KR0147938B1 - Bit allocation circuit for keeping uniform peak signal noise ratio - Google Patents

Abstract

The present invention relates to a video encoding apparatus for encoding a digital video signal to be transmitted in a group unit (GOP), so that a bit allocation amount of a frame image to be currently encoded in consideration of the peak signal-to-noise ratio (PSNR) of a previously encoded frame image In order to provide a bit allocation circuit for computing a bit allocation circuit according to the present invention, an image restoring unit for restoring the encoded image; A memory for storing the applied digital video signal in units of frames; A peak signal-to-noise ratio measurement unit for measuring the peak signal-to-noise ratio (PSNR) of the currently encoded frame image by using the frame image signal output from the image restoration unit and the frame image signal read out from the memory; The current transmission obtained by calculating the first bit allocation (T _i , T _p , T _b ) calculated using the peak signal-to-noise ratio output from the peak-to-noise ratio measurement unit, the quantization parameter output from the encoder, and the value of the output bit. And a bit allocation calculator for providing a second bit allocation (T _i ^′ , T _p ^′ , T _b ^′ ) of the frame image to be encoded. Therefore, it is possible to uniformly provide the image quality of a continuously reproduced image.

Description

Bit Allocation Circuit for Maintaining Uniform Peak Signal-to-Noise Ratio of Image

1 is a block diagram of an image encoding apparatus having a bit allocation circuit according to the present invention.

FIG. 2 is a detailed circuit diagram of the bit allocation calculator shown in FIG.

3 illustrates an example of a frame-by-frame transmission sequence of an encoded image.

* Explanation of symbols for main parts of the drawings

10: encoder 20: quantization parameter counter

30: Beat Counter 40: Video Restoration Unit

50: PSNR measurement unit 60: frame memory

70 bit allocation calculator 51 intra frame I detector

52: predictive (P) frame average detector 53: bidirectional (B) frame average detector

54 to 56: first to third selectors 57: calculation unit

The present invention relates to a bit allocation circuit provided in an encoding apparatus for compressing an image signal. In particular, the present invention relates to a bit allocation circuit so that a peak signal to noise ratio (hereinafter, referred to as PSNR) of an encoded image is uniformly maintained. The bit allocation circuit to allocate.

Conventional video encoding apparatuses perform compression processing suitable for group-by-group transmission by making 12 or 15 frames into one frame group (called GOP (Group Of Picture)) according to the format of the input source. Within the group transmitted in this compression process, there are intra (I or I) frames (also called pictures), predicted (P or P) frames, interpolated (B or Bidirectionally) frames, as is well known.

An intra frame (hereinafter, referred to as an I frame) is a frame for performing discrete cosine transform, encoding, and transmitting all data without motion compensation, and is mostly transmitted as the first frame of each group. This is to prevent the prediction error due to the motion compensation of the previous group from accumulating in the group. A prediction frame (hereinafter referred to as P frame) is a frame obtained by performing discrete cosine transform and encoding of an error predicted by a motion vector obtained from a previous I frame or P frame, and an interpolation frame (hereinafter referred to as B frame) It is a frame obtained by performing discrete cosine transform and encoding of an error predicted by a motion vector obtained bidirectionally from I or P frames. The transmission bit amount of each of these frames has been calculated and allocated by Equation 1 so far.

In Equation 1, T _i , T _p , and T _b are target bit amounts of I frames, P frames, and B frames, respectively, where R is the remaining Group Of Picture (GOP) bit allocation, and N _p is the remaining bit amount. The number of P frames, N _b is the number of remaining B frames, K _p and K _b are general constants, K _p = 1.0 and K _b = 1.4. X _i , X _p , and X _b are the values of 160 × (bitrate / 115), 60 × (bitrate / 115), 42 × (bitrate / 115), respectively. A value obtained by multiplying the total bit amount S generated by coding a frame by the average value AQ of the quantization coefficients S × AQ divided by the number of macroblocks MBNO in the frame (S × AQ) / MBNO ) Is used. If the target bit amount obtained for each frame obtained using these variables is smaller than the result value calculated by the right expression (bitrate / (8x frame rate)) of Equation 1, the result value calculated by the right expression is applied. Provided as the target bit amount of the frame.

