KR100647202B1 - Low noise encoding and decoding method - Google Patents

Abstract

Adaptive digital image processor 22 is located before MPEG2 encoder 24. The processor receives a high definition (1920x1080 pixels per image) video signal for broadcast or storage and appropriately low-pass filters the signal. The signal is low-pass two-dimensional filtered 223 to remove encoding defects and associated noise. The video signal is then horizontally down-sampled 226 to produce a lower resolution hybrid signal (1280 × 1080 pixels per image). The receiver decodes and decompresses the hybrid signal. Hybrid signals are up-sampled to original resolution using existing hardware and software with software modifications.

Description

LOW NOISE ENCODING AND DECODING METHOD}

1 shows a configuration of a video compressor according to the present invention.

FIG. 2 is a detailed view of the unit 22 of FIG. 1.

3 shows one possible response in the adaptive filter included in the unit 22.

4 is a flow diagram of an exemplary transmission system using the present invention.

5 is a flow diagram of an exemplary receiving system using the present invention.

The present invention relates to image compression for reducing bandwidth requirements of digital video encoders.

The United States Federal Communications Commission (FCC) approves the Digital High Definition Television (HDTV) standard proposed by the Grand Alliance (GA), and clarifies the way for terrestrial digital television broadcasting in the United States. . The GA HDTV system applies the Moving Picture Experts Group (MPEG2) video compression standard. For details, see "Information Technology-General Coding of Video and Related Audio Information: Video" {ISO / IEC 13818-2: 1996 (E)}. By using state-of-the-art sophisticated video compression methods such as source processing, motion estimation and compensation, transform representation and statistical coding, the MPEG compression system can reduce the transmission bit rate by more than 50 times. A full HD signal for one second requires almost a billion bits before compression. As suggested by the GA specification, 1920 x 1080 pixels (picture elements) at 60 field speeds per second are compressed at 18 megabits per second for digital broadcast.

A GA video compression system generally includes two main subsystems, the two main subsystems being a pre-processor and an MPEG2 video encoder, followed by an output buffer. The input to the pre-processor is analog video in RGB form. The pre-processor digitizes the input signal and performs gamma correction on each color component to compensate for the non-linear response of the imaging camera. Gamma correction reduces the visibility of the compressed image, especially the quantization noise included in the shaded region of the image. Next, the pre-processor linearly converts the digitized, gamma corrected RGB sample to the SMPTE 240MYC1C2 color space. Finally, the resulting chrominance components are sub-sampled to form a 4: 2: 0 digital video input signal. In addition to the functions just described, the pre-processor may perform image conversion. For example, in a broadcast digital satellite system, video signals are horizontally decimated from 720 pixels per line to 544 pixels per line to further reduce bandwidth requirements. This signal is sent to the MPEG2 video encoder.

The MPEG2 video encoder compresses the input digital video signal by removing some temporal overlap between frames and some spatial overlap within the frame. In general, compression is achieved by successively using several different techniques as described above. By adjusting the quantization accuracy, the encoder can generate a compressed bitstream at any rate specific to the application. Quantization in the MPEG2 system is performed on DCT coefficients of a data block, which may be initial image information or residual information from motion estimation. By using a quantization matrix combined with a scalable quantization step size, the quantizer quantizes only a small fraction of the DCT coefficients from all DCT blocks for transmission, resulting in significant data reduction. Bring. The quantization matrix can vary based on the frame depending on the statistical variance of the DCT coefficients and the content of the video signal. For other regions in the frame, the quantization can be finely adjusted for each macroblock by adjusting the quantization step size based on the complexity of the macroblock. The output buffer will provide the control signal used in the encoder at a predetermined output bit rate to adjust the quantization step size for a particular frame to maximize the quantization resolution within the effective bandwidth.

Ideally, the video compression system removes high frequency components that will not be recognized because the viewer cannot see them when the image is reconstructed and displayed. The remaining low frequency components are quantized to fit within the effective bandwidth. Also, the quantization noise embedded in the signal is invisible to the viewer even when the image is reconstructed. In a real system, however, there is a trade-off between the information to transmit and the quantization step size for the effective bandwidth. If the system does not reduce enough coefficients for quantization, the system increases the quantization step size causing block-like defects in the reconstructed image. If the image loses too much high frequency information during the compression process, the reconstructed image will have other significant edge defects.

