KR100248265B1 - Digital image coder and decoder for minimizing of eye sore - Google Patents

Abstract

The present invention relates to a digital image decoding / coding apparatus for reducing annoyance phenomenon of a digital image compression extension algorithm, and detects a unit block in which an encoding apparatus is expected to appear and transmits the detected information to the decoding apparatus. In this case, the decoding apparatus low-pass filters only the reconstructed unit blocks estimated to appear in accordance with the transmitted information to reduce the blurring phenomenon, thereby forming the subjective image quality of the image.

Description

Digital Video Coding / Encoding Device to Reduce Eyestrain

1 is a block diagram of a conventional digital video coder.

2 is a block diagram of a conventional digital video decoding apparatus.

3 is a block diagram of a digital video encoding apparatus for reducing the eye-eye phenomenon of the present invention.

4 is a configuration diagram of a digital image decoding device for reducing the eye-eye phenomenon of the present invention.

* Explanation of symbols for main parts of the drawings

1, 15: memory 2: motion estimator

3, 16: motion compensator 4: discrete cosine converter

5: quantizer 6, 13: inverse quantizer

7, 14 inverse discrete cosine converter 8: variable length coder

9, 11: buffer 10: image analyzer

12 variable length decoder 17 delay memory

18: Low pass filter SW1, SW2: switch

The present invention finds a part in which a partitioning phenomenon may appear in a coding apparatus to remove eye anomalies, especially partitioning, in a digital image compression and expansion algorithm, and uses a low pass filter in a decoding apparatus according to the sent information to reduce partitioning. The present invention relates to a digital image decoding / coding apparatus for reducing annoyance.

The conventional digital video encoding apparatus includes a memory 1 for storing a previous frame image, as shown in FIG. A motion estimator (2) for detecting a degree of motion by comparing an input image with a previous frame stored in the memory (1); A motion compensator (3) for extracting an image from a previous frame stored in the memory (1) according to the motion information detected by the motion estimator (2); A discrete cosine converter with an 8 * 8 transform coefficient based on the image extracted by the motion compensator 3; A quantizer 5 for calculating and quantizing the quantization interval of the coefficients transformed by the discrete cosine transformer 4; An inverse quantizer 6 for restoring the data quantized by the quantizer 5 into discrete cosine transform coefficients; An inverse discrete cosine converter 7 that receives the 12-bit coefficients output from the inverse quantizer 6 and restores the 9-bit pixel data; A variable length encoder (8) for variable length coding the discrete cosine transform coefficients and other information quantized by the quantizer (5); It consists of a buffer 9 for preventing underflow and overflow caused by the output coded by the variable length encoder 8 being transmitted at a constant bit rate.

A conventional digital video decoding apparatus includes a buffer 11 for adjusting the number of bits transmitted at a constant bit rate in an encoding apparatus, as shown in FIG. A variable length decoder 12 for decoding the variable length code of the number of bits adjusted by the buffer 11; An inverse quantizer 13 for restoring the variable length decoded data by the variable length decoder 12 into a discrete cosine transform coefficient; An inverse discrete cosine converter 14 which receives the 12-bit coefficients output from the inverse quantizer 13 and restores the 9-bit pixel data; A memory 15 for storing the pixel data recovered by the inverse discrete cosine converter 14 and the information decoded by the variable length decoder 12; A motion compensator (16) for extracting an image from a previous frame stored in the memory (15) according to the pixel data reconstructed by the inverse discrete cosine converter (14); And a switch SW1 for selectively outputting the output of the motion compensator 16 or the output of the inverse discrete cosine transducer 14.

The conventional digital video decoding / coding apparatus configured as described above is generally known as H.261 and Motion Picture Expert Group (MPEG) or a plurality of commonly used digital video information compression / storage or transmission. In digital high definition television (HDTV) algorithms, the number of bits generated to adjust the output of the encoder to a constant bit rate is controlled. As a result, when the image is complicated or there is a lot of motion, encoding is performed at a low bit rate, and thus a frustration phenomenon, in particular, a segmentation phenomenon appears.

However, the video expert group (MPEG) or the high-definition television (HDTV) introduced in the digital compression algorithm to reduce the angling phenomenon of the digital compression algorithm, proposes to reduce the bleeding phenomenon by adjusting the quantization interval on the encoder side However, the processing method on the decoding device side has not been proposed, and there is a limit to remove the eye strain in the case of a fairly complicated image.