As described above, the bit allocation of the frame to be encoded currently is calculated by calculating the bit allocation remaining in the group by using the value of counting the quantization parameter of the previous frame and the data (or output bit) encoded and output of the previous frame. Provided. That is, in allocating so that the total amount of bits allocated to the GOP is divided into the frames forming the GOP, for example, when the frames are transmitted in the order of I frames, P frames, and B frames, the amount of bits generated after the I frames are encoded. Using the value subtracted from the total bit amount R allocated to the GOP to provide the bit allocation amount of the next P frame, and according to the allocated bit amount, the P frame is encoded and the generated bit amount is transferred to the previous frame. The bit allocation of the next B frame is provided by using a value subtracted from the bit allocation of the remaining GOP after bit allocation. However, this bit allocation method does not consider the uniformity of the image quality of successive frame images in the group at all, which causes a problem that the peak signal-to-noise ratio of the image quality continuously reproduced is not maintained uniformly.

Accordingly, the present invention has been made to solve the above-described problem, by calculating the amount of bits allocated to the frame image to be currently encoded using a value measured from the peak signal-to-noise ratio (PSNR) of the previously encoded frame image, An object of the present invention is to provide a bit allocation circuit for maintaining a peak signal-to-noise ratio of a continuously reproduced image quality uniformly.

Bit allocation circuit according to the present invention in order to achieve the above object,

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of an image encoding apparatus having a bit allocation circuit according to an embodiment of the present invention, in which an encoder 10 for encoding an applied digital image signal is compressed and a quantization parameter output from the encoder 10 is counted. The quantization parameter counter 20 for reproducing, the bit counter 30 for counting bits output from the encoder 10, and the image reconstructor 40 for reconstructing the compressed video signal output from the encoder 10 to an original state. PSNR measurement unit 50 for measuring the peak signal to noise ratio (hereinafter, referred to as PSNR) of each frame image by the signal output from the image reconstructor 40, and digital image signal in units of frames. Frame counts 60 to be stored and provided to the PSNR measurement unit 50 by the count values output from the frame memory 60, the quantization parameter counter 20, and the bit counter 30. Calculates a target bit amount of the frame based on the calculated value of the target bit amount and the PSNR value of the frame provided from the PSNR measuring unit 50 to the bit allocation calculator 70 for providing the encoder 10 to the encoder 10. Is done.

FIG. 2 is a circuit diagram provided in the bit allocation calculator 70 shown in FIG. 1 according to the present invention. The average of PSNRs (hereinafter referred to as PSNR _i ) of I frames transmitted from the PSNR measurement unit 50 is shown in FIG. The first selector 54 and the PSNR measuring unit 50 for selectively outputting the PSNR _i output from the I frame average detector 51 and the PSNR measuring unit 50 and the values output from the I frame average detector 51. P-frame average detector 52, which calculates an average of PSNRs (hereinafter referred to as PSNR _p ) of P frames transmitted from the P-frame average detector 52 and PSNR _p transmitted from the PSNR measurement unit 50, and data output from the P-frame average detector 52 are selectively selected. From the second frame selector 55 and PSNR measurement unit 50 to obtain the average of the PSNR (hereinafter referred to as PSNR _b ) of the B frame transmitted from the PSNR measurement unit 50 Selects the signal output from the transmitted PSNR _b and B frame average detector 53 The target bit of each frame obtained by Equation 1 using the count result value transmitted from the third selector 56 and the quantization parameter counter 20 and the count result value transmitted from the bit counter 30. By calculating the signals output from the quotas (T _i , T _p , T _b ) and the first to third selectors 54 to 56, the bit allocations (T _i *, T _p *, T _b *) for each frame are calculated. And a calculation unit 57 for outputting to 10).

FIG. 3 is an example of a transmission sequence of frame images encoded and transmitted in a GOP unit, and two B frames are transmitted before I frames. Here, the B frame is a frame considering the motion vector of the I frame of the group.

Next, the operation of the present embodiment will be described in detail with reference to the accompanying drawings. The quantization parameter counter 20, the bit counter 30, the image restoring unit 40, the frame memory 60, the encoder 10, and the like, which are operated in the related art, will be briefly described.

First, when encoding is performed on the digital video signal applied by the encoder 10, the quantization parameter counter 20 counts the quantization parameter generated during encoding, and the bit counter 30 is output from the encoder 10. Count the data bits. The results counted by the quantization parameter counter 20 and the bit counter 30 are transmitted to the bit allocation calculator 70, respectively.