In addition, the quantization difference between each frame allows a frame within a picture group (GOP) to contain variable high frequency components. In one example, an I frame can have a significant amount of high frequency coefficients reduced during encoding, while P and B frames have high frequency components corresponding to the reduced high frequency coefficients in the I frame. The reconstructed GOP will now have defects because the high frequency information varies between the frames used to reconstruct each other.

This problem occurs within the GA system, as is generally defined. Compressing the HD video signal will only further degrade the displayed picture quality. Satellite broadcast providers do not prefer HD signal transmission because only one program can be sent to a transponder at a time. Until now, a method of sufficiently compressing HD programs to simultaneously fit two programs to one satellite channel (eg, 24 MHz 4-PSK modulation) has resulted in unacceptable degradation of image quality. Therefore, satellite broadcast providers are reluctant to broadcast HDTV due to inefficient channel usage. Similarly, terrestrial broadcast providers will not attempt to provide full HD programs if one program fully occupies one channel where several SD programs may exist.

In accordance with the principles of the present invention, the digital image processor identifies the video signal type and optionally converts the original signal format to another format, if necessary. The converted signal is filtered and, if necessary, inversely converted to the original signal. The filtered signal is converted to a lower resolution and compressed at the target bit rate. Finally, the compressed signal is delivered to the output data channel.

An MPEG2 encoder comprising an apparatus according to the principles of the present invention has a two-dimensional (eg vertical and horizontal) filter prior to the encoder. Encoders, output buffers, and filters each generate information that can be used in other units to improve overall efficiency. This information relates to, for example, image movement, image contrast, quantization matrix selection, scale factor selection, bit rate of each unit, and image texture. The information is transferred between the units via a controller that monitors the encoding process or through each controller present in each unit.

The controller evaluates the incoming information and identifies commonalities for groups of pictures, frames, or partial frames, the commons of which operate the filters and / or encoders to efficiently encode groups, frames, or partial frames according to target bit rates. Can be effectively used to change In general, the filter is tuned because adjusting the filter generates less noise than adjusting the encoder. Also, a filter is actually a set of filters that allows for the best flexibility by adjusting each filter coefficient, if necessary. These filters are horizontal anti-aliasing low-pass filters, vertical low-pass filters, and two-dimensional low-pass filters, and are generally provided only in sequential order. The controller evaluates the received information for current filter and encoder settings and adjusts one or more filters and / or encoders according to one or more important commonalities. As a result, the input signal is generally low-passed through the filter in a manner that allows the encoder to encode the image uniformly through groups of pictures, frames, or partial frames with respect to the significant commonalities of the uniformly encoded data. Is filtered.

The encoded signal can be transmitted with an effective bandwidth and then reconstructed so that it can be displayed without a defect, otherwise there may be a defect. For high definition signals with 1920 x 1080 pixels per image frame, the horizontal resolution is reduced to 1280 pixels per line before encoding to further reduce the bandwidth of the transmitted signal after filtering. The result is a hybrid video resolution that the HD receiver can receive, decode, and display with small software changes.

An exemplary configuration of a video compression system according to the present invention is shown in FIG. An input video signal is received at the movie detector 20, which identifies whether the signal is a movie (film) signal reconstructed from 24 frames per second to 30 frames per second via the telecine method. The reconstructed cinematic signal is directed to the corresponding section of the adaptive image processor 22 as described. If the input signal is not a reconstructed cinematic signal, the signal is passed to another portion of the adaptive image processor 22. The identification of the telecine movie signal is made through known methods.

The processor 22 receives control information from the output buffer 26 and the MPEG2 encoder 24 via the controller 28, and the encoder 24 efficiently encodes the frame within the effective bit rate without noticeable defects. To filter the video frames. The processor 22 needs to improve the reconstructed picture quality of the MPEG2 encoded bitstream constrained by the average bit rate, so that the signal is two-dimensional (2-D) (e.g., horizontal and vertical). To filter. This is to improve MPEG2 coding efficiency in a manner that minimizes damage to the MPEG2 reconstructed picture with respect to picture clarity and encoding defects by changing the local 2D frequency content of the source. The filtering of the signal may be performed on predetermined data such as, for example, a group of pictures (GOP) or a frame, a single frame, or each pixel.

