KR101259063B1 - A image processing device and blocking effect removal method of the same - Google Patents

Abstract

The present invention relates to an image processing apparatus and an apparatus for removing blocking phenomenon thereof, wherein an image processing apparatus according to an embodiment of the present invention comprises an 8x8 encoder and a dequantizer, an inverse quantizer, an inverse quantizer, It is comprised by the decoder provided with a block dispersion part.

The block spreader corresponds to an 8x8 progression unit for storing the decoded video signal in units of blocks, and a target pixel for compensating the luminance difference value and the luminance value between blocks through the video signal stored in the 8x8 progression unit. And a control unit for determining a compensation variable, a multiplier for determining a compensation value through the compensation variable, and an adder for adding the target pixel and the compensation value.

According to an embodiment of the present invention, there is provided a method of removing a blocking phenomenon of an image processing apparatus. Determining a value, and adding the compensation value to the luminance value of the target pixel.

Therefore, the video signal can be processed in real time without undergoing the repeated filtering process through the filter, and the block dispersing phenomenon can be efficiently prevented by compensating the luminance difference between blocks with a simple block dispersing unit.

Description

Image processing device and blocking effect removal method of the same}

1A and 1B are block diagrams schematically illustrating the configuration of a general encoder and a decoder.

2 is a graph for explaining a signal obtained by performing repeated low pass filtering on a video signal and a signal obtained by performing repeated low pass filtering.

3A and 3B are block diagrams schematically illustrating a configuration of an image display apparatus according to an embodiment of the present invention.

FIG. 4 is a block diagram specifically illustrating a block spreader of the decoder of FIG. 3B.

FIG. 5 is a schematic diagram for describing a method of processing an image signal in an 8 × 8 progression unit of the block dispersion unit of FIG. 4.

FIG. 6 is a flowchart for describing a method of removing a block phenomenon of the block spreader of FIG. 4.

<Name of main part of drawing>

80: decoder 82: 8 × 8 progression section

84 control unit 86 multiplier

88: adder

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly, to an image processing apparatus for removing a blocking phenomenon occurring during quantization and dequantization during image signal compression and reconstruction.

Recently, a large amount of analog information has been converted to digital, and various video signal compression and decompression techniques have been proposed to efficiently use a limited data transmission bandwidth when transmitting / receiving or storing video information having a very large size.

Here, the compression of the video signal means removing redundancy inherent in the video signal. The redundancy inherent in the video signal includes a temporal redundancy in which the front screen and the back screen are equal to each other except for an active part. In addition, there are spatial redundancy that is similar to each other among neighboring pixels even in a screen, and statistical redundancy that a frequency of occurrence of a specific value is high or low after encoding.

In general, as a method for removing such redundancy, a discrete cosine transform (DCT) process is used. The DCT process is a process of converting a signal of a time sequence into a signal of a frequency sequence, and a pixel value of a time sequence. Is a process of converting the signal into a frequency series signal according to a matrix operation. The DCT is based on the high correlation of the video signal in the spatial direction. The DCT concentrates the energy dispersed in all the pixels of the image on a few conversion coefficients having a low frequency including a DC component, and then uses the advantages of the DCT. That's the way. Therefore, the DCT data, that is, the DCT coefficients, concentrate the matrix calculation result in the low frequency direction, and then reduce the size of the data through the quantization process.

Quantization is a process for representing each DCT coefficients having a real value with a limited data length. In such a quantization process, quantization noise occurs and data loss occurs. In order to reduce such data loss, the low frequency components of the DCT coefficients in which the data are concentrated are densely quantized, and the high frequency components of the DCT coefficients with little data are sparsely quantized to minimize the data loss to reduce the amount of data. But some loss had to be paid.

1A and 1B schematically illustrate the configuration of a typical encoder and decoder, and as illustrated in FIG. 1A, the encoder 1 may perform an image smoothing process. A filter 2 for removing noise, a DCT unit 3 for converting a time sequence region of a signal into a frequency sequence region, and a quantizer 4 for reducing the size of actual data.

The filter 2 is a general band-limiting filter, and removes high frequency components by passing only low frequency signals included in an input video signal of a predetermined unit, and outputs the low frequency signals from which the high frequency components have been removed to the DCT unit 3. do.

As a means of removing the spatial redundancy, the DCT unit 3 converts the DCT coefficients by block units (8 × 8) by using a high correlation between adjacent pixels in the screen, and the quantizer 4 Will output

The quantizer 4 receives DCT coefficients having a real value and divides the DCT coefficient values into a set of predetermined step intervals to express each frequency component with a limited data length, and finds a set to which each input DCT coefficient value belongs. Generates a bitstring encoded by the representative value of the set.