When the bit allocation calculator 70 transmits the result counted from the quantization parameter count 20 and the bit counter 30, the target bit amount T _i , T _p , Calculate T _b ).

On the other hand, the compressed video signal output from the encoder 10 is transmitted to the transmission path for transmission to the receiver not shown, and at the same time it is transmitted to the image restoration unit 40 is restored. The reconstructed frame video signal is transmitted to the PSNR measurement unit 50.

The PSNR measurement unit 50 calculates the bit allocation by using the frame image signal transmitted from the image restoration unit 40 and the frame image signal read out from the frame memory 60 to obtain the PSNR used as a measure of the image quality of the frame image. Transfer to section 70. At this time, when the corresponding frame is an I frame, the PSNR transmitted to the bit allocation calculator 70 becomes PSNR _i , and the PSNR transmitted when the P frame is PSNR _p , and the PSNR transmitted when the B frame is B frame. Becomes PSNR _b . The PSNR of which frame is transmitted in this way is determined by the frame order distributed in the corresponding GOP layer. For example, if the frame order distributed in the GOP layer is as shown in FIG. 3, PSNR _b , PSNR _b , PSNR _i , PSNR _b , PSNR _b , PSNR _p , PSNR _b , ... PSNR is output in order.

The bit allocation calculator 70 calculates the bit allocation amount T _i , T _p , T _b and the PSNR measurement unit 50 for each frame calculated using the signals transmitted from the quantization parameter counter 20 and the bit counter 30. The bit allocations (T _i ′, T _p ′, T _b ′) of the current frame to be encoded are calculated by using the PSNR transmitted from the C 1.

That is, when PSNR _b is applied from the PSNR measurement unit 50 by the sequence shown in FIG. 3, the bit allocation calculator 70 converts the applied PSNR _b to the third selector 56 and the B frame average detector 53. To send simultaneously.

The B frame average detector 53 transmits the applied PSNR _b to the third accumulator 531 to accumulate, and the third counter 532 counts the number of times the PSNR _b is input. According to the example shown in FIG. 3, since two B frames are applied successively, the count value of the third counter 532 becomes two, and the value accumulated by the third accumulator 531 is divided into two B frames. This is the result of accumulating the PSNR generated for. The accumulated result value and count value are transmitted to the third average value detector 533 to detect the average value of the currently transmitted B frame.

The average value detection method performed by the third average value detector 533 is the same as the general average value detection method. That is, the average value is detected by dividing the accumulated result value output from the third accumulator 531 by the counted value output from the third counter 532. At this time, the average value detection is performed on the frame image output from the encoder 10 until the frame image of the other component is output, as can be seen in the accumulated frame described above.

The third selector 56 selectively outputs the PSNR _b transmitted from the PSNR measurement unit 50 and the average value output from the third average value detection unit 533. The signal for controlling the selection of the third selector 56 is a signal generated by detecting a case where a continuous normal video signal is applied and a case where it suddenly changes. That is, when a normal video signal is transmitted, a control signal is generated so that the average value output from the third average value detector 533 is selected and transmitted through the third selector 56, and when the screen changes abruptly, Since it is difficult to cope with the generated average PSNR, a control signal is generated so that the PSNR _b applied to use the PSNR of the immediately processed image is selected and transmitted. At this time, the third accumulator 531 and the third counter 532 are reset for the newly applied PSNR _b . The PSNR _b transmitted through the third selector 56 is transmitted to the calculator 57.

When the PSNR _b is applied from the third selector 56, the calculation unit 57 calculates the following equation (2).

In Equation 2, T _b is previously obtained as mentioned in the bit allocation calculator 70 described above, and PSNR _i and PSNR _p are obtained from the first selector 54 and the second selector 55 which will be described later. The value sent. T _b ^′ obtained by Equation 2 is provided as a bit allocation of a frame to be encoded in the current encoder 10.

When the I frame is encoded and output from the encoder 10, the PSNR _i is output from the PSNR measuring unit 50. Accordingly, the bit allocation calculator 70 transmits the applied PSNR _i to the I frame average detector 51 and the first selector 54. The operation of the I frame average detector 51 and the first selector 54 transmits the corresponding PSNR _i to the calculation unit 57 in the same manner as the above-described third selector 56 and the B frame average detector 53. do.

When PSNR _i is applied from the first selector 54, the calculation unit 57 calculates T _i ^′ as a bit allocation amount of a frame to be encoded by the encoder 10 by calculating as in Equation 3 below.