The present invention is characterized in that it is possible to remove the eye bleeding phenomenon regardless of the input image on the decoding device side in view of such a conventional problem. In other words, as a way to reduce the annoyance phenomenon of the digital compression algorithm, if the encoder detects a unit block that is expected to appear the annoyance phenomenon and sends the detected information to the decoder, the decoder sends the annoyance sent from the encoder. The low-pass filtering was performed on the reconstructed unit blocks according to the estimated information to reduce the phenomenon of annoyance.

When described in detail with reference to the drawings as follows.

The digital video encoding apparatus for reducing the eye strain of the present invention includes a memory (1) for storing a previous frame image, as shown in FIG. A motion estimator (2) for detecting a degree of motion by comparing an input image with a previous frame stored in the memory (1); A motion compensator (3) for extracting an image from a previous frame stored in the memory (1) according to the motion information detected by the motion estimator (2); A discrete cosine transformer (4) for discrete cosine transforming an input image into an 8 * 8 transform coefficient by the image extracted by the motion compensator (3); A quantizer 5 for calculating and quantizing the quantization interval of the coefficients transformed by the discrete cosine transformer 4; An inverse quantizer 6 for restoring the data quantized by the quantizer 5 into discrete cosine transform coefficients; An inverse discrete cosine converter 7 which receives the 12-bit coefficients output from the inverse quantizer 6 and restores the 9-bit pixel data; A variable length encoder (8) for variable length coding the discrete cosine transform coefficients and other information quantized by the quantizer (5); A buffer 9 for preventing underflow and overflow caused by the output coded by the variable length encoder 8 being transmitted at a constant bit rate; The image analyzer 10 determines a part in which a partitioning phenomenon may occur using the complexity of the input image and the quantization interval.

As shown in FIG. 4, a digital video decoding apparatus for reducing the eye-eye phenomenon of the present invention includes: a buffer 11 for adjusting the number of bits transmitted at a constant bit rate from an encoder; A variable length decoder 12 for decoding the variable length code of the number of bits adjusted by the buffer 11; An inverse quantizer 13 for restoring the variable length decoded data by the variable length decoder 12 into a discrete cosine transform coefficient; An inverse discrete cosine converter 14 that receives the 12-bit coefficients output from the inverse quantizer 13 and restores the 9-bit pixel data; A memory 15 for storing the pixel data reconstructed by the inverse discrete cosine transformer 14 and the information decoded by the variable length decoder 12; A motion compensator (16) for extracting an image from a previous frame stored in the memory (15) according to the pixel data reconstructed by the inverse discrete cosine converter (14); A switch SW1 for selectively outputting the output of the motion compensator 16 or the output of the inverse discrete cosine transducer 14; A delay memory 17 for delaying the video information selected and output by the switch SW1 for a predetermined time; A switch SW2 for selectively outputting the video information delayed through the delay memory 17 according to the data information decoded by the variable length decoder 12; By selecting this switch SW2, only the low-pass filter 18 which reduces the blurring phenomenon by filtering low-pass only the unit block extracted from the coding apparatus is comprised.

Referring to the operation and effects of the present invention configured as described above in detail.

In the decoded image when the image is encoded at a low bit rate, a kind of annoying phenomenon such as a blocking effect that is well seen in a flat place, a color bleeding that appears at the boundary of the image having a flat background, In order to reduce this noticeable partitioning phenomenon, we can find out the segmentation phenomenon and apply a lowpass filter in the decoder to improve the subjective quality of the image.

Accordingly, the present invention can be broadly classified into two types. First, unit partition extraction to be applied to the filter, and second, low pass filter is applied to the unit partition extracted from the encoding device in the decoding device. First, looking at the method of extracting the unit partition to apply the filter in the encoder, it is necessary to find the necessary factor to find the part where the partitioning phenomenon can appear in the image analyzer 10, wherein the necessary factor is the complexity of the image, the quantization interval There are the followings, and the extraction unit of each factor uses a large segment, which is a unit for obtaining a motion vector, as an extraction unit. The complexity of the image, which is one of the factors to be extracted, is generally used as the complexity factor by calculating the injection value (Vbi; i-th large-segment variance value) in units of large compartments by using an example using variance as it is. The dispersion value Vbi which is an example of this calculation is as follows.

