WO2005039188A1

WO2005039188A1 - Technique for bit-accurate comfort noise addition

Info

Publication number: WO2005039188A1
Application number: PCT/US2004/033712
Authority: WO
Inventors: Jill Mac Donald Boyce; Cristina Gomila; Joan Llach; Alexandros Tourapis; Jeffrey Allen Cooper; Peng Yin
Original assignee: Thomson Licensing
Priority date: 2003-10-14
Filing date: 2004-10-12
Publication date: 2005-04-28
Also published as: CN1868216B; RU2340944C2; US8238613B2; BRPI0415404A; SG147440A1; HUE046111T2; EP1673944B1; AU2004306878A1; KR100989897B1; CA2540852A1; KR20060132813A; JP2007508789A; CA2540852C; CN1868216A; MXPA06004099A; MXPA06004097A; BRPI0415307A; CN1868215B; AU2004306878B2; PL1673944T3

Abstract

The addition of comfort noise to an image serves to hide compression artifacts. To facilitate comfort noise addition, supplemental information accompanying a video image contains at least one parameter that specifies an attribute regarding comfort noise. Typically, the supplemental information includes parameters that function to turn the comfort noise on and off, as well as to indicate the level of noise to add, based on the expected level of compression artifacts.

Description

TECHNIQUE FOR BIT- ACCURATE COMFORT NOISE ADDITION

CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority under 35 U.S.C. 119(e) to U.S. Provisional Patent

Application Serial No. 60/511,026, filed on October 14, 2003, the teachings of which are incorporated herein.

TECHNICAL FIELD

This invention relates to a technique for adding comfort noise to hide compression artifacts.

BACKGROUND ART

The decoding of a video stream compressed at low bit rate often yields visible artifacts noticeable to a viewer. Blockiness and structured noise patterns are common artifacts that arise when using block-based compression techniques. The human visual system has a greater sensitivity to certain types of artifacts, and thus, such artifacts appear more noticeable and objectionable than others. The addition of random noise to the decoded stream can reduce the noticeability of such compression artifacts, but large frame-to-frame differences created by adding random noise can itself produce artifacts that appear noticeable and objectionable. The addition of a dither signal can reduce human sensitivity to image artifacts, for example to hide contouring and blocking artifacts. One prior art technique has proposed adding a random noise dither that is based on film grain to an image to disguise block effects. The rationale for adding such random noise is that random error is more forgiving than structured or correlated error. Other prior art techniques have proposed adding a dither signal to a video stream to hide compression artifacts. One past technique has proposed adding a random noise dither in the video encoding and decoding process in the in loop deblocking filter for the ITU/ISO H. 264 video coding standard, commonly known as the JNT coding standard. The amount of dither to be added depends on the position of a pixel with respect to a block edge. Another prior technique has proposed adding that random noise subsequent to video decoding (i.e., adding noise as a "post process"), for use as comfort noise. The amount of noise added depends on the quantization parameter and on the amount of noise added to spatially neighboring pixels. The term "comfort noise" comes from the use of noise in audio compression to indicate noise pattern generated at the receiver end to avoid total silence that is uncomfortable to a listener. Past comfort noise addition techniques, while effective, have not afforded a desired level of control over the addition of comfort noise, as well as the level of noise to be added.

BRIEF SUMMARY OF THE INVENTION Briefly, in accordance with a preferred embodiment of the present principles, there is provided a technique for reducing artifacts in a video image. The method commences by the receipt of supplemental information that accompanies the video image. The supplemental information includes at least one parameter that specifies an attribute of comfort noise for addition to the video image. Temporally correlated comfort noise is generated and is then added to the image in accordance with the at least one parameter in the supplemental information.

BRIEF DESCRIPTION OF THE DRAWINGS FIGURE 1 depicts an apparatus in accordance with the present principles for generating comfort noise in connection with processing of the image block; and FIGURE 2 depicts an apparatus for use with the apparatus of FIG. 1 for adding comfort noise on a pixel-by-pixel basis.

DETAILED DESCRIPTION

In accordance with the present principles, the addition of comfort noise to an image serves to hide compression artifacts. To facilitate comfort noise addition in accordance with the present principles, supplemental information accompanying a video image contains at least one parameter that specifies an attribute regarding comfort noise. Typically, the supplemental information includes parameters that function to turn the comfort noise on and off, as well as to indicate the level of noise to add, based on the expected level of compression artifacts. In the illustrated embodiment, the video image typically undergoes compression in accordance with the H.264 video compression standard. Accordingly, the supplemental information containing the at least one comfort noise parameter is carried by a registered user data Supplemental Enhancement Information (SEI) message.

Comfort noise SEI message

The use of a registered user data SEI message serves to indicate the use of comfort noise. This message applies to all pictures that follow it until an IDR picture or a new comfort noise or film grain SEI message arrives. Comfort noise SEI messages can only precede I pictures, and only one comfort noise SEI message can precede a particular I picture. I pictures are indicated by slice_type equal to 7, or by nal_ref_idc equal to 5. Table 1 below lists the SEI message elements related to comfort noise:

TABLE 1

comfort_noise_flag equal to 1 indicates that comfort noise addition is used. comfort_noise_flag equal to 0 indicates that comfort noise addition is not used.

comfort_noise_qp_offset_idc indicates the quantization parameter offset used in the calculation of the additive comfort noise level, and may range in value from -51 to 52. comfort_noise_qp_ eight_offset_idc indicates a quantization parameter weight offset used in the calculation of the additive comfort noise level, and may range in value from -6 to 7.