The 2-D filter low-pass filters the image. Optimally, the high frequency information that is removed is either redundant or invisible to the viewer. In practice, some high frequency information visible to the viewer can be removed to obtain the required bit rate. However, a system including a processor 22 before MPEG2 encoding produces a better image than a system without a processor 22 as described.

The filtered signal is encoded via MPEG2 encoder 24, which receives image parameters from processor 22 and output buffer 26 via controller 28 and adjusts the MPEG2 compression to match the effective bit rate. do. Compression occurs in the same way as described in the GA specification. Encoder 24 sends the compressed data to output buffer 26. The buffer 26 is encoded, modulated using known signal processing techniques, and allows compressed data transmitted over the transmission channel to be delivered at a predetermined rate. Prior to modulation, the compressed signal can be sent to a statistical multiplexer to be multiplexed through several programs for transmission on a single channel. The signal processing unit following the buffer 26 is well known and is not shown in FIG. 1 for the sake of simplicity.

The video compression system can be configured to accept any type of video signal. The system of FIG. 1 is configured to accommodate both televisions (cameras) and movies (films) formed to known industry standards. For the system of FIG. 1, one common configuration, for example, is that it is configured to receive output from the pre-processor as already described in the background. The system can be configured to accept other types of video signals by adding appropriate hardware and / or software. This configuration is not shown to simplify FIG. 1.

The movie detector 20 recognizes that there is a specific relationship to the input signal that can be used to improve the coding efficiency: that is, a source interlaced at (type 1) 60 fields / second, (type 2) 60 fields / second Interlaced 30 frames / sec film, (Type 3) Interlaced scans at 60 fields / sec, 24 frames per second film, (Type 4) Sequentially scanned sources, (Type 5) 30 frames / sec sequentially scanned at 30 frames / sec Specific relationships such as movies, and 24 frames / second movie sequentially scanned at 60 frames / second (type 6). Detection occurs in response to an external control signal (not shown) or through known techniques such as those used in current standard definition (SD) MPEG2 encoders. The signal format information is conveyed to the adaptive image processor 22 with the signal as described later. The movie detector 20 also detects whether the signal is interlaced scan or sequential scan type, and passes the information to the processor 22. This type of scan is exemplary and defines the parameters to which the signal is directed through the processor 22. Implementations for other fields and frame rates may also be used.

Adaptive image processor 22 performs several programmable functions to reduce the amount of data to be compressed by encoder 24. Processor 22 generally operates for each frame so that the processed frame can be best encoded to remove or greatly reduce the noise visible to the viewer. Processor 22 can be considered as a spatially variable 2-D low-pass filter, because the processor spatially down-samples each image frame and fits the selected 2-D high frequency component from the signal. Because it filters. Adaptive filtering can be adjusted over a series of frames, for a single frame, or for each pixel to produce a processed frame.

Processor 22 may facilitate encoding for any type of signal. However, in this embodiment, processor 22 is programmed to operate on HD data as defined in the GA specification. This may be 1920 × 1080 pixels per image or 1280 × 720 pixels per image. According to the GA specification, each HD format requires almost 18 megabits per second for broadcast. To simplify the discussion, only the 1920 × 1080 format will be discussed in detail. This discussion is equally applicable to the 1280 × 720 format or any other format.

2 shows a detailed view of the adaptive image processor 22. According to the signal format information received from the detector 20, the image signal is passed through the controller 28 (FIG. 1) to the interlaced-to-sequential converter 221 (type 1) and the de-telecine unit 222. Pass through the spatial band-limiting, low-pass filter 223 without being sent to (types 2,3) or altered (types 4-6). Filter 223 receives the output of these units after units 221 and 222 have processed the signals.

The converter 221 receives the signal and converts the signal into sequential frames at a rate of 60 frames per second if the format of the signal includes a 60 Hz interlaced field. A sequential frame contains all the image information in each frame. In general, filtering sequential scan signals does not introduce defects such as may occur when filtering field information of interlaced signals. Converter 221 uses known methods to convert interlaced fields into sequential frames.

