KR100229794B1 - Image decoder having function for restructuring error of motion vector - Google Patents

Abstract

The present invention relates to an image decoder having an error detection function for motion vector information received from an image decoder and an error recovery function for motion vector information for restoring the detected error to original motion vector information.

According to the present invention, non-error motion vector information, that is, candidate motion vector information is generated by inverting bit by bit from the most significant bit to the least significant bit of the motion vector information bit in which an error occurs, and among the generated candidate motion vector information. After selecting only the candidate motion vector information included in the block range, the motion prediction prediction error signal and the result of the motion prediction are performed on each of the candidate motion vector information selected by the previous frame image signal stored in the frame memory. The candidate block signal having the highest correlation is determined based on the previous frame image signal among the added candidate block signals, and the frame image signal predicted from the motion vector information corresponding to the determined candidate block signal and the motion vector information in which no error is generated are determined. Add motion compensated prediction error signal By restoring to over the original image signal, even if an error or the like caused by the transmission channel environment, there is an effect that it is possible to effectively restore the more the error occurs the motion vector.

Description

Image Decoder with Error Recovery Function for Motion Vector Information

The present invention relates to an image decoder having an error compensation function. More particularly, the present invention relates to an error vector for motion vector information received from an image decoder and to a motion vector information for restoring the detected error to original motion vector information. An image decoder having an error recovery function.

As is well known in the art, the transmission of discrete video signals can maintain better image quality than analog signals. When a video signal consisting of a series of image "frames" is represented in digital form, a significant amount of data must be transmitted, especially for high quality televisions (aka HDTVs). However, since the usable frequency range of the conventional transmission channel is limited, in order to transmit a large amount of digital data, it is necessary to compress the transmitted data and reduce the amount thereof. Among the various compression methods for compressing data, hybrid coding, which combines stochastic coding with temporal and spatial compression, is known to be the most efficient, and these methods have already been proposed by the World Organization for Standardization. It is widely disclosed in the recommendations of enacted MPEG-1 and MPEG-2.

Most hybrid coding techniques use motion-compensated DCPM (prediction pulse code modulation), two-dimensional DCT (discrete cosine transform), quantization of DCT coefficients, variable-length coding (VLC), and the like. The motion compensation DPCM determines a motion of an object between a current frame and a previous frame and predicts the current frame according to the motion of the object to generate a differential signal representing the difference between the current frame and the predicted value.

In general, two-dimensional DCT converts digital image data blocks, for example, 8x8 blocks, into DCT conversion coefficients by using or removing spatial redundancy between image data.

The DCT conversion coefficient may be processed through a quantizer, a zigzag scan, a VLC, or the like to effectively hide (or compress) the amount of data to be transmitted.

More specifically, the motion compensation DPCM predicts the current frame from the previous frame according to the motion of the object estimated between the current frame and the previous frame. The estimated motion may be represented by two-dimensional motion vector information indicating the displacement between the previous frame and the current frame.

Typically, there are several approaches to estimating the displacement (vector) of an object. These are generally classified into two types, one of which is a block-based motion estimation method using a block matching algorithm, and the other is a pixel-based motion estimation method using a pixel cyclic algorithm.

In the motion estimation method for estimating the displacement of an object as described above, the displacement is calculated for all of the pixels using the pixel-based motion estimation method. This method has the advantage of being able to estimate pixel values more accurately and easily handle scale changes (e.g., zooming, a movement perpendicular to the image plane), while motion vector information for each pixel Since it is determined, a large amount of motion vector information is generated and it is impossible to transmit substantially all the motion vector information to the receiver.

In addition, in block-by-block motion estimation, a block having a predetermined size of the current frame is moved by one pixel in a search range of a previous frame and compared with the corresponding blocks to determine an optimal matching block having a minimum error value. From this, the interframe displacement vector (the extent to which the block has moved between frames) for the entire block is estimated for the current frame to be transmitted. Here, for determining the similarity between two corresponding blocks between the current frame and the previous frame, the average absolute difference, the mean square difference, etc. are mainly used, as is well known in the art.

On the other hand, in the above-described encoding scheme, that is, a bitstream encoded through a coding scheme such as motion compensation DCPM, two-dimensional DCT, quantization of DCT coefficients, and VLC (or entropy encoding) is transmitted to an output side of an image encoding system When the next transmission time is stored in the buffer, it is sent to the video decoder. The image decoder separates the variable length coded data and the motion vector information from the bitstream received from the transmission buffer, and obtains a differential image signal obtained by performing variable length decoding, inverse quantization, and inverse DCT on the variable length coded data. The motion vector information is reconstructed as original motion vector information by adding a predicted video frame whose motion is predicted based on a previous frame stored in a frame memory and the difference image signal.