Bit-accurate implementation of comfort noise

The addition of comfort noise in accordance with the present principles involves certain operations performed at the block level, and operations performed at the pixel level of a video image.

Block Level Operations

FIGURE 1 depicts an apparatus 300 in accordance with an illustrative embodiment of the present principles for performing operations on 8 x 8 blocks of luma pixels for the calculating the relative weights of the three noise value terms w₀, Wi_. and W that control comfort noise addition. The apparatus 300 of FIG. 1 includes a processing block 302 that receives decoded luma pixels, the picture QP = (pic_init_qp_minus26 + 26) and the comfort_noise_qp_offset_idc and comfort_noise_qp^weight_offset_idc parameters from the comfort noise SEI message. The term t indicates the current picture number, which undergoes a reset to 0 at the I picture that follows a comfort noise SEI message. The processing block 302 computes the average of the 8x8 luma pixel block of the current picture, t, as block_avg(t). A first threshold comparator 305 makes a comparison of this value to a threshold value, say 10. If block_avg(t) > 10, the comparator 305 sets the block_sum_level = 1, otherwise the value of the block_sum_level becomes 0. A delay 304 element delays value of block_avg(t) by one picture interval for input to an absolute difference block 306 which computes the absolute difference between the delayed value from the delay element 304 and the value of block_avg(t) directly received from the processing block 302. A second comparator 310 compares the difference computed by the absolute difference block 306 to a threshold value, say 3. If the threshold comparator determines that |block_avg(t) - block_sum(t-l)| > 3, the comparator 310 sets the value of the term block_absdiff_level = 0, otherwise block_absdiff _level = 1. If t is equal to 0, block_absdiff_level = 1. For Standard Definition (SD) television resolutions and below, all pixel and block operations occur using the display resolution. For High Definition resolutions, block operations occur using 2x2 sub-sampled pixels (i.e. using the upper left pixel of each 2x2 block), so the 8x8 luma pixel avg involves adding 8x8 = 64 pixel values, but these values are spread over a 16x16 pixel range. For SD resolutions, storage of the block_avg values for the entire picture requires storage of 1/64 the size of a frame store. For HD resolutions, storage of block_avg values requires a storage capacity of 1/256 the size of a frame store. A first look-up table (LUT) 312 receives the values of block_sum_level and block_absdiff_level from the threshold comparators 305 and 310, uses the values to find values for three sub-noise values sw₀, swi and sw_f, which as described in greater detail, undergo scaling by a scaling factor w₉ generated by a second look-up table 314, to yield the noise values wo, Wi and W_f. Table 2 depicts the relationship between the block_sum_level and block_absdiff_level and the sub-noise values sw₀, swi. and swf provided by the look up table 312 TABLE 2

The look-up table 314 receives the value of the picture QP, and the values of the parameters comfort_noise_qp_offset_idc, and comfort_noise_qp_weight_offset_idc from the SEI message, and uses those values to calculate the scaling factor w_g for scaling the sub- noise values sw₀, SW] and sw_f, to yield the noise values w_υ, wi and W , which are used in the pixel level operations performed by the apparatus 400 described hereinafter with respect to FIG. 2. Table 3 depicts the relationship between the value of Q and weight(Q) a precursor to the scaling factor w₉. TABLE 3

The scaling factor w_g is given by the relationship: w_q = clip(( weight (clip(QP +comfort_noise_qp_offset_idc,0,51)) + comfort_noise_qp_weight_offset_idc),0,7) As discussed, scaling the sub-noise values sw₀, swi and SW_f by the scaling factor w_g yields the noise values w₀, wi and W_f as follows: Wo = SWo * Wq Wi = SW! * W_q Wf = SWf * W_q Note that the determination of the current picture structure depends on the display rather than coding structure. For an interlaced display, pixels from the current field and previous field serve as the basis for calculations. For progressive display, pixels from the current frame and previous frame serve as the basis for calculations.

Pixel Level Operations

FIGURE 2 illustrates an apparatus 400 that adds comfort noise on a pixel-by-pixel basis in accordance with the noise values wo, wi and W_f obtained from the apparatus 300 of FIG. 1. As described hereinafter, the apparatus 400 makes use of the same pre-stored Gaussian random number list and primitive polynomial generator. The apparatus 400 includes three distinct primitive polynomial uniform random number pattern generators 402, 404, and 406. For SD resolution, the pixel level operations are performed for all luminance pixels, and the generated luma noise value is applied to both luma and chroma pixels. Under such circumstances, the relationship pic_width = PicWidthInSamples_L (as defined in the H.264 recommendation) applies. For HD resolutions, the pixel level operations are applied on ^lA of the pixels, using a 2x2 sub-sampling, and the generated noise value is applied to all pixels, using a 2x2 pixel repeat. Under such circumstances, the relationship pic_width = PicWidthInSamplesτ > 1 will apply. The first uniform random number pattern generator 402 initializes its seed value xo(i, seed_Xo(t)) with the seed value seed_xo(0)=3, upon the arrival of an SEI message containing comfort noise information. At each new frame t, the uniform random number pattern becomes initialized to seed_xo(t) where seed_x₀(t)= seed_x₀(t-l) + 2. The random noise generator 404 initiates its seed value (i, seed_.x^t)) with the seed value seed_xι(t)= seed_x₀(t-l). This implies the relationship that that xι(i, seed_x₁(t))= Xo(i, seed_xo(t-l)). The uniform random pattern number generators 402 and 404 will generate two new numbers UN₀(t,i) and UNι(t,i), respectively, where UNι_.(t,i)= UNo(t-l,i), in accordance with the following relationships:

