresponding to the calculated dispersion value (S107).

The resolution conversion apparatus 100 may obtain an optimal alpha value corresponding to the variance value based on a table stored in advance. For this, the resolution conversion apparatus 100 may include a separate storage unit, and at least one alpha value table may be included in the storage unit.

In addition, a plurality of alpha value tables may exist. For example, as shown in FIG. 5, when the scaler is adaptively applied according to the divided area, the alpha value table for obtaining the variance value may be different. For example, the resolution converter 100 may include a 4-tap alpha value table corresponding to a 4-tap convolutional scaler, and a 6-tap alpha value table corresponding to a 6-tap convolutional scaler may be separately included. As described above, since the expression intensity or the like of an edge may differ depending on each alpha value, the designer of the resolution conversion apparatus 100 can designate an appropriate alpha value in advance as a table.

Then, the resolution conversion apparatus 100 performs resolution conversion using the obtained alpha value (S109).

As described above, when the alpha value is acquired, the resolution conversion apparatus 100 can generate a kernel function using the kernel function, and performs cubic convolution scaling using the generated kernel function to generate a resolution-converted image block . In addition, the alpha value may be differently applied depending on whether the kernel function is a general 4-tap kernel or a 6-tap kernel according to the present invention.

Thereafter, the resolution conversion apparatus 100 determines whether the current block is the last block (S111). If the current block is the last block, the resolution conversion apparatus 100 outputs the resolution-converted image (S113).

The image converted by the resolution conversion apparatus 100 may be output frame by frame or block by block. On the other hand, if the resolution conversion apparatus 100 is not the last block, the resolution conversion apparatus 100 may perform the operation again from the dispersion calculation step (S105) corresponding to the next block.

Thus, by applying a kernel using different alpha values according to the characteristics of the image block, the image quality can be improved. Although the intra-block variance value is used in the embodiment of the present invention, this is only an example, and it is obvious that the present invention does not limit the method of classifying the feature of the image block. In addition, the present invention can be applied to both cases where a 4-tap cubic convolutional scaler or a 6-tap cubic convolutional scaler is used, and can be extended even when adaptively applied by having a plurality of tables.

7 is a flowchart for explaining a selective scaling method according to a third embodiment of the present invention.

In the selective scaling method according to the third embodiment of the present invention, a method of selectively using an optimal scaling method for each block is proposed.

Referring to FIG. 7, the resolution conversion apparatus 100 receives an original image (S201). The original image may include already decoded image data, as described above.

Then, the resolution conversion apparatus 100 blocks the image (S203).

The resolution conversion apparatus 100 can block an image by, for example, dividing each frame of the image into N x N blocks. In this case, the shape of the block may be a square having the same length and the same length, and may be a rectangle having a different length and a different length. Each blocked image can be individually subjected to resolution conversion in a different manner. When the image is blocked, the processing order corresponding to each block can be determined, and the resolution conversion can be sequentially performed according to the order.

Then, the resolution conversion apparatus 100 performs cost calculation to acquire J4 and J6 (S204), and compares the two values to determine whether J4 is smaller or equal (S205).

Here, the resolution conversion apparatus 100 according to the third embodiment of the present invention can selectively determine an optimal scaling method for each block. In particular, a more suitable scaling scheme, preferably a 4-tap cubic convolutional scaler and a 6-tap cubic convolutional scaler, may be selected.

At this time, it may be different depending on which scaler is used in each block, that is, which scaler is optimal. For example, a scaler with a smaller cost value can be selected after comparing cost values J4 and J6 calculated for each block. Here J4 is the cost value when the corresponding block is a 4 tap Cubic convolution scaler, and J6 is the cost value when the 6 tap Cubic convolution scaler is used. In addition, J4 and J6 may be defined as distortions that occur when the corresponding block is resolved by 4 tap or 6 tap Cubic convolution scaler.

In addition, the cost function for J4 and J6 can be determined based on at least one of image quality, calculation amount, and degree of distortion. Also, a weight for each variable may be set, and a new cost function for resolution conversion generated based thereon may be defined. Therefore, in the embodiment of the present invention, the condition of the cost function is not limited, and it is possible to determine whether to perform an optimal scaling method in various ways.

If J4 is less than or equal to J6, the resolution conversion apparatus 100 may perform enlargement or reduction using the 4-tap cubic convolutional scaler (S207).

If J6 is smaller than J4, the resolution conversion apparatus 100 may perform enlargement or reduction using the 6-tap cubic convolutional scaler (S209).

Thereafter, the resolution conversion apparatus 100 determines whether the current block is the last block (S211). If the current block is the last block, the resolution conversion apparatus 100 outputs the resolution-converted image (S213).

