KR20010066560A - shaking compensation circuit using motion estimation and motion compensation in videophone - Google Patents

Abstract

PURPOSE: A shaking compensation device using moving prediction and moving compensation in video phone is provided to enable a user to compensate for imaging position and posture etc by processing moving prediction and moving compensation and inputting again an image whose shaking is compensated for to a terminal monitor in a call looking at the other with each other. CONSTITUTION: An encoder for video phone consists of an encoding controller, a transformer to remove space overlap of input image, a quantizer, a reverse quantizer, a reverse transformer and a prediction memory to compare moving of input image with that of recorded past image. A video input unit(1) receives an image of a size referring to a shaking compensation region and transmits it to a shaking compensator(30) and the prediction memory. The shaking compensator(30) compensates for a moving generated on account of a user's shaking. A video output unit(32) outputs the image whose shaking is compensated for by the shaking compensator(30) again to a terminal monitor.

Description

Shaking compensation circuit using motion estimation and motion compensation in videophone

In the method of transmitting data, voice, and video in both directions using wireless communication such as IMT-2000, which is the next generation mobile communication standard, in particular, video has a great advantage that users can face to face with each other. At this time, since the amount of data transmitted and received in both directions is large, the amount of information of voice and video is reduced by using a so-called compression method called source coding. In particular, in the case of video, the amount of information is larger than that of voice, and thus spatial compression using a DCT (Discrete Cosine Transform) and a quantizer is performed to remove redundancy between data such as H.261 or H.263 by compressing the information. We use a method that removes temporal redundancy simultaneously using redundancy, motion estimation and motion compensation. In this case, in order to use the motion prediction and the motion compensation, the motion information is obtained by using the past or future screen in addition to the current picture, the motion compensation is performed, and then the difference component with the input screen is obtained. To this end, the prediction memory is stored in the image processing unit of the terminal, and the screen information for comparing the motions is stored therein, and the motion vector transmitted from the receiver is obtained by measuring a motion vector of the current or incoming screen and measuring the amount of motion. By reconstructing the image using vectors, the image information is compressed.

1 is a block diagram showing a conventional video phone encoder.

As shown in FIG. 1, the image received from the image input unit 1 of the terminal is distributed to the prediction memory 14 and the subtractor 2, respectively. The distributed input images are first compared with the amount of motion by using the images of the past (or future) in the prediction memory 14. First, in order to compare the motions, the input image is decomposed into macroblocks. For example, the image processing unit first divides an image of luminance signal N × M pixel into an 8 × 8 pixel block, which is called a DCT block, and has a 16 × 16 size of four DCT blocks. Blocks are called macroblocks, and the macroblocks are grouped together into slices. The macroblock divided as described above calculates a position having the smallest difference component compared with the stored image, which is called motion estimation, and the amount of motion is called a motion vector 15. Each motion vector 15 thus calculated is transmitted to the other receiver and used to recover the motion. The process of calculating the motion vector 15 and reconstructing the input image from the past or future image information close to the input image is called motion compensation, and the motion compensated image is input to the subtractor 2 again. Subtracted from the input image. In addition, the difference image from which the temporal redundancy output from the subtractor 2 is removed is output through the quantizer 9 through the DCT converter 8 to remove spatial redundancy again, and then a reference for motion prediction and motion compensation. In order to construct an image, the image is input to the adder 13 through an inverse quantizer 11 and an inverse DCT converter 12. In the adder 13, the motion compensated image and the reconstruction difference image output from the inverse DCT converter are added to reconstruct the reference image to be used for the next motion prediction and the motion compensation, and store the resultant image in the prediction memory 14. Repeat the above process with the original image being input.

In the above-described video encoder for a videophone, all motions generated by user's shaking are recognized as motion of image information. Therefore, when the video image is processed up, down, left, and right, the motion of the original video is compensated as it is. Compared to the case of processing, the amount of motion increases, which causes a problem of increasing the amount of data of the motion vector to be transmitted to the other party. In addition, since the image reconstructed by the receiver also restores the image which is shaken up, down, left and right according to the transmitted motion vector value, there is a problem that the receiver cannot see the unstable image.

The present invention has been presented to solve the above problems, and an object of the present invention is to calculate the degree of shaking using motion prediction and motion compensation in a videophone, and to offset the amount of motion, thereby increasing the compression efficiency and restoring a stable screen upon reception. To provide a device that can.

As a technical idea for achieving the object of the present invention, the motion of the encoding controller, the transformer for removing the spatial redundancy of the input image, the quantizer, the inverse quantizer, the inverse transformer and the stored past image and the input image A video phone encoder comprising a predictive memory to be compared, comprising: a video input unit for receiving an image having a size considering a shake compensation region and transmitting the shake compensator to the predictive memory; And a video output unit for outputting a shake compensated image by the shake compensator back to a terminal monitor, wherein the predictive memory has a shake amount and a motion by comparing the image received from the video input unit with a past or future reference image. Calculate the amount and the A shake vector and a motion vector are transmitted to a receiver, the shake vector is compensated in units of frames, the motion vector is compensated in units of macroblocks (16 × 16), and the shake compensator receives the input image transmitted from the video input unit. The present invention is characterized by outputting the image size of the compressed transmission by compensating for the amount of shaking transmitted from the prediction memory.

1 is a block diagram showing a conventional video phone encoder.

2 is a schematic diagram showing the effect of shaking.

3 is a schematic diagram showing the amount of motion vector according to the shake compensation. FIG. 3 (a) shows before the shake vector is removed and FIG. 3 (b) shows after the shake vector is removed.

4 is a block diagram showing an encoder for a videophone according to the present invention.