UN₀(t,i) = x₀(i, seed_x₀(t)) % 32 - 16 UNι(t,i) = UN₀(t-l,i) = xι(i, seedjtι(t)) % 32 - 16

The third uniform random number generator 406 becomes initialized to 1 at the beginning of each displayed frame and serves to generate a fixed noise image for receipt by a random noise line generator 408. The random noise line generator 408 generates offsets into the Gaussian_list[ ] stored in the database 26 to access a line of 8 random numbers using the following operations, where i increments for each 8 values:

UN_f= x(i, l) for n=0..7, G[n] = (Gaussian_list[(UN_f + n)%2048] + 1) » 1

The outputs of the uniform noise generators 402 and 404 undergo scaling by a separate one of scaling blocks 410 and 412, whereas the output of the random noise line generator undergoes shifting by a shift register 414. Multipliers 416, 418 and 420 each serve to multiply the output of a separate one of the scaling blocks 410 and 412 and the shift register 414 by a separate one of the noise values w₀, wi, and w_f, respectively. A first summing block 422 sums the outputs of the multipliers 418 and 420, while a second summing block 424 sums the output of the first summing block 422 with the output of the multiplier 416. A shift register 426 shifts the output signal of the second summing block 426 to yield a noise value at position [r][s] of the noise image equal to: m = pic_width*r+s noise[r][s] = (w_f*G[s%8]+ w₀*UN₀(t, m) + Wι*UN₀(t-l, m) + 512 ) » 10 where WQ, WI, and W_f change at the block boundaries.

For SD sequences, the final noise luma_noise equals the added noise, while for HD sequences the final noise luma_noise is generated by performing a 2x2 upsampling of noise using pixel repetition.

The chroma noise is half the value of the final luma noise chroma_noise[r][s] = (luma_noise [r*2][s*2] + 1 ) » 1 A first summing amplifier 428 sums the luma-noise with decoded luma pixels and the resultant sum undergoes clipping within the range of [0, 255] by a first clipper 430 to yield display luma pixels. A shift register 431 further shifts the output signal of the shift register 426 to yield the chroma_noise signal for summing by the summing amplifier 432 with decoded chroma pixels. The resultant sum undergoes clipping within the range of [0, 255] by a second clipper 436 to yield display chroma pixels. The foregoing describes a technique for adding comfort noise to a video image. Appendix The list of the 2048 Gaussian distributed random numbers are:

char Gaussian[2048] = {

OxFB, 0x05, 0x33, OxFB, 0x14, OxEF, 0x06, OxlD, 0x26, 0x30, OxD5, 0x01, 0x20, 0xD9, 0x16, OxlB, 0xE7, OxOA, 0x06, OxFB, OxF6, 0xF7, 0x10, OxCl, 0x08, OxFE, OxCC, 0x09, 0x09, 0x23, 0x17, OxFB, OxED, 0x15, OxFF, 0x25, OxDF, OxlA, OxD3, 0x10, 0xE9, OxOA, OxFF, OxE5, 0x18, 0x00, 0xE4, 0xEC, 0x00, 0x3C, OxCl, OxCB, OxE8, 0x04, 0x07, Ox3F, Ox3D, 0x36, 0x19, Ox3F, 0x00, 0x03, 0x38, 0x09, OxOE, 0x06, 0x26, 0x38, 0x28, 0xE2, OxCl, 0x37, 0xE7, 0xF2, 0x01, OxE8, 0xF5, OxlD, 0xF2, OxDC, 0x05, 0x38, 0x21, 0x27, OxFF, 0xC7, OxD5, OxFE, OxFE, 0x14, OxlD, OxD8, 0x18, 0xF3, OxFl, OxEF, OxCC, 0x19, 0x08, 0xF4, OxEF, OxFA, 0xF9, OxCl, OxE5, OxF5, OxE5, OxCl, OxC8, 0x02, 0xF4, OxDC, 0x3F, Ox3F, OxFF, 0x14, 0x2B, OxEO, 0xF9, OxlB, 0x09, 0x2D, OxD8, OxEO, OxEO, Oxll, OxFD, OxE5, 0x31, OxFD, 0x2C, Ox3E, OxF3, 0x2D, 0x00, OxlF, OxlD, 0xF9, OxF5, 0x38, OxFO, 0x3 A, 0x06, OxOC, 0x19, OxF8, 0x35, OxFD, OxlA, 0x13, OxEF, 0x08, OxFD, 0x02, 0xD3, 0x03, OxlF, OxlF, 0xF9, 0x13, OxEE, 0x09, OxlB, 0x08, 0xE7, 0x13, 0x10, OxEE, Ox3E, OxED, OxC5, 0x08, OxFl, 0x00, 0x09, 0x31, OxlE, 0x32, OxFA, OxDC, OxF8, 0xE7, 0x31, 0x01, 0x01, OxlD, 0x10, OxFF, OxFF, 0x04, OxEC, OxCC, OxEE, 0x06, 0x3F, 0x07, OxCl, OxFl, OxD5, OxED, OxE5, 0x16, OxEC, 0x25, OxOB, 0xF7, 0xF5, OxDD, 0x25, 0xE6, 0x00, 0x10, OxEA, 0x08, 0xD2, OxlD, OxEO, OxDF, OxlB, OxCE, 0xF2, 0xD5, OxEF, 0xD2, 0x21, 0x02, OxDC, 0xE2, 0x2E, OxEB, 0x06, 0xF4, OxEE, OxCl, 0xF8, 0x07, OxCl, OxlF, Oxll, OxOF,