In this case, a channel error in reconstructing the original video signal in the video decoder is generated in a bitstream of motion vector information obtained by motion estimation and motion compensation prediction error compensated through motion vector information.

In this case, the image quality deterioration due to an error occurring in the motion vector adversely affects viewing. Therefore, in order to solve the problem, the average of the motion vectors of the neighboring partitions is used to restore the error vector to the original motion vector information. The replacement method is used as a zero motion vector. However, this method alone does not accurately compensate for errors.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to detect motion vector information in which an error has occurred and generate non-error candidate motion vector information based on the detected motion vector information. By selecting candidate motion vector information having an optimal correlation based on the candidate motion vector information of the error and the neighboring block information, the motion vector information for accurately reconstructing the motion vector information in which the error has occurred is returned to the original motion vector information. An image decoder having an error recovery function is provided.

According to the present invention for achieving the above object, when the video encoding side receives the motion vector information to which the parity bit is added for each data of a predetermined block unit and the data variable-length encoded into the bitstream, the corresponding motion vector by the parity bit. An image decoder having an error recovery function for motion vector information for detecting whether an error occurs and restoring motion vector information to original motion vector information, wherein the image decoder corresponds to a result of detecting whether an error occurred in the motion vector information. A first switch for switching and outputting motion vector information; A second switch for selectively outputting a motion compensation prediction error signal according to a result of detecting whether an error of the motion vector information occurs; After selecting only candidate motion vector information included in the screen block range among non-error candidate motion vector information generated from the error-producing motion vector information output through the first switch, the selection is performed based on a previous frame image signal. A candidate selector configured to provide a candidate block signal according to a result of performing a motion prediction on each of the candidate motion vector information and a motion compensation prediction error signal outputted through the second switch; An error compensator for determining a candidate block signal having the largest correlation among candidate block signals provided by the candidate selector based on the previous frame signal, and providing motion vector information corresponding to the determined candidate block signal; And a decoder configured to add the motion compensation prediction error signal and the predicted frame image signal according to decoding, inverse quantization, and inverse discrete cosine transform to the variable length coded data, and reconstruct the original image signal. The image signal may be predicted by the previous frame image signal stored in the frame memory, the motion vector information according to the switching operation of the first switch, and the motion vector information provided by the error compensator.

1 is a block diagram of an image decoder having an error reconstruction function for motion vector information according to the present invention.

(A) and (b) of FIG. 2 are diagrams showing an example of motion vector information for checking whether an error of motion vector information according to the present invention occurs.

3 is a diagram illustrating an example of candidate motion vector information with respect to motion vector information in which an error occurs.

<Description of the code | symbol about the principal part of drawing>

100: receiving buffer 200: decoder

300: error detection unit 420, 440: switch

500: candidate selection unit 600: error compensation unit

Hereinafter, the present invention will be described in detail with reference to the illustrated drawings.

1 is a block diagram of an image decoding apparatus having an error reconstruction function for motion vector information according to an embodiment of the present invention, and includes a receiving buffer 100, a decoding unit 200, an error detecting unit 300, and a first and second switches. 420 and 440, the candidate selector 500, and the error compensator 600.

The reception buffer 100 includes motion vector information to which parity bits are added to data of a predetermined block unit to implement a video, and transmits the variable length coded data transmitted in a bitstream form of the image encoding side, which is not shown. The variable length coded data received from the transmission buffer is provided to the ping 210 of the decoder 200, and the motion vector information is provided to the error detector 300.

The decoder 200 includes a VLD 210, an IQ and IDCT 220, a frame memory 230, a motion predictor 240, and an adder 250 to decode the video signal of the current frame from the previous frame video signal. The output will be restored.

The VLD 210 variable length decodes the variable length encoded data received in the reception buffer 100 and provides the variable length decoded data to the IQ and IDCT 220 to be described later. The IQ and IDCT 220 are variable length decoded in the VLD 210. A motion compensated prediction error signal obtained by inverse quantization and inverse discrete cosine transform is provided to the adder 250 and the second switch unit 440.

The adder 250 adds a motion compensation prediction error signal obtained by inverse quantization and inverse DCT of the frame image signal predicted by the motion predictor 240 and the variable length decoded data, as described below. The image frame signal to be restored is provided to the frame memory 230 and the display device.