The de-telecine unit 222 removes duplicate fields of the 60 Hz interlaced scanned movie and reconstructs the original sequential scanned movie. The sequential format ensures that subsequent vertical low-pass filtering does not produce motion artifacts. If the movie source (type 2 or type 3) input was treated as a type 1 source, vertical low-pass filtering degrades the performance of the MPEG2 encoder to detect and properly process movie source material. Coding efficiency worsens. Unit 222 converts the signal to sequential format and removes duplicate fields / frames before filtering, because the filtering may filter the duplicate information differently. If duplicate information is not removed before filtering, the information may not be the same after filtering, and the encoder may not recognize the signal as a type 2/3 signal. The encoder then encodes the information otherwise the information will be removed due to duplication.

In addition, the design of processor 22 is simplified by providing a single output clock from unit 222. If the unit 222 provides an output sequential film image at a rate of 24 frames per second and a rate of 30 frames per second, two output clocks and supporting circuitry will be required.

The initially generated signal in sequential format of 30 frames per second is passed directly to filter 223. The filter 223 requires video information represented as a complete picture frame. Spatial low-pass filter 223 is actually a filter set. In one example, the first filter is a semi-aliasing horizontal low-pass filter. The second filter is a vertical low pass filter. The final filter is the previously described 2-D low-pass filter. The coefficient of each filter tap can be suitably set according to the control information from the encoder 24 and the buffer 26 as shown in FIG. The sequential signal is horizontally low-pass filtered to remove aliasing resulting from continuous down-sampling at the sampling rate converter 226. The final horizontal output of 1920 pixels per line will be 1280 pixels per line as described. To remove the alias noise in the final signal, the low pass filter 223 has a cutoff frequency of 640 cycles per line. The horizontal anti-alias filter included in unit 223 may be a 17 tap finite impulse response (FIR) filter having the following tap coefficients.

[f0, f1, ..., f15, f16] = [-4,10,0, -30,48,0, -128,276,680,276, -128,0,48, -30,0, 10, -4] / 1024

Encoding an HD video signal at a reduced bit rate generally requires additional vertical low-pass filtering to further reduce the bandwidth of the video signal. Removing vertical high frequency energy before MPEG encoding is necessary to achieve acceptable overall picture quality. The vertical frequency range with the highest phase sensitivity is attenuated. The vertical cutoff frequency is set as part of the Nyquist frequency. As an example, a cutoff frequency that is nearly one half of the line rate of the HD input signal may be appropriate for some video material. For HD signals with 1080 lines / picture height (l / ph), this corresponds to a blocking of 540 l / ph. This frequency may be programmable and the programmable cutoff frequency will be determined from the parameters available from encoder 24 and buffer 26 of FIG. 1 via controller 28 (ie, required bit rate, quantization matrix, etc.). . The vertical low pass filter included in the unit 223 may be seventeen tap FIR filters having the following tap coefficients.

[f0, f1, ..., f15, f16] = [-4, -7,14,28, -27, -81,37,316,472,316,37, -81, -27,28,1 4, -7,- 4] / 1024

Alternatively, the cutoff frequency may be twice the line rate of the SD signal. In general, the vertical low-pass filter follows the horizontal anti-alias filter.

Processor 22 performs vertical filtering rather than vertical portion removal, thereby maintaining a constant vertical line resolution. In general, filtering is more appropriate than interpolation for interlaced video signals. Converting the vertical line resolution for interlaced image sequences requires complex hardware and software, resulting in expensive receiver costs. Vertical sample rate conversion degrades vertical high frequency performance because of the increased complexity of taps combined with Nyquist sampling (ie, no oversampling). Cost considerations for the receiver generally counteract the reduction in vertical resolution to reduce defects and encoded bit rates. The displayed picture will be significantly degraded by using the current technology in a vertical sampling rate converter instead of the vertical low-pass filter described above. However, an effective and cost effective vertical sample rate converter may replace the vertical filter described herein without departing from the principles of the present invention.

The coefficients for both the horizontal and vertical low-pass filters can be changed via software and, if necessary, applied to the pixel level to obtain the target bit rate without creating defects in the reconstructed image. In general, coefficient changes based on frames are possible. An alternative to the slower processor is to program a number of different coefficient sets for the filter and select the most appropriate set for the image information to be processed. The greater flexibility of the adaptive filter allows the entire system to generate a data stream with little defect on the system without the adaptive filter.