Ox2E, 0x08, 0xE7, OxE3, 0x23, 0x26, 0x28, Ox3F, Ox3F, OxlE, 0x10, OxCC, 0xD2, 0x00, 0x00, 0x25, OxDE, 0x23, Ox3F, 0xF7, 0xC9, OxOE, OxOB, 0x07, 0x01, 0x13, Ox2D, 0x02, 0x14, 0x00, OxFE, 0x13, 0x07, 0x38, 0xF2, OxEE, 0x19, 0x15, 0x35, OxOD, Ox3B, 0x03, 0xD9, OxOC, OxDE, 0xF6, 0x2E, OxFB, 0x00, 0x09, 0x14, 0xE7, 0x27, OxCl, OxEB, Ox3F, 0x08, 0x05, 0xF6, OxOF, OxE7, OxOD, 0xD4, OxD3, OxED, 0xF7, OxFC, OxOC, 0xC6, 0x23, 0xF4, OxEB, 0x00, 0x05, 0x2A, OxCB, 0x13, OxFO, OxCl, 0x17, 0x19, 0xF4, 0xF6, 0x16, 0x00, 0x07, OxEF, OxDE, 0x00, OxDC, OxOC, OxFD, 0x00, OxOE, OxFF, 0x16, 0x10, OxFO, 0x3A, OxEA, 0x27, OxF5, OxF8, OxCA, OxFB, OxDD, 0x2C, 0xE9, OxOB, OxD3, Ox3B, OxEE, 0x18, OxCl, OxlD, 0x10, OxD8, OxFB, OxF8, OxFD, 0x16, OxCl, 0xF9, 0x2C, 0x3F, 0x08, 0x31, OxED, OxFO, 0x12, 0x15, OxED, OxFl, OxF6, 0x34, 0xF7, 0x09, 0x09, 0xE3, OxFC, OxOF, 0x00, OxCl, 0x10, Ox3F, OxD6, 0x25, OxOB, OxEC, 0xE8, OxCl, OxCB, 0xF9, 0x16, OxDB, 0x00, OxOE, 0xF7, 0x14, OxDE, OxED, 0x06, 0x3F, OxFF, 0x02, OxOA, OxDC, OxE3, OxCl, OxFF, OxFF, 0xE6, OxFE, OxC5, 0x2E, Ox3B, OxD8, OxE8, 0x00, 0x09, OxEA, 0x21, 0x26, OxFA, 0xF6, OxCl, Oxll, OxEC, OxlB, Ox3B, OxFE, 0xC7, OxF5, 0x22, OxF9, OxD3, OxOC, 0xD7, OxEB, OxCl, 0x35, 0xF4, OxEE, 0x13, OxFD, OxFD, 0xD7, 0x02, 0xD5, 0x15, OxEF, 0x04, OxCl, 0x13, 0x22, 0x18, OxEl, 0x24, 0xE8, 0x36, OxF3, 0xD4, 0xE9, OxED, 0x16, 0x18, OxFF, OxlD, OxEC, 0x28, 0x04, OxCl, OxFC, 0xE4, OxE8, 0x3E, OxEO, 0x17, Oxll, Ox3A, 0x07, OxFB, OxDO, 0x36, 0x2F, 0xF8, 0xE5, 0x22, 0x03, OxFA, OxFE, 0x18, 0x12, OxEA, Ox3C, OxFl,