Vbi = [∑Sn-(∑Sn) ² ] / Npm

Sn represents a pixel value and Npm represents the number of pixels of a large compartment.

In addition, the quantization interval, which is another one of the factors to be extracted, is used as a factor for finding a part where the partitioning phenomenon may appear as it is using the quantization interval used in the quantizer 5 according to the existing algorithm.

With the complexity and quantization interval of the extracted large-segment unit, it is necessary to determine whether the filter is applied in the decoding apparatus. The large-segment determination algorithm uses the following determination equation.

Q_step is a quantization interval, Th1 is a threshold for complexity, and Th2 is a quantization interval threshold. If F_flag is "1", the decoder applies a filter. If "F" is 0, the filter is not applied. The F_flag information is encoded by the variable length encoder 8 and transmitted to the decoder. The image information transmitted to the decoding apparatus passes through the low pass filter 18 when the F_flag is "1" according to the F_flag information detected by the variable length decoder 12 in the large-segment unit through the delay memory 17, and " 0 "is output through the switch SW2 as it is.

When the switch SW2 is connected to the low pass filter 18, the delay memory 17 also outputs peripheral pixel values according to the number of filter taps in addition to the large block to be processed.

As described in detail above, the present invention provides a partitioning phenomenon by using the complexity of the image and the quantization interval in the encoding apparatus in order to reduce the partitioning phenomenon which is noticeable in the decoded image when the image is encoded at a low bit rate. It is possible to reduce the segmentation phenomenon and improve the subjective quality of the image by finding a part that can appear and using a method of using a low pass filter in a large block unit according to the information sent from the encoder in the decoder. have.

Claims (2)

A memory 1 storing a previous frame image; A motion estimator (2) for detecting a degree of motion by comparing an input image with a previous frame stored in the memory (1); A motion compensator (3) for extracting an image from a previous frame stored in the memory (1) according to the motion information detected by the motion estimator (2); A discrete cosine transformer 4 for discrete cosine transforming an input image into an 8 * 8 transform coefficient by the image extracted by the motion compensator 3; A quantizer 5 for calculating and quantizing the quantization interval of the coefficients transformed by the discrete cosine transformer 4; An inverse quantizer 6 for restoring the data quantized by the quantizer 5 into discrete cosine transform coefficients; An inverse discrete cosine converter 7 which receives the 12-bit coefficients output from the inverse quantizer 6 and restores the 9-bit pixel data; A variable length encoder (8) for variable length coding the discrete cosine transform coefficients and other information quantized by the quantizer (5); A buffer for preventing underflow and overflow caused by the output coded by the variable length encoder 8 being transmitted at a constant bit rate; And an image analyzer (10) for detecting a unit partition in which a partitioning phenomenon is expected to occur using the complexity of the input image and the quantization interval. A buffer 11 for adjusting the number of bits transmitted at a constant bit rate in the encoder; A variable length decoder 12 for decoding the variable length code of the number of bits adjusted by the buffer 11; An inverse quantizer 13 for restoring the variable length decoded data by the variable length decoder 12 into a discrete cosine transform coefficient; An inverse discrete cosine converter 14 that receives the 12-bit coefficients output from the inverse quantizer 13 and restores the 9-bit pixel data; A memory 15 for storing the pixel data reconstructed by the inverse discrete cosine transformer 14 and the information decoded by the variable length decoder 12; A motion compensator (16) for extracting an image from a previous frame stored in the memory (15) according to the pixel data reconstructed by the inverse discrete cosine converter (14); A switch SW1 for selectively outputting the output of the motion compensator 16 or the output of the inverse discrete cosine converter 14; A delay memory 17 for delaying the video information selected and output by the switch SW1 for a predetermined time; A switch SW2 for selectively outputting the video information delayed through the delay memory 17 according to the data information decoded by the variable length decoder 12; The digital image decoding device for reducing the eye strain characterized in that it comprises a low pass filter 18 to reduce the eye strain phenomenon by low-pass filtering only the unit block extracted from the encoder by the selection of the switch (SW2) .