In Equation 3, only PSNR _i is a currently applied value, and the rest are values previously obtained by the calculation unit 57.

On the other hand, when the video encoded and output by the encoder 10 is a P frame, the PSNR _p is output from the PSNR measurement unit 50. The bit allocation calculator 70 transmits the applied PSNR _p to the second selector 55 and the P frame average detector 52, respectively. The second selector 55 and the P frame average detector 52 are also driven in the same manner as the above-described selectors 54 and 56 and the frame average detectors 51 and 53 to output corresponding PSNR _p . The output PSNR _p is transmitted to the calculator 57.

When the PSNR _p is applied, the calculation unit 57 calculates _{P p} ^' as a bit allocation amount of the frame to be encoded by the current encoder 10 by calculating as in Equation 4.

In Equation 4, as in Equation 3, PSNR _p is a value currently output from the second selector 55, and the rest uses a value previously held. For example, when the frames are transmitted in the order as shown in FIG. 3, when calculating the bit allocation using the PSNR _p applied by the calculation unit 57, the PSNR _i is applied to the I frame existing three frames earlier. using the PSNR for _i which was generated during the encoding and, _b PSNR is used in the PSNR _b that was generated during the encoding for the two B frames transmitted in the immediately preceding P frame.

In this way, the PSNRs are not simultaneously output from the first to third selectors 54 to 56, but are output in the order of the transmitted frames. That is, according to the example shown in FIG. 3, the PSNR is first outputted from the third selector 56 to the calculation unit 57, and then the PSNR is outputted from the first selector 54 to calculate the calculation unit ( 57, and then the PSNR is output from the third selector 56 to the calculation unit 57, and then the PSNR is output from the second selector 55 to the calculation unit 57. Is sent. The calculation unit 57 calculates a new bit allocation for the PSNR applied in this way and outputs the new bit allocation to the encoder 10.

As described above, in the bit allocation circuit according to the present invention, the peak signal-to-noise ratio (PSNR) of a previously encoded frame image and the values (T _i , T _p , and T _b ) used in the previous bit allocation are different. By assigning the target bit of the frame to be currently encoded by calculating the PSNR value of), there is an effect that can provide an equal picture quality during continuous video signal reproduction.

Claims (4)

A bit allocation circuit of an image encoding apparatus having an encoder for compressing an applied digital image signal, comprising: an image restoring unit for restoring an image output from the encoder to an original state; A memory for storing the applied digital video signal in units of frames; A peak signal-to-noise ratio measurement unit for measuring a peak signal-to-noise ratio (PSNR) of the currently encoded frame image by using the frame image signal output from the image restorer and the frame image signal read out from the memory; The first bit allocation (T _i , T _p , T _b ) calculated using the peak signal-to-noise ratio output from the peak signal-to-noise ratio measurement unit, the quantization parameter output from the encoder, and the value of the output bit is calculated. And a bit allocation calculator for providing a second bit allocation (T _i ^′ , T _p ^′ , T _b ^′ ) of the frame image to be transmitted to the encoder. Bit allocation circuit for holding. 2. The apparatus of claim 1, wherein the bit allocation calculator detects an average value of the peak signal-to-noise ratio output from the peak signal-to-noise ratio measurement unit, and outputs the peak signal-to-noise ratio output from the peak signal-to-noise ratio measurement unit and the average detector. And a selector for selectively outputting according to a sudden change of an image signal for transmitting a signal, and a calculator for calculating the second bit allocation by calculating the signal output from the selector and the first bit allocation. Bit allocation circuitry to maintain a uniform peak signal-to-noise ratio (PSNR) of the image. 3. The apparatus of claim 2, wherein the average detector comprises: an accumulator for accumulating the peak signal-to-noise ratio output from the peak signal-to-noise ratio measurement unit, a counter for counting the number of occurrences of the peak signal-to-noise ratio transmitted from the peak signal-to-noise ratio measurement unit, and And a mean value detector for dividing a value accumulated in the accumulator by a count value and detecting an average value of a corresponding frame and outputting the average value of the corresponding frame to the calculator. 3. The apparatus of claim 2, wherein the bit allocation calculator comprises the average detector and the selector to separately process the peak signal-to-noise ratios of the intra frame, the predicted frame, and the bi-directional frame transmitted from the peak-to-noise ratio measurement unit. Bit allocation circuit for maintaining a uniform peak signal-to-noise ratio of an image.