OxDA, 0x14, OxEA, 0x02, 0x01, 0x22, 0x08, 0xD9, 0x00, 0xD9, 0x02, Ox3F, 0x15, OxOD, Ox3F, OxCl, OxOD, OxE5, 0xF3, OxlB, 0x37, 0x17, 0x35, 0x00, OxDA, 0x00, OxlA, OxFC, 0xF5, OxEB, Ox3D, 0x36, 0x3F, 0x32, 0x21, 0x17, 0x02, 0x00, Ox3D, OxFA, OxE5, OxFO, OxE8, 0x2C, 0x20, OxCC, OxFE, 0x2F, OxE6, OxlF, 0x16, OxOE, 0x17, 0x09, OxEF, 0x07, 0x14, 0x17, OxDO, 0xF4, 0x2F, OxDB, Ox3F, 0xC7, Ox3F, OxDF, 0x00, OxF8, 0x19, OxDl, 0x17, 0x05, Oxll, OxEA, OxDB, 0x2C, OxCB, OxFC, 0xE4, 0xF2, OxCA, 0xF4, 0x3F, 0xE2, OxFA, 0x26, OxEA, 0x08, 0x09, 0x29, OxF5, 0x04, Ox3F, OxDF, OxlA, 0x01, OxOC, 0x06, 0x37, 0x15, OxC8, OxF5, 0x05, 0xF4, 0x29, 0x21, OxFA, 0x25, 0xC3, OxlD, Ox3F, OxFB, 0x31, 0xF7, OxlF, OxED, OxlA, 0x04, 0x03, OxlE, OxE5, 0x01, 0xE4, 0x38, OxCC, OxE3, 0x01, OxFC, 0xE9, 0x24, 0x2A, OxE5, OxEF, 0x06, Ox3B, OxOD, 0x2E, OxDD, 0x06, OxCF, OxDD, 0xF6, OxOE, 0x23, OxDl, 0x09, 0xE6, 0x20, OxFA, OxEl, 0xF4, 0x20, 0x24, OxFC, 0x3F, 0x00, OxCl, 0x33, OxF6, OxDC, OxC9, OxCD, OxFD, OxOE, OxEC, 0xF6, OxE3, 0xF2, 0xF4, 0x09, OxFE, 0xE7, 0x2F, OxE3, OxDl, OxEE, Oxll, 0x09, OxDE, Ox3F, 0xF7, OxCl, 0xF5, OxC5, 0xE6, 0x12, 0x25, OxCl, 0x00, OxFB, OxC5, OxE6, 0xF3, 0x13, 0x22, 0x08, 0x08, 0xC7, 0x2C, OxlF, OxOC, 0x12, 0xF5, 0x18, OxCE, OxFl, OxFC, OxDl, 0xE6, 0x02, 0x2E, OxF5, OxE8, OxFC, 0x19, 0x01, OxDB, 0xD4, OxFB, OxED, Ox3F, OxD5, OxF5, 0x09, OxOA, 0x38, 0x25, 0x19, OxFl, 0x2E, OxEl, 0x03, OxFB, 0x17, 0x12, 0x32, OxEB, OxF8, 0xE6, OxFD, OxEE, OxDA, OxFl, 0xF6, OxlF, OxOF, OxlF, OxOA, OxCl, OxOF, OxlF, 0x12, 0x33, 0xD6, OxFC, 0x26,

0x27, OxlD, 0xD9, OxED, Oxll, 0x04, 0x28, 0xF4, OxFC, 0x01, OxF8, 0x23, Ox3F, 0x29, OxD5, OxlB,

0x09, OxC5, OxC3, 0x12, 0x05, 0x3F, OxlC, OxE5, 0x38, 0x06, OxOC, 0x10, OxFA, 0xE9, OxOA, OxFA,

0x02, OxlC, OxOD, OxOC, OxOC, OxFB, OxEE, 0x12, 0xD2, 0x26, 0x28, 0x04, 0x19, 0x06, 0x21, OxFA,

0x00, 0x10, 0x16, OxDB, 0x10, OxED, 0xF5, OxE8, OxCl, 0xF3, OxOF, OxFC, Oxll, 0x06, 0x23, 0x06,

OxlC, 0x05, 0xE6, OxD6, OxlA, OxEA, OxEF, 0x00, 0x3F, 0x05, OxDF, OxEA, 0x17, 0xC7, 0x01, 0x05,

OxlC, OxEF, 0x3B, 0xF7, 0xE2, OxlA, 0xE3, OxCl, OxE8, OxF5, 0x01, OxFE, 0x08, OxD8, OxFE, Ox3F,

OxOC, 0x27, 0x21, OxlF, 0xF4, 0x06, OxEO, OxEE, OxCl, 0xF2, OxOA, OxEl, 0x20, 0xE6, OxEC, 0x36, OxEl, 0x07, 0xF6, 0x06, OxOE, OxEl, OxOA, OxOD, 0x2F, OxEA, OxE3, 0xC6, OxFC, 0x27, OxE8, OxOB, OxEB, OxF8, 0x17, 0xE9, 0xC4, OxEF, OxF2, 0xE6, OxEA, OxOE, Ox3F, OxFA, 0x18, OxFC, OxCl, 0x25, OxF3, 0xF5, 0x2C, OxlD, 0x05, OxDl, 0x28, OxE3, OxlD, OxlE, 0xF4, 0x14, 0xD3, OxFF, 0xF6, OxE3, OxEA, OxE3, 0xF5, OxE6, 0x23, 0xF2, 0x21, OxFl, OxF5, 0x07, 0xF8,.0xDF, 0xF4, 0xF2, 0xE2, 0x17, 0x12, 0x08, 0x07, OxEE, OxF5, OxFB, 0x04, OxF3, 0xF7, OxlD, 0x16, OxE8, 0xE9, OxFF, OxF6, OxD8,

OxOE, OxDF, OxCl, 0x25, 0x32, 0x02, OxF8, 0x30, Oxll, OxEO, 0x14, 0xE7, 0x03, OxE3, OxOB, 0xE4,

0xF7, OxF4, OxC5, OxDC, 0x2D, 0x07, 0xF9, 0x27, OxFO, 0xD9, OxCl, OxEF, 0x14, 0x26, 0xD7, 0x00,