The frame memory 230 stores a previous frame image signal provided by the adder 250, and the stored previous frame image signal includes a motion predictor 240, a candidate block generator 530, and an error compensator 600 which will be described later. Is provided.

The motion predictor 240 predicts the motion based on the previous frame video signal stored in the frame memory 230 and formal motion vector information, that is, motion vector information in which no error occurs, and predicts the predicted frame video signal. The adder 250 is provided.

The error detector 300 connected to the reception buffer 100 detects whether an error of the motion vector information is generated based on the parity bit of the motion vector information received from the reception buffer 100, and an error is detected in the motion vector information. While not generated, a switching control signal (high level logic signal) for motion prediction is provided to the first and second switches 420 and 440 based on the motion vector information, and an error occurs in the motion vector information. If so, the first and second switches 420 and 440 provide a switching control signal (a low level logic signal) for error correction of the error vector motion vector information.

In this case, the first and second switches 420 and 440 may output an output path of the motion vector information and the motion compensation prediction error signal according to a switching control signal according to whether an error occurrence of the motion vector information of the error detector 300 is detected. And to control.

The candidate selector 500 includes a candidate generator 510, a candidate selector 520, a candidate block generator 530, and adders 5401 to 540n.

The candidate generator 510 is generated by inverting only one bit from the most significant bit to the least significant bit of the motion vector information bit in which an error is generated based on the motion vector information including the error according to the switching operation of the first switch 420. The candidate motion vector information is provided to the candidate selector 520.

The candidate selector 520 selects only candidate motion vector information included in the screen block range among the candidate motion vector information provided from the candidate generator 510 and provides the candidate motion vector to the candidate block generator 530.

The candidate block generator 530 is a candidate block corresponding to the candidate motion vector information based on the selected candidate motion vector information provided from the candidate selector 520 and the previous frame image signal provided from the frame memory 230. After performing the motion prediction for each of them, the result signal is configured to be provided to the adders 5401 to 540n described later. The adders 5401 to 540n are configured to correspond to the blocks of the candidate block generator 530 so that the candidate blocks provided with the candidate block generator 530 and the second switch 440 are predicted. The candidate block signal reconstructed by adding the motion compensation prediction error signal inputted through the signal is provided to the error compensator 600.

The error compensator 600 determines the candidate block signal having the highest correlation among the reconstructed candidate block signals provided by the adders 5401 to 540n based on the previous frame image signal provided from the frame memory 230, The motion vector information corresponding to the candidate block signal determined therefrom is provided to the motion predictor 240.

A detailed operation of an image decoder having an error reconstruction function for motion vector information configured as described above will be described in detail.

First, assuming that normal motion vector information is received, the following will be described.

Referring to (a) and (b) shown in FIG. 2, it is first checked whether the bit [0 1 0 1 1 0 0] representing the motion vector information shown in (a) of FIG. 2 is an odd value or an even value. Will be done. If the result of the check is an odd value, the parity bit (X) is set to 0 [0 1 0 1 1 0 0 ] Bit is received from the encoding side, and if the bit [0 1 0 1 0 0 0] indicating the motion vector information shown in (b) of FIG. 2 is an odd value or an even value and is an even value, Parity bit (X) equals 1 [0 1 0 1 0 0 0 ] The bit is received from the encoding side.

For example, as can be seen from the above, the variable length coded data is received from the reception buffer 100 by receiving a bitstream including motion vector information to which parity bits are added and variable length coded data is stored in the VLD 210. After variable length decoding through the VLD 210, the motion compensation prediction error signal is calculated by the IQ and the IDCT 220, and is output to one side of the adder 250 and the second switch 440.

At this time, if the motion vector information received by the reception buffer 100 is provided to the error detector 300, the error detector 300 determines whether an error is referred to by referring to the parity bit added to the motion vector information bits. As assumed above, when the motion vector information received through the reception buffer 100 is determined to be normal motion vector information, the error detection unit 300 transmits a high level logic signal on the line 2 to the first and the second. The second switches 420 and 440 are provided.

Accordingly, the first switch 420 switches the Souch terminal to the contact point a according to a high level logic signal provided from the error detector 300 so that the detected motion vector information is provided to the motion predictor 240. do. In addition, the second switch 440 is switched off according to a high level logic signal provided from the error detection unit 300 so that the motion compensation prediction error signal provided from the IQ and IDCT 220 is selected as a candidate selector ( Blocking the output to the adders (5401 to 540n) of 500).