<Description of the symbols for the main parts of the drawings>

1, 1 ': Video input 2: Subtractor

3: switch 4: coding controller

5: In-frame / inter-frame identification flag 6: Transmission / non-transmission identification flag

7: quantization characteristic designation information 8: converter

9: quantizer 10: level of transform coefficients

11: inverse quantizer 12: inverse transformer

13: Adder 14: Predictive Memory

15: motion vector 21, 21 ': screen shaken up, down, left and right

22, 22 ': Original video 23: Predictive memory video

24: shake compensator output image 30: shake compensator

31: shake vector 32: video output

Hereinafter, with reference to the accompanying drawings with respect to the configuration and operation of the embodiment of the present invention will be described.

2 is a schematic diagram showing the effect of shaking.

As shown in FIG. 2, the original screen 22 and the screens 21 and 21 ′ which are shaken to the upper (bottom) left (right) are shown. In the conventional video phone, all the motions generated by the shaking of the user are displayed. Recognize it as a movement.

3 is a schematic diagram showing the amount of motion vector according to the shake compensation, in which FIG. 3 (a) shows before the shake vector and FIG. 3 (b) shows after removing the shake vector.

As shown in FIG. 3 (a), when the original input image 22 ′ and the predictive memory image 23 are compared to determine the shake vector when the overall motion vector exists, the shake compensator output image ( 24). When compensating for the motion of the original image as it is, compared with the case of processing the shake compensation, the amount of motion becomes large, resulting in an increase in the amount of data of the motion vector to be transmitted to the counterpart. In addition, the image reconstructed by the receiver also restores the image which is shaken up, down, left and right according to the transmitted motion vector value, so that the receiver has no choice but to view the unstable image.

4 is a block diagram showing a video phone encoder according to the present invention.

As shown in FIG. 4, the encoding controller 4, a transformer 8 for removing spatial redundancy of an input image, a quantizer 9, an inverse quantizer 11, and an inverse transformer 12 are stored. In the video phone encoder comprising a prediction memory 14 for comparing the motion of the past image and the input image, the image of the size considering the shake compensation region is input to the shake compensator 30 and the prediction memory 14. A video input unit 1 'for transmitting, a shake compensator 30 for compensating for movement caused by the shaking of the user, and a video output unit for outputting the shake compensated image by the shake compensator 30 back to the terminal monitor. It consists of 32.

The size of the image inputted from the input unit 1 'of the video phone is not N × M, but an image having a size of N' × M '(N'> N, M '> M) considering a shake compensation region. And send it to the prediction memory 14 and the shake compensator 30. The distributed image is first compared to the image of the past (or future) size of N × M in the prediction memory 14 with the amount of shake and the amount of motion. First, in order to find out the amount of shake and the amount of motion, the value is obtained by comparing with the image stored in the prediction memory 14 as in the conventional invention. At this time, the amount of shake and the reference image as a reference for calculating the amount of motion are opposite to each other, and the amount of shake is obtained in picture units rather than macroblock units, unlike the amount of motion.

The shake compensator 30 receives the shake vector 31 output from the predictive memory 14 and receives an N '× M' size image input from the video input unit 1 'to the shake compensator 30 by this value. To compensate for the motion compensation image of the final size N × M size. The shake compensation image output from the shake compensator is output to the monitor unit of the subtractor 2 and the video phone, and the image signal output to the monitor unit indicates the position or posture of the user's image input from the image input unit such as a camera. Allow correction. Then, the video signal output to the monitor unit of the subtractor 2 and the video phone allows the user to correct the position or posture of his / her own image input from an image input unit such as a camera. In addition, the shake compensation image input to the subtractor 2 is processed by the subtractor 2 and the image compensated by the amount of motion in macroblock units in the prediction memory 14. The video of the difference component from which the motion information is removed through the subtractor is subjected to the motion-compensated image in the prediction memory 14 through the DCT converter 8 and the quantizer 9 and vice versa as in the conventional method described above. Combined to create a motion compensation reference image that will be used next.

As can be seen from the above description, in the present invention, the user holding the video terminal and illuminates his / her face during the call, so that the image is compensated for the shake through the process of motion prediction and motion compensation while talking to each other. By inputting it back to the terminal monitor, the user can easily correct his or her image position and posture, and reduce the amount of movement during motion compensation, thereby reducing the compressed image data transmitted.

Claims (3)

An encoder for a videophone comprising a coding controller, a transformer for removing spatial redundancy of an input image, a quantizer, an inverse quantizer, an inverse transformer, and a predictive memory for comparing a movement between a stored past image and an input image, A video input unit which receives an image having a size considering a shake compensation region and transmits it to a shake compensator and the prediction memory; A shake compensator to compensate for movement caused by a shake of the user, And a video output unit for inputting the shake compensated image by the shake compensator back to the terminal monitor. The prediction memory calculates a shake amount and a motion amount by comparing the image received from the video input unit with a reference image of the past or future, and transmits the shake vector and the motion vector to the shake compensator and the receiver. Frame-wise and the motion vector is macroblock-wise. The shake compensator is a motion compensation device using the motion prediction and compensation in the video phone, characterized in that for outputting the video size to compensate for the amount of the shake transmitted from the prediction memory to the image transmitted by the video input unit. The shake compensation apparatus of claim 1, wherein the shake compensated image is inputted again through a monitor input of a terminal so that a user can correct his or her image position and posture. The method of claim 1, wherein the prediction memory first compensates for the shaking of an image signal having a size considering the shaking compensation region transmitted from the video input unit by using an image signal having a size of a compression or transmission stored in the past or in the future. And a motion compensation device using motion prediction and compensation in a video phone, the compressed video data being reduced by reducing the amount of motion.