OxlB, OxOB, OxDB, Ox3F, OxF8, 0xF6, 0x06, OxOF, OxlB, OxC8, OxCl, 0x2C, OxlB, OxlE, 0x06, OxlB, OxFA, 0xC8, 0xF9, OxOF, 0x18, OxDF, 0xF8, 0x2D, OxFC, 0x00, OxOA, 0x22, OxDD, 0x31, 0xF7, OxC8,

0x20, OxD3, OxFC, OxFC, OxDD, 0x3F, 0x19, OxD8, 0xE8, OxOC, OxlE, 0xE2, 0xC9, 0x03, OxEC, Ox3F, 0x2B, OxEO, 0x35, OxCl, OxFE, Oxll, 0xF9, 0x14, OxE8, 0x06, 0x06, 0x24, OxCE, OxF3, 0x26, Ox3F, OxFD, OxCE, 0x2C, 0x12, Ox3C, 0x2C, 0xC2, OxE3, 0x06, 0xD2, 0xC7, OxOA, OxDF, OxD5, OxDl, OxC5, 0x15, 0xF2, OxFl, 0x08, 0x02, 0xE6, 0xE2, OxOA, OxEB, 0x05, OxDA, OxE3, 0x06, OxOE, 0x01, 0x03, OxDC, 0x13, OxE3, OxFB, 0x36, OxE6, 0x14, 0x21, OxFA, OxCl, OxCl, OxE8, OxOB, OxOE, 0x17, 0x11, 0x2D, Oxl 1 , OxFO, 0x39, 0xE7, OxFO, 0xE7, 0x2D, 0x03, 0xD7, 0x24, 0xF4, OxCD, OxOC, OxFB, 0x26, 0x2A, 0x02, 0x21, OxD8, OxFA, OxF8, OxFO, 0xE8, 0x09, 0x19, OxOC, 0x04, OxlF, OxCD, OxFA, 0x12, Ox3F, 0x38, 0x30, Oxll, 0x00, OxFO, OxE5, Ox3F, OxC3, OxFO, OxlE, OxFD, Ox3B, OxFO, OxCl, OxE6, OxEB, OxlF, 0x01, OxFE, 0xF4, 0x23, 0xE4, OxFO, OxEB, OxEB, 0x10, 0xE4, OxCl, Ox3F, OxOC, OxEF, OxFB, 0x08, OxD8, OxOE, 0xE4, 0x14, OxCl, OxCl, OxOA, 0xE9, OxFB, OxEF, OxEl, 0xE7, OxFO, OxD8, 0x27, OxDA, OxDC, 0x04, OxOD, OxDC, OxFC, OxDB, 0xD6, 0xD6, 0χE4, OxOC, 0x27, OxFC, OxDO, Oxll, OxEO, 0x04, 0xE3, 0x07, 0x00, OxEC, 0x10, OxD5, OxEA, 0x08, OxFF, OxFC, OxlD, 0x13, 0x05, OxCA, OxED, OxOB, 0x10, 0x08, 0xF2, 0x01, 0x19, OxCA, OxFE, 0x32, 0x00, 0x20, OxOB, 0x00, 0x3F, OxlE, 0x16, OxOC, OxFl, 0x03, 0x04, OxFD, 0xE8, 0x31, 0x08, 0x15, 0x00, OxEC, 0x10, OxED, 0xE6, 0x05, OxCA, 0xF7, OxlC, OxCl, 0x22, OxOD, 0x19, 0x2E, 0x13, OxlE, 0xE7, 0x16, OxED, 0x06, 0x2A, 0x3C, OxOD, 0x21, 0x16, 0xC9, 0xD7, OxFF, OxOF, 0x12, 0x09, OxEE, OxlD, 0x23, 0x13, OxDA, OxE9, OxlD, OxD9, 0x03, OxEl, OxEF, OxFA, OxlE, 0x14, OxCl, 0x23, OxFE, OxOB, OxE5, 0x19, OxCl, 0x21, OxFE, OxEC, OxOE, OxEl, OxlD, OxFF, 0x00, 0xF7, OxEA, 0xD2, OxD8, OxDO, 0xF9, 0xE6, OxFB, OxFB, OxDA, 0x06, 0x00, 0x03, OxDF, OxCl, 0x3F, OxF3, OxOD, OxFA, 0x08, OxFA, OxF3, 0x00, 0x04, 0xE9, OxFO, OxF9, OxOD, OxFl, 0xE3, OxlD, 0x26, 0xC4, OxOD, 0x13, OxE5, OxEl, OxFl, 0xF6, OxEE, OxFl, OxFD, OxCl, 0xF4, 0xE2, 0x23, OxCl, 0x38, OxCl, Ox3F, 0x2B, OxFD, 0x39, 0x36, OxlA, 0x2B, OxCl, 0x01, 0x07, OxOB, 0x25, OxCC, 0xE7, 0x01, 0x24, OxD8, 0xC9, OxDB, 0x20, 0x28, OxOC, OxlA, Ox3F, OxEA, 0xE7, OxCD, OxEC, OxEO, 0xF2, 0x27, OxDF, 0x20, OxFO, OxFl, OxFD, Ox3F, 0x00, OxFA, OxE7, 0x21, OxF9, 0x02, 0xD2, OxOE, OxEF, OxFD, OxD3, 0xE4, OxFF, 0x12, 0x15, 0x16, OxFl, OxDE, OxFD, 0x12, 0x13, 0xE7, 0x15, OxD8, OxlD, 0x02, 0x3F, 0x06, OxlC, 0x21, 0x16, OxlD, OxEB, OxEB, 0x14, OxF9, OxC5, OxOC, 0x01, OxFB, 0x09, OxFA, 0x19, OxOE, 0x01, OxlB, OxE8, OxFB, 0x00, 0x01, 0x30, OxF7, OxOE, 0x14, 0x06, 0x15, 0x27, OxEA, OxlB, OxCB, OxEB, 0xF7, Ox3F, 0x07, OxFB, 0xF7, 0xD8, 0x29, OxEE, 0x26, OxCA, 0x07, 0x20, OxE8, 0x15, 0x05, 0x06, OxOD, OxOD, OxlE, OxlC, OxOF, OxOD, 0x35, 0xF7, OxlB, 0x06, 0x30, 0x02, OxFD, 0xE2, OxCD, 0x2F, 0x35, OxEB, OxlA, OxOD, 0xE9, OxFC, 0x34, 0xE6, 0x17, 0x2C, 0x33, OxFO, 0x13, OxEF, OxlB, 0x19, 0x23, OxDl, OxEF, OxD5, OxCB, 0xF7, OxFl, 0x04, 0xF7, 0x27, 0xF9, 0x26, 0x02, 0xF7, OxCB, 0x2A, OxOA, OxEA, OxED, OxEC, 0x04, 0xF2, 0x25, 0x17, OxDB, OxlE, OxCl, 0x3C, 0xC9, 0xE4, OxFl, 0x14, 0x03, 0x27, 0x25, 0x21, OxlC, 0x14, 0xF4,