After the signal is low-pass filtered in the horizontal and vertical directions by the unit 223, the controller 28 is able to encode the signal uniformly through the encoder 24 based on the frame without generating significant quantization noise. Decide if you can. If so, the signal is delivered to either of the units 224, 225 or 226 depending on the format of the signal as described. However, if the encoding process is likely to cause noise and / or defects in the signal, the signal is sent to a two-dimensional lowpass filter in unit 223 for further adaptive filtering. Control parameters from processor 22, encoder 24, and output buffer 26 (FIG. 1) allow controller 28 (each unit controller in unit 223) to determine if additional filtering is needed. do. In one example, the parameters used to make this determination are measurements of shift and contrast, an effective quantization table, encoding efficiency and the current target bit rate.

The 2-D filter in unit 223 preferentially reduces the high frequency information from the image frame along the diagonal, not only in the horizontal or vertical direction. The human eye is very sensitive to high frequency noise in the vertical and horizontal directions compared to the diagonal direction. Diagonal removal of significant high frequency information in order to uniformize quantization through encoder 24 generally results in a better quality signal with less observable noise. As with all previous filtering, the diagonal filter operates and is programmable over the entire picture frame.

The diagonal filter may be compatible with the quantization matrix at the encoder. Quantization matrices often use diamond shaped matrices for I frame quantization. However, such matrices often contain noise because B and P frames use other types of quantization matrices that retain high frequency components during the compression and motion compensation processing occurring at encoder 24. The filter of processor 22 removes high frequency information from each video frame before MPEG2 encoder 24 processes the data in I, P, and B frames into the motion estimation network. Therefore, high frequency components are generally removed from P and B frames as well as I frames. During reconstruction, the picture generally does not have the defect created by MPEG2 encoding as known.

Referring to FIG. 1, in practice, the controller 28 evaluates a signal parameter (eg, shift, contrast, etc.) before filtering the provided frame through the processor 22, and includes a unit that includes the necessary diagonal filtering. Determine coefficient settings for all filters in (223). During the filtering process of a given frame, controller 28 monitors the signal parameters from processor 22, encoder 24 and buffer 26 and counts as needed to maintain a target bit rate with minimal defects / noise. Change Each frame is filtered based on the most recent signal parameter, passed to encoder 24 for compression, and then to buffer 26 when subsequent information is entered into filter 223.

If the signal is generated with a movie signal of 24 frames per second, the filtered signal is applied to a 3: 2 pull down unit 224. Unit 224 doubles the frame selected to provide an output signal of 30 frames per second. This happens through known methods. The signal is then transmitted from unit 224 to horizontal down-sampling converter 226.

Field sub-sampling unit 225 converts the sequential signal from filter 223 into a sequential-interlaced scan format. This conversion is accomplished through known methods. Without inversely converting to interlaced format, the signal can acquire twice as much data as the total data, because the sequential frame rate from unit 221 is 60 Hz. The interlaced signal is applied to the transducer 226.

Sampling rate converter 226 receives the sequential signal from filter 223 directly at 30 frames per second. Units 224 and 225 also provide the signal to transducer 226, as described above. Converter 226 down-samples the HD signal to the selected transmission format. This format does not have to be a standard format. What is needed may be any image ratio and frame size. However, non-standard formats will require receiver modifications.

When converter 226 receives a 1920 × 1080 GA HDTV signal, converter 226 down-samples the horizontal information and outputs a 1280 × 1080 hybrid pixel frame format. The GA HDTV compatible receiver may receive an image frame including 1920 × 1080 pixels and 1280 × 720 pixels. Therefore, the GA compatible receiver can be modified to support 1280 pixel horizontal resolution and 1080 pixel vertical resolution. The compatible receiver hardware up-samples 1280 horizontal pixels to 1920 horizontal pixels in connection with an increase in vertical resolution. However, a GA compatible receiver is not needed to receive a video frame resolution of 1280 x 1080 pixels (horizontal x vertical) and is also not programmed. The hardware is in the right place to receive and decode these resolutions in the defined format, but support software must be added to decode and increase only the horizontal resolution. Adding software is simpler and cheaper than redesigning and adding new hardware needed for other non-standard formats.