Through this process, the encoded bitstring is reconstructed into a video signal by the decoder 5. Referring to FIG. 1B, the decoder 5 receives an encoded bitstring and multiplies a specific constant divided during quantization, and an inverse DCT unit 7 for performing an inverse DCT process. It consists of

The reconstructed video signal output through this process generates a blocking effect due to quantization noise while undergoing quantization and inverse quantization. That is, in general, since the DCT process is performed in block units (8 × 8), a block boundary becomes noticeable due to a difference in luminance between adjacent blocks due to quantization and inverse magnetization.

In order to solve this problem, a method for repeatedly low-pass filtering having a plurality of low-pass filter in the filter (2) has been disclosed, the "blocking phenomenon" published in the Journal of the Institute of Electronics Engineers B, No. 32, 1994 It will be described schematically with reference to the paper "Repeating critical low pass filter for removal" as follows.

Referring to the above paper, a repetitive low pass filtering process is used to remove a block phenomenon occurring during encoding, and a threshold value is used to prevent blurring of image quality caused by the repetitive low pass filtering process. Set (T). The equation for this is shown in Equation 1 below.

..............식 1

Equation 1

은 t번 저역여파과정을 거친 출력화소이고, in[i,j]는 저역여파하고자 하는 영상의 화소이며, LPF는 저역여파기이다. 다수개의 저역여파기를 구비하고, t번 저역여파하여 원 영상신호와의 차이가 임계값(T)를 넘으면 반복적인 저역여파를 멈추고 임계값(T)을 넘지 않으면 저역여파를 반복한다.Where output

Is the output pixel after t low pass filtering, in [i, j] is the pixel of the image to be low filtered, and LPF is the low filter. A plurality of low-pass filters are provided, and the low-pass filter is repeated t times, and when the difference with the original video signal exceeds the threshold value T, the repeated low frequency filter stops.

Accordingly, in the repeated critical low pass method, as shown in FIG. 2, unlike the signal that has undergone only the repeated low pass process, the original signal is decoded into a reconstruction signal that is almost the same as that of the original video signal, thereby effectively preventing the blocking phenomenon. Due to the repetitive characteristics, processing time is delayed due to the change of the video signal. Therefore, there is a difficulty in application in the field in which the image is processed in real time.

An image processing apparatus and a method of removing a blocking phenomenon thereof according to the present invention provide an image processing apparatus for removing a blocking phenomenon without undergoing an iterative processing of a low pass filter to prevent a blocking phenomenon occurring during encoding and decoding of an image. .

Hereinafter, an image processing apparatus according to an embodiment of the present invention will be described with reference to FIGS. 3A and 3B.

3A and 3B illustrate an image processing apparatus according to an exemplary embodiment of the present invention. FIG. 3A illustrates an encoder 10 encoding an input original image signal using a DCT scheme, and FIG. 3B illustrates an encoded bitstring. And a decoder 50 for decoding the reconstructed video signal.

The encoder 10 is composed of a DCT unit 20 and a quantizer 30, and the DCT unit 20 is subjected to an iterative threshold low pass process to remove a blocking phenomenon generated during encoding and decoding an original video signal. Instead, the raw video signal is directly input in block units (8x8) and subjected to discrete conversion to convert the video signal of the time component into the DCT coefficient of the frequency component. In addition, the quantizer 30 receives a DCT coefficient having a real value converted into such a form and expresses each frequency component as a limited data length, and divides the DCT coefficient into a set of predetermined step intervals and inputs each input DCT coefficient. Find the set to which the value of belongs and generate a bitstring encoded with the representative value of the set.

The decoder 50 is composed of an inverse quantizer 60, an inverse DCT unit 70, and a block spreader 80. The inverse quantizer 60 inversely quantizes an encoded bitstring. Through 70, the inverse quantized DCT coefficients are reconstructed into a video signal.

The block dispersion unit 80 has an incidence of the block boundary due to the luminance difference in the block boundary that occurs on the vertical and horizontal lines in units of blocks (8 × 8). In order to eliminate the blocking phenomenon, the luminance difference between adjacent blocks is distributed stepwise to four pixels left and right by the vertical block boundary.

FIG. 5 is a block diagram illustrating the block diverter of FIG. 4 in detail, and is controlled through an 8 × 8 progression unit 82 receiving a decoded video signal and pixel information stored in the 8 × 8 progression unit 82. The controller 84 generates a signal, and a multiplier 86 and an adder 88 which receive the control signal and compensate the luminance value of the target pixel.

The 8x8 progression unit 82 receives the decoded video signal from the inverse DCT unit (70 in FIG. 3B) in block units (8x8), and stores it in a form that the control unit 84 can process.

The controller 84 determines the relative position from the boundary between the blocks of the pixels arranged in the 8x8 progression unit 82, extracts the luminance difference between the blocks, and determines the compensation variable.