OxOF, 0x12, 0xE9, OxEE, 0x15, OxDC, OxEE, OxlF, Ox3F, OxDE, 0xE7, 0x2C, OxFO, 0xE2, OxlD, OxE5, 0x15, 0x07, 0x02, OxDF, 0x06, OxD3, OxlF, OxOE, OxED, OxFF, 0x29, OxFF, OxED, OxD6, 0xD6, OxlC, Oxll, OxDE, 0xE2, OxOE, OxEE, OxDl, 0xD9, 0x02, OxOF, OxFE, OxFO, 0xD9, 0xF6, OxFC, OxDA, 0x16, 0x03, 0xD2, OxDD, 0x20, 0x04, 0xE8, Ox3F, OxDE, OxOC, OxFB, OxED, OxC7, OxlF, OxCl, OxCE, 0x02, OxFl, 0x37, OxOB, 0xE3, 0x20, OxCE, OxOD, OxEB, OxOA, 0xE3, 0xF3, OxDC, 0x01, 0xD2, 0x02, Ox3F, 0x02, 0x25, 0xD5, OxFC, OxEB, OxCE, Ox3F, 0x00, 0x3E, 0x2D, OxEl, 0x19, OxlC, 0x01, 0x28, OxCl, Ox3F, 0x27, Ox3F, 0xF2, OxOE, Ox3A, OxDB, OxF8, 0xE4, 0x34, 0x18, 0x16, OxOC, OxDD, 0x18, OxED, OxCB, OxOF, OxFO, 0x01, OxFB, 0x14, OxCl, 0x19, OxCC, OxEB, OxEE, 0x19, 0x00, 0x17, 0x2B, OxFC, 0x26, OxOD, OxEC, 0xF4, 0x2D, 0x2B, OxE5, 0x25, 0x05, 0x10, 0x26, OxlD, Ox3F, Ox3F, OxFD, OxDC, 0x18, OxFO, OxCB, OxEF, 0x12, OxlC, OxlA, OxF8, OxFE, 0x29, OxlA, OxCB, OxlA, 0xC2, OxOE, OxOB, OxlB, OxEB, OxD5, OxF8, OxFD, 0x17, OxOB, OxFC, 0x00, OxFA, 0x37, 0x25, OxOD, 0xE6, OxEE, OxFO, 0x13, OxOF, 0x21, 0x13, 0x13, OxEl, 0x12, 0x01, OxOA, OxFl, 0xE7, OxF3, OxlA, OxED, OxD5, OxOA, 0x19, 0x39, 0x09, OxD8, OxDE, 0x00, 0xF9, 0xE9, OxEA, OxFF, Ox3E, 0x08, OxFA, OxOB, OxD7, 0xD7, OxDE, 0xF7, OxEO, OxCl , 0x04, 0x28, OxE8, OxlE, 0x03, OxEE, OxEA, OxEB, OxlC, OxF3, 0x17, 0x09, 0xD6, 0x17, OxFA, 0x14, OxEE, OxDB, 0xE2, 0x2A, 0xD9, OxCl, 0x05, 0x19, 0x00, OxFF, 0x06, 0x17, 0x02, 0x09, 0xD9, 0xE5, 0xF3, 0x20, OxDD, 0x05, OxCB, 0x09, 0xF8, 0x05, OxFl, OxlF, 0xE5, 0x12,