Accordingly, the motion predictor 240 predicts the motion from the previous frame video signal based on the previous frame video signal stored in the frame memory 230 and the motion vector information input through the first switch 420. The predicted frame image signal is provided to the adder 250.

The adder 250 outputs the frame image signal reconstructed by adding the frame image signal predicted by the motion predictor 240 and the motion compensation prediction error signal provided from the IQ and IDCT 220 to the display device. The frame memory 230 is provided.

In contrast to the above, a description will be given of a state that assumes that abnormal (error) motion vector information is received.

For example, the receiving buffer 100 receives the bitstream including the motion vector information to which the parity bits are added and the variable length encoded data, decodes the variable length, and then provides the decoded data to the IQ and IDCT 220. The motion compensation prediction error signal is calculated and output to one side of the adder 250 and the second switch 440, and the motion vector information is provided to the error detector 300. Here, the error detector 300 determines whether there is an error based on the parit bit added to the motion vector information bit.

Referring to (a) and (b) of FIG. 2, first, it is first checked whether a bit [0 1 0 1 1 0 0] representing the motion vector information shown in (a) is an odd value or an even value. If the result of the check is an odd value, the parity bit (X) is set to 0 [0 1 0 1 1 0 0 ] Bit is received from the encoding side, and if the bit [0 1 0 1 0 0 0] indicating the motion vector information shown in (b) of FIG. 2 is an odd value or an even value and is an even value, Parity bit (X) equals 1 [0 1 0 1 0 0 0 ] The bit is received from the encoding side.

However, when a bit representing the motion vector information to which the parity bit X is added means an even value, it is determined that an error has occurred.

As assumed above, if it is determined that the abnormal motion vector information, the error detector 300 provides a low level logic signal to the first and second switches 420 and 440 on the line 2, Accordingly, the first switch 420 switches the switch terminal to the contact b to provide the motion vector information detected by the error detector 300 to the candidate generator 510 of the candidate selector 500. The second switch 440 switches the switch terminal to on so that the motion compensation prediction error signal provided from the IQ and IDCT 220 is output to the adders 5401 to 540n of the candidate selector 500. Do this.

The candidate generator 510 generates a candidate motion vector generated by inverting only one bit from the most significant bit to the least significant bit of the motion vector information bit in which an error is generated by the motion vector information including the error provided from the first switch 420. Information is provided to the candidate selector 520.

Referring to (a) to (i) of FIG. 3 as an example, if the bit input from the first switch 420 to the candidate generator 510 is as shown in (a) of FIG. The candidate motion data bits as shown in (b) to (i) of FIG. 3 are generated.

First, if only the most significant bit [0] among the bits of (a) of FIG. 3 in which an error occurs is inverted to [1] as shown in (b) of FIG. do. Also, as shown in (c) of FIG. 3, only the first lower bit [1] of the most significant bit of the motion vector in which the most significant bit of the bits of FIG. When inverted, these bits become candidate motion data as non-error data, and only the second lower bit [0] of the most significant bit among the bits of FIG. 3 (a) where an error occurs is shown in FIG. When inverted to [1], the bits become candidate motion data as non-error data.

By the above operation, candidate motion data are generated as shown in FIGS. 3B to 3 and provided to the candidate selector 520. The candidate selector 520 selects only candidate motion vector information included in the screen block range among the candidate motion vector information provided from the candidate generator 510 and provides the candidate motion vector to the candidate block generator 530. Herein, the screen block range is limited to a predetermined size at which the frame video signal is displayed.

The candidate block generator 530 may select candidate blocks corresponding to the candidate motion vector information based on the selected candidate motion vector information provided from the candidate selector 520 and the previous frame image signal from the frame memory 230. After performing the motion prediction for each, the resulting signal is provided to the adders 5401 to 540n.

The adders 5401 to 540n add error candidates predicted by the candidate block generator 530 and the motion compensation prediction error signal inputted through the second switch 440 to compensate for the restored candidate block signal. Provided to the unit 600.

Therefore, the error compensator 600 has the highest correlation among the candidate block signals reconstructed based on the previous frame image signal provided from the frame memory 230, that is, the neighboring blocks (the upper block in which the error occurred and the left block). The candidate block signal is determined, and motion vector information corresponding to the determined candidate block signal is provided to the motion predictor 240.