More specifically, the arrangement of the pixels of the four blocks A, B, C, and D stored in the 8 × 8 progression unit 82 is shown in FIG. 5. Here, the blocking phenomenon occurs in the adjacent vertical block boundary (V) and the horizontal block boundary (H) between each block, and in particular, the blocking phenomenon of the vertical block boundary (V) is higher than that of the horizontal block boundary (H). It is remarkable.

As shown in FIG. 5, the luminance value of a pixel is defined by following formula (2).

..............식 2

Equation 2

In Equation 2, i and j represent rows and columns of pixels stored in the 8 × 8 progression unit 82, and m and n represent relative positions on the horizontal line and the vertical line of the pixel with respect to the block boundary, respectively. . That is, the m and n values have integer values from -4 to 3 around the vertical and horizontal block boundaries V and H, and the m and n values of the first pixel of one block are 0 and 0, respectively.

Accordingly, the luminance difference between blocks in the vertical block boundary V is defined by Equation 3 below.

..............식 3

Equation 3

Preferably, the control unit 84 calculates a difference between luminance values of pixels in i + n rows and j-1 columns and luminance values of pixels in i + n rows and j columns around the vertical block boundary V. It includes.

In addition, the controller 84 determines a relative position on the horizontal line with respect to the pixels input to the 8x8 progression unit 82, generates a power factor K corresponding thereto, and outputs the result to the multiplier 86. do.

The multiplier 86 multiplies the luminance difference of the vertical block boundary V input from the controller 84 with the compensation variable to determine the compensation value. In this case, the closer to the vertical block boundary (V), the greater the compensation, and the farther apart, the smaller the compensation. To this end, the compensation variable is determined as a differential value depending on the relative position with respect to the vertical block boundary, thereby gradually dispersing the luminance difference of the vertical block boundary (V) to each pixel.

으로 정의된다. 여기서 이 보상변수 의 거듭제곱계수(K)는 휘도차를 보상하고자 하는 복원 영상신호의 목적화소와 수직블록경계(V)와의 이격정도이다. 이에 따라, 상기 제어부(84)로부터 입력되는 판별신호에 대응하여, 상기 거듭제곱계수(K)는 1 내지 4의 정수값이 되고, 보상변수가 1/2의 거듭제곱의 형태이므로 이 승산기의 승산과정은 비트쉬프트연산만으로 구현가능하다. 따라서, 상기 승산기(86)는 보상정도계수에 대응하여 수직블록경계(V)에서 블록간의 휘도차를 라이트 비트쉬프트(Right bit shift)하고, 가산기(88)로 출력한다.Preferably, the compensation variable is

Is defined. Here, the power factor K of the compensation variable is the distance between the target pixel and the vertical block boundary V of the reconstructed video signal to compensate for the luminance difference. Accordingly, in response to the discrimination signal input from the controller 84, the power factor K is an integer value of 1 to 4, and since the compensation variable is in the form of a power of 1/2, the multiplier of this multiplier is multiplied. The process can be implemented only by bit shift operation. Accordingly, the multiplier 86 performs a right bit shift on the luminance difference between blocks in the vertical block boundary V in response to the compensation accuracy coefficient, and outputs the result to the adder 88.

In addition, when the compensation variable is not defined in the form of a power of 1/2, the multiplication process is implemented as a general multiplication operation rather than a bit shift operation. In this case, the luminance difference between the blocks and the compensation variable are multiplied in the vertical block boundary. To determine the compensation value and input it to the adder 88.

Thereafter, the adder 88 adds and outputs the luminance value of the target pixel and the compensation value input from the multiplier.

The block dispersion unit of the above-described operation is represented by the following equation (4).

..............식 4

Equation 4

In the above equation, all variables are integers, x is a pixel from which blocking has been removed, f is a target pixel, B is a luminance difference between blocks in a vertical block boundary, m and n are rows and columns, and k is a power factor. Indicates.

Where k is equal to the absolute value of n, if n is less than 0, i.e., if the target pixel is to the left of the vertical block boundary, then n + is greater than or equal to 0, Determined by a value of 1.

In addition, unless the compensation variable is in the form of a power of 1/2, it is natural that the value multiplied by the luminance difference of the vertical block boundary in the compensation value is different.

6 is a flowchart of a blocking phenomenon removing method according to an embodiment of the present invention, the blocking phenomenon removing method according to an embodiment of the present invention comprises the steps of storing the luminance value of the pixel in blocks (S1); Extracting a luminance difference between blocks (S2); Extracting a column value n of the target pixel to compensate for brightness (S3); Defining a power factor according to the position of the target pixel (S4-1 to S4-4); Determining a compensation value (S5); And adding the luminance value of the target pixel to the compensation value (S6).

The step S1 of storing the luminance value of the pixel in block units includes storing the decoded video signal input from the inverse DCT in block units (8x8).