0x25, 0xF8, 0x3F, OxDC, OxFO, 0xF2, OxC5, 0x34, 0x21, 0x35, OxCD, OxCC, 0x23, OxlE, 0x01, OxOB,

OxFF, 0x10, OxFE, 0xF9, OxDF, 0xF9, 0xF5, 0xE5, 0x07, OxEl, 0x25, OxlC, 0xC9, 0x00, 0x29, OxF3,

OxOA, 0x25, OxED, OxF8, OxFB, 0x20, 0xF8, OxCl, OxE5, OxEO, OxOF, 0x2F, Ox3A, 0x01, OxC8, OxFD, OxCA, OxEl, 0x30, 0x04, 0x19, 0x03, 0x25, OxF3, 0x24, 0x38, OxEE, 0xC9, 0x2F, 0xE7, OxOB, OxFA, OxF7, OxlB, OxOA, OxOB, 0x2D, 0x2D, OxOB, OxE8, 0x08, OxDB, OxOB, 0x04, OxE8, OxDO, OxEE, 0x18, OxEF, Oxll, OxCl, 0xD6, 0x15, Ox3F, OxF5, 0xF4, 0x2A, 0x29, OxEF, OxFO, OxFA, 0x36, 0x33, OxED, 0x19, OxDF, Oxll, 0x09, OxF5, 0x18, OxFl, Ox3F, 0x14, OxOC, 0xD2, OxFF, OxFF, 0x34, 0x01, 0xE4, OxF8, 0x03, Ox3F, OxF8, 0x3E, 0x21, 0x22, 0xE2, OxOF, OxEF, OxlA, OxE4, OxF5, 0x08, 0x15, OxEF, OxF3, 0xE4, OxDF, OxF6, OxFC, OxE8, 0x21, 0x06, 0x20, 0x02, 0x17, OxlB, Ox3F, OxDB, 0x16, 0x2C, OxEO, OxFA, OxDA, OxD8, OxD3, OxOB, OxOE, 0x10, OxED, OxD5, OxFO, 0x30, OxD3, 0x13, , 0x04, OxEl, OxFF, OxFB, 0x3F, OxE8, OxEE, 0xE5, OxOB, OxEF, OxEF, 0xE6, 0x2C, 0xD3, 0x00, 0x18, 0x26, OxFE, OxCl, 0x08, 0x16, OxFE, OxDC, 0x00, 0xE4, 0xF7, OxDC, OxOE, 0x2E, OxlD, 0x18, OxOA, 0x08, 0x37, OxC9, 0x10, 0xD7, 0x17, 0x17, OxFB, Oxll, OxD5, 0x15, OxlC, OxDO, 0x3F, 0xF8, 0x00, 0x00, OxED, OxCl, OxFF, 0x00, OxlF, 0x2E, 0x00, 0x12, OxEO, 0xE2, 0xF7, 0x13, OxCl, OxlC, 0x18, OxF8, Ox3F, 0x2C, OxEB, OxCA, 0xE7, OxF8, 0x03, OxEE, 0x22, 0x17, 0xF9, 0x35, 0x14, OxlC, 0x03, 0x09, 0x03, 0x01, 0x2B, 0xD4, 0xD2, 0xF8, 0xF6, 0xF5, 0x06, 0x03, OxFE, OxDA, 0xD3, OxFF, 0x03, OxEF, OxFE,

0x09, 0x01, OxC9, 0x02, OxDF, OxD8, Ox3C, 0xF7, OxFO, OxEE, OxD6, 0x3F, 0x21, 0x16, 0x08, 0x17

};

Claims

CLAIMS 1. A method for reducing subjective artifacts in a video image, comprising the steps of: receiving supplemental information that includes at least one parameter that specifies an attribute of comfort noise for addition to an image; generating the temporally correlated noise in accordance; and adding temporally correlated noise to the image to hide artifacts.

2. The method according to claim 1 further comprising the step of generating the temporally correlated noise by the steps of: obtaining a block pixel average; accessing a look-up table using the block pixel average and picture quantization parameters to obtain weights of temporal correlation factors for weighting the added noise.

3. The method according to claim 2 wherein the step of accessing a look-up table further comprises the step of accessing a look-up table containing Gaussian random numbers.

4. The method according to claim 2 wherein the step of adding temporally correlated noise includes the step of adding temporally correlated noise to one of luma or chroma pixels.

5. The method according to claim 4 further comprising the step of adding temporally correlated noise includes adding noise to both luma and chroma pixels.

6. Apparatus for reducing subjective artifacts in a video image, comprising: . means for receiving supplemental information that includes at least one parameter that specifies an attribute of comfort noise for addition to an image; means for generating the temporally correlated noise in accordance; and means for adding temporally correlated noise to the image to hide artifacts. 7 The apparatus according to claim 6 further comprising: means for obtaining a block pixel average; means for accessing a look-up table using the block pixel average and picture quantization parameters to obtain weights of temporal correlation factors for weighting the added noise. 8. The apparatus according to claim 6 wherein the look up contains Gaussian random numbers. 9. The apparatus according to claim 6 wherein the means for adding temporally correlated noise adds temporally correlated noise to one of luma or chroma pixels. 10. The apparatus according to claim 9 wherein the means for adding temporally correlated noise adds temporally correlated noise to both luma and chroma pixels.