Processor 22 provides a hybrid 1280 × 1080 format because current display technology cannot display 1920 pixel / line resolution. In general, the best television monitors can display the resolution at resolutions of nearly 1200 pixels / line to 1300 pixels / line. Therefore, limiting the output resolution in the horizontal direction to 1280 pixels / line has almost no adverse effect on image quality. By providing a display resolution (1280x1080) supported by existing decoding and decompression receiver hardware, receiver manufacturers are minimally affected because only software changes are needed. For certain receivers, such as broadcast satellites, software changes can be downloaded via satellite links and installed remotely. There is no need for a service technician to be involved with such a receiver.

The hybrid format is advantageous because terrestrial and satellite program providers are reluctant to transmit HD programs. Satellite transponders transmit data streams of nearly 24 megabits per second (Mbps). Terrestrial HDTV broadcasts can carry up to 19 Mbps, including 18 Mbps HD programs and other information (such as audio, program guides, conditional access, etc.). Current satellite transponders deliver at most one HDTV program each, and satellite program providers' claimed HDTV programs are not profitable enough. Simply reducing the horizontal frame resolution from 1920 to 1280 is not enough to enable the continuous transmission of two HD programs through a single satellite transponder. The filtering provided by processor 22 advantageously allows this dual HD transmission over a single channel.

The filtering characteristics provided by the processor 22 can take many forms, including diamonds, intersections, and hyperbolic from axis to axis, in which case the filtering is diagonal for each filter. One possible form, 2-D hyperbola, is particularly advantageous in this application and has an amplitude-frequency response as described in FIG. 3. The adjustable filter cutoff frequency is generally set such that the GOP, frame or selected partial frame is uniformly compressed by the encoder 24. If necessary, additional horizontal and vertical high frequency information can be filtered, but this is not generally required. As the picture complexity changes, or as the effective bit rate increases, the amount of data filtered by the diagonal filter and other previous previous filters decreases. The 2-D filter may be described as, for example, a two-dimensional FIR (13 × 13), or 2-D infinite impulse response (IIR) filter with 13 taps in each direction.

The 2-D FIR filter included in unit 223 may be a 13 × 13 tap filter with the following tap coefficients:

0 0 0 1 -3 5 -5 5 -3 1 0 0 0

0 0 1 -1 -3 9 -11 9 -3 -1 1 0 0

 0 1 -1 -4 4 6 -13 6 4 -4 -1 1 0

 1 -1 -4 1 14 -13 4 -13 14 1 -4 -1 1

-3 -3 4 14 5 -47 61 -47 5 14 4 -3 -3

 5 9 6 -13 -47 -55 193 -55 -47 -13 6 9 5

-5 -11 -13 4 61 193 556 193 61 4 -13 -11 -5

 5 9 6 -13 -47 -55 193 -55 -47 -13 6 9 5

-3 -3 4 14 5 -47 61 -47 5 14 4 -3 -3

 1 -1 -4 1 14 -13 4 -13 14 1 -4 -1 1

 0 1 -1 -4 4 6 -13 6 4 -4 -1 1 0

 0 0 1 -1 -3 9 -11 9 -3 -1 1 0 0

 0 0 0 1 -3 5 -5 5 -3 1 0 0 0

For this coefficient, the DC gain is 1024. Coefficients represent octant symmetry giving 28 independent coefficients. The symmetric coefficient range allows for faster setting of the adjustable filter. However, as an example, when the filtered image or range exhibits different characteristics in one portion of the image, each octant may be different.

The filter response of processor 22 may vary continuously from one set of coefficients to another based on each pixel. Thus, processor 22 may exhibit different operating parameters to maintain good image quality under bit rate constraints as described.

As mentioned previously, the processor 22 may be modified as appropriate for the filter depending on the parameter (s) used to define the filter suitability. For example, variations in image frames can be used to segment the image into ranges for other processing. An edge is an important image feature where an important edge masks coding errors in the vicinity, and the edges can also define an image range. Colorimetry can be used to identify low complexity areas such as texture and sky. Textures can also be identified and processed as regions. Texture is generally less important than edges. Therefore, the texture identifies a range that can be filtered more than other ranges. In addition, components relating to movies can be used to locate important features or operations that require higher coding efficiency and hence less filtering. The background is generally softened by the depth of the camera optics field and can be filtered more deeply. Pan and scanning information may be used to define the center of the relevant image for other processing through processor 22.