In the extracting of the luminance difference between blocks (S2), the luminance value of the left pixel is subtracted from the luminance value of the right pixel among the pixels adjacent to the block boundary of each row in the image signal stored in the block unit (8x8). It consists of steps.

The step S3 of extracting the column value n of the target pixel to compensate for the luminance value comprises extracting the column value n of the target pixel to compensate for the luminance values from -4 to 3 around the block boundary. .

Defining the power factor k according to the position of the target pixel (S4-1 to S4-4), if the value of the column value n of the extracted target pixel is less than 0, the power factor ( k) is determined to be equal to the absolute value (| n |) of the column value n of the target pixel, and when the column value n of the target pixel is greater than or equal to 0, the column value n of the target pixel. It is determined by the value (n + 1) to add 1 to.

The determining of the compensation value (S5) includes performing a bit shift circuit in the block boundary V to right bit shift the luminance difference by the power factor k.

The adding of the target pixel value and the compensation value (S6) may include adding the compensation signal and the target signal to be compensated among the decoded video signals inputted from the inverse DCT.

Accordingly, the image processing apparatus according to the embodiment of the present invention reduces the filter provided in the encoder for compressing the original video signal, and further includes a block dispersing unit in the decoder for restoring the compressed video signal, thereby reducing the luminance difference between blocks. Gradually spread out from.

 It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the invention as defined in the appended claims. It can be understood that

Accordingly, the image processing apparatus according to the embodiment of the present invention reduces a plurality of low pass filters that are inevitably provided in the encoder to remove a block phenomenon occurring when the encoded image signal is decoded, and performs a simple operation on the decoder. A block dispersing unit may be provided to efficiently remove a blockage phenomenon occurring during encoding of a video signal processed in real time.

Claims (10)

An encoder comprising a DCT unit for receiving a video signal in block units and converting a video signal of a time sequence component into a DCT coefficient of a frequency sequence component, and a quantizer for converting the DCT coefficient into a coded bit string of a specific set; A decoder comprising an inverse quantizer for converting the bitstring encoded by the encoder into a DCT coefficient, and an inverse DCT unit for converting the DCT coefficient of the frequency sequence component into a decoded video signal of a time sequence component; A block spreader which receives the decoded video signal output from the decoder in units of blocks and distributes the luminance difference between the blocks of the decoded video signal to other pixels in the vicinity; The block dispersion unit An 8x8 progress unit which receives and stores the decoded video signal in units of blocks; A control unit which receives a video signal stored in the 8x8 progress unit, extracts a luminance difference between blocks in units of rows, and determines a compensation variable that is a luminance compensation degree of a target pixel to compensate for the luminance value; A multiplier for determining a compensation value by multiplying the luminance difference and the compensation variable; An adder for adding the compensation value and the luminance value of the target pixel; The compensation variable is a power of 1/2 Image processing apparatus, characterized in that. The method of claim 1, wherein the decoder An image processing apparatus, comprising: no filter for blocking removal. delete The method of claim 1, wherein the luminance difference between the blocks And a result obtained by subtracting the luminance value of the first left pixel from the luminance value of the first right pixel centering on the boundary between blocks. delete The multiplier of claim 1 wherein the multiplier And write bit shifting the luminance difference value by the number of times corresponding to a position value of the target pixel based on a block boundary. An 8x8 progression unit which receives and stores the decoded video signal in block units; A control unit which receives a video signal stored in the 8 × 8 processing unit, extracts a luminance difference in units of rows between blocks, and determines a compensation variable indicating a compensation degree of a target pixel to compensate for the luminance value; A multiplier for determining a compensation value by multiplying the luminance difference and the compensation variable; In the block processing phenomenon removal method of the image processing apparatus, characterized in that it comprises an adder for adding the compensation value and the luminance value of the target pixel. Storing luminance values of pixels in block units; Extracting a luminance difference between blocks; Determining a compensation value according to the luminance compensation degree of the target pixel; Adding the compensation value to the luminance value of the target pixel; Determining a compensation value according to the luminance compensation degree of the target pixel Extracting a column value of the target pixel to compensate for the luminance value; If the extracted column value of the target pixel is less than 0, right-bit shifting the luminance difference between the blocks by an absolute value of the column value of the target pixel;  If the column value n of the target pixel is greater than or equal to 0, performing a light bit shift on the luminance difference between the blocks by a value obtained by adding 1 to the column value of the target pixel. Blocking phenomenon removal method of the image processing apparatus, characterized in that. The method of claim 7, wherein storing the luminance value of the pixel in block units And storing the decoded video signal in block units (8x8). 8. The method of claim 7, wherein extracting the luminance difference between blocks And subtracting the luminance value of the left pixel from the luminance value of the right pixel among the pixels adjacent to the block boundary of each row. delete

Images