In more detail, the error compensator 600 firstly includes a previous frame image signal provided from the frame memory 230 (corresponding to the reference numeral Top in FIG. 4, the same position of the error generating block) and the adder 5401. The sum of the squared error with the predicted frame image signal (for example, (b) of FIG. 3) provided from FIG. The square error of the previous frame image signal provided from the frame memory 230 (corresponding to reference numeral L in FIG. 4 and the same position of the error generating block) and the predicted image frame signal provided from the adder 5401. Find the sum of. If the sum average of the squared errors obtained here is obtained, this is obtained from the candidate block generator 5301 to obtain first information related to the correlation of the candidate data provided first.

In addition, the error compensator 600 may include a previous frame image signal provided from the frame memory 230 (corresponding to the reference numeral Top of FIG. 4, the same position of the error generating block) and a predicted value provided from the adder 5402. The sum of squared errors with a frame image signal (e.g., (b) of FIG. 3) is obtained. The square error of the previous frame image signal provided from the frame memory 230 (corresponding to reference numeral L in FIG. 4 and the same position of the error generating block) and the predicted image frame signal provided from the adder 5402. Find the sum of. If the sum average of the squared errors obtained here is obtained, this is obtained from the candidate block generator 5301 to obtain first information related to correlation of candidate data first provided.

By performing the operation up to the data shown in (i) of FIG. 3 in this way, it is possible to obtain information that can know the correlation. That is, the above operation process is formulated as follows.

More specifically,

Where C _j ^ERR is the j th DCT coefficient of the block in which the error occurs, C _j ^T is the j th DCT coefficient of the block immediately above the error, and C _j ^L is the block in which the error occurred and the left block in which the error occurred. E _TOTAL represents the sum of the top block (Top) and the left block (L) where the error occurred.

The error compensator 600 selects a motion vector having the largest correlation among pieces of information that can be correlated, and provides the motion vector to the motion predictor 240 through the line 4.

Therefore, the motion predictor 240 predicts the motion based on the previous frame image signal stored in the frame memory 230 and the motion vector information from the error compensator 600, and adds the predicted image frame signal. By providing the reference numeral 250, the motion vector in which the error occurs can be compensated.

As described above, the present invention detects an error in motion vector information based on a parity bit due to a transport channel environment, and generates a candidate motion vector for the detected motion vector information. By selecting candidate motion vector information that is closest to the original motion vector information, the original motion vector information can be accurately restored.

Claims (2)

When the video encoding side receives the motion vector information to which the parity bit is added and the variable length coded data into the bit stream, the parity bit detects whether or not an error occurs in the motion vector information. A video decoder having an error recovery function for motion vector information for restoring vector information to original motion vector information, comprising: a first switch for switching and outputting motion vector information corresponding to an error detection result of the motion vector information; 420; A second switch 440 for selectively outputting a motion compensation prediction error signal according to a result of detecting whether an error of the motion vector information occurs; After selecting only the candidate motion vector information included in the screen block range among the non-error candidate motion vector information generated from the motion vector information of the error occurrence outputted through the first switch 420, based on the previous frame image signal Selecting a candidate block signal based on a result of performing a motion prediction on each of the selected candidate motion vector information and a motion compensation prediction error signal output through the second switch 440; 500); An error compensator for determining a candidate block signal having the largest correlation among candidate block signals provided by the candidate selector 500 based on the previous frame signal, and providing motion vector information corresponding to the determined candidate block signal ( 600); And a decoder 200 that adds the motion compensation prediction error signal and the predicted frame image signal according to decoding, inverse quantization, and inverse discrete cosine transform to the variable length coded data, and reconstructs the original image signal. The predicted frame image signal is based on the previous frame image signal stored in the frame memory, the motion vector information according to the switching operation of the first switch 420, and the motion vector information provided from the error compensator 600. An image decoder having an error reconstruction function for motion vector information, characterized in that it is predicted. The bit selector of claim 1, wherein the candidate selector 500 sets the least significant bit from the most significant bit of the motion vector information bit in which the error is generated by the motion vector information including the error provided by the first switch 420. A candidate generator 510 for providing the non-error candidate motion vector information generated by inverting only the candidate motion; and a candidate motion vector included in a screen block range among candidate motion vector information provided from the candidate generator 510. A candidate selector 520 that selects and provides only information, the candidate motion vector information provided by the candidate selector 520, and the candidate based on a previous frame signal provided by the decoder 200. A candidate block generator 530 which performs a motion prediction on candidate blocks corresponding to the motion vector information and then provides a result signal; The candidate block signal predicted by the lock generator 530 and the motion compensation prediction error signal provided through the second switch 440 are added to the error compensator 600. And an adder (541 to 542n) provided.