The image converted by the resolution conversion apparatus 100 may be output frame by frame or block by block. On the other hand, if the resolution conversion apparatus 100 is not the last block, the resolution conversion apparatus 100 can perform the operation again from the step S204 of calculating the cost J4 and J6 of the next block.

In this way, the optimal scaler for each block can be selected, thereby improving not only the image quality of the resolution-converted image, but also optimizing the calculation performance and image quality according to the cost function design.

8 is a block diagram for explaining the configuration of the resolution conversion apparatus 100 according to the fourth embodiment of the present invention.

Referring to FIG. 8, the resolution conversion apparatus 100 according to the fourth embodiment of the present invention may include a horizontal six-tap scaling unit 103 and a vertical four-tap scaling unit 105. [

According to the fourth embodiment of the present invention, the resolution conversion apparatus 100 can perform 6-tap cubic convolution scaling as described above only in the horizontal direction. In the vertical direction, 4-tap cubic scaling can be performed.

According to the embodiment of the present invention, when the resolution of an image is converted, the result is converted into a 6-tap cubic convolution scaler having generally good performance in the horizontal direction and a 4-tap cubic convolution scaler in the vertical direction .

Such an embodiment of the present invention can be advantageously applied in hardware implementation. Generally, resolution conversion in the horizontal direction can be performed using one line memory regardless of the number of taps.

However, since a plurality of line memories must be used according to the number of taps in the resolution conversion process in the vertical direction, it is necessary to use four line memories (6-tap cubic convolution scaler) tap memory can reduce the memory requirement when the 4-tap cubic convolutional scaler uses only vertical memory.

Therefore, according to the embodiment of the present invention, the resolution conversion apparatus 100 includes the 6-tap cubic convolution scaling unit 103 only in the horizontal direction, so that the performance can be improved while maintaining the line memory requirement.

The method according to the present invention may be implemented as a program for execution on a computer and stored in a computer-readable recording medium. Examples of the computer-readable recording medium include a ROM, a RAM, a CD- , A floppy disk, an optical data storage device, and the like, and may also be implemented in the form of a carrier wave (for example, transmission over the Internet).

The computer readable recording medium may be distributed over a networked computer system so that computer readable code can be stored and executed in a distributed manner. And, functional programs, codes and code segments for implementing the above method can be easily inferred by programmers of the technical field to which the present invention belongs.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It should be understood that various modifications may be made by those skilled in the art without departing from the spirit and scope of the present invention.

Claims (10)

A method of scaling a resolution of a video signal,
Receiving a decoded video signal or a digitally processed video signal based on the encoded bitstream;
Selecting a scaling scheme based on the video signal;
Performing adaptively scaling corresponding to a first direction and a second direction of the video signal according to the selected scaling scheme; And
And outputting a resolution-converted image according to the scaling,
The step of selecting the scaling scheme
Dividing the video signal into blocks;
Calculating a variance value corresponding to the block according to a pixel value included in the divided block;
Obtaining an alpha value corresponding to the calculated variance value; And
Generating a kernel corresponding to the obtained alpha value, and selecting a scaler corresponding to the kernel by the scaling method. The method according to claim 1,
And storing a scaler-specific alpha value table for obtaining the alpha value. The method according to claim 1,
Wherein the alpha value is a selectable tuning parameter, and the intensity of the edge partial representation of the resolution-converted image is varied according to the alpha value selection. The method according to claim 1,
Wherein the scaler corresponding to the kernel is a 4-tap or 6-tap cubic convolution scaler. The method according to claim 1,
Wherein the first direction is a horizontal direction corresponding to the video signal and the second direction is a vertical direction corresponding to the video signal. An apparatus for converting a resolution of a video signal,
An input unit for receiving a decoded video signal or a digitally processed video signal based on the encoded bit stream;
At least one scaler for selecting a scaling method based on the video signal and performing adaptive scaling according to a first direction and a second direction of the video signal according to the selected scaling method; And
And an output unit for outputting a resolution-converted image according to the scaling,
Wherein the scaler divides the image signal, calculates a variance value corresponding to the block according to a pixel value included in the divided block, obtains an alpha value corresponding to the calculated variance value, And generates a kernel corresponding to the alpha value, and selects the scaling method corresponding to the kernel. The method according to claim 6,
And a storage unit for storing a scaler-specific alpha value table for obtaining the alpha value. The method according to claim 6,
Wherein the alpha value is a selectable tuning parameter and the edge partial intensity of the resolution-converted image is varied according to the alpha value selection. The method according to claim 6,
Wherein the scaler corresponding to the kernel is a 4-tap or 6-tap cubic convolution scaler. A recording medium on which a program for causing a computer to execute the method according to any one of claims 1 to 5 is recorded.

Images