The operation of encoder 24 is compatible with the MPEG2 standard. Encoder 24 may provide information via controller 28 that processor 22 may use to improve performance. Such information may include, for example, bit rate information. Such bit rate information may include the average bit rate, frame bit rate, and macro-block or block bit rate of the GOP. Other information that may improve the performance of the processor 22 includes discrete cosine transform complexity, the type quantization matrix to be used, and the quantization matrix step size to be used. In addition, processor 22 may provide information to encoder 24 through controller 28 to adjust the operation of the processor to improve encoding performance.

After the HD signal is formed in the transport packet datastream using known techniques, the HD signal is transmitted to the receiver in a known manner, as described, for example, in the Grand Alliance specification. Except for the up-sampling required for full HD pixel resolution at the receiver, the signal processing provided by processor 22 is clarified to the decoder in the Grand Alliance compatible receiver.

At the receiver, the datastream is demodulated and the transport stream is processed to recover data packets and program information using known techniques. For the above-described hybrid format HD program, the signal is up-sampled in the horizontal direction in the display processor when the display needs the complete HD signal. The number of vertical lines in the video signal is unchanged. This reconstructs a complete HD signal with 1920 × 1080 resolution for display through a high definition image playback device. If the video display device requires less full HD signals, the signals are appropriately down-sampled during video reconstruction prior to display through known methods. Existing grand alliance compatible receivers require software modifications to reconstruct the hybrid signal. The software change allows the horizontal and vertical hardware and software processing routines assigned to the grand alliance normalization mode to be independently selected when the incoming signal is needed.

4 is a flowchart illustrating a flow of a video image signal through an encoding system. In step 30, the signal format is identified and the identification information is conveyed with the video signal. The format information may indicate, for example, whether the signal is the first signal from a movie having a format of 24 frames per second, and whether the signal is interlaced or sequential. If the video signal is interlaced, the signal is converted into a sequential signal at 60 frames per second in step 31. If the signal was converted by 3: 2 pull down, the duplicate frame is removed in step 32. If the video signal is already in camera sequential format, the video signal is passed directly to the filter of step 34. In step 34, the video signal is spatially low-pass filtered. The filtering includes vertical, horizontal and diagonal filtering as described above. In step 35, the sequential signal converted from the interlaced signal in step 31 is reconverted into an interlaced signal. In step 36, the signal is pulled down 3: 2 to recover duplicate frames that were previously deleted in step 32. The video signal transmitted directly from step 30 to step 34 above is now passed directly from step 34 to step 38. In step 38, the video signal from any of steps 34, 35 and 36 is subsampled to the hybrid 1280x1080 HD signal resolution defined above or to another desired output format. In step 40, the hybrid signal is MPEG2 encoded as previously described. Step 42 transmits the encoded signal, and in step 44 modulates the transmitted signal as required for transmission over an output channel, such as an RF terrestrial broadcast channel. Finally, step 46 transmits the modulated signal. Steps 42 to 46 occur through known methods.

5 is a flowchart illustrating a flow of an image signal transmitted through a receiver. This flowchart assumes that the image resolution of the received signal is a hybrid 1280 × 1080 HD signal as defined above. In step 50, the transmitted signal is received and demodulated by the tuner. The demodulated signal is MPEG2 decoded and decompressed in step 52. In step 54, the video signal resolution is identified at 1280 x 1080 pixels per image through the control information sent with the signal. The Grand Alliance MPEG2 protocol provides image resolution information along with the transmitted signal. In general, a hybrid signal is identified by a particular unique code, such as any other defined resolution. Hybrid signals may also be defined differently through information in user data of transmitted data, for example. The hybrid video signal is up-sampled horizontally into a full 1920 × 1080 HD signal during display processing in step 56. The hybrid signal is up-sampled with new software with respect to existing hardware and software present in the receiver as previously described. Finally, the complete HD video signal is displayed on the 1920 × 1080 display in step 58. Steps 50, 52 and 58 use known methods.

The apparatus and method described above can be applied in various configurations to achieve improved image reconstruction for high definition displays. Adaptive and non-adaptable options may be used depending on the requirements of the particular system. Some of these choices are discussed below.

The non-adaptive method sets the frame filtering of the processor 22 to the target bit rate and allows all images to be processed uniformly. Another non-adaptation method assumes that the center of the displayed image is the most important area. The non-adaptive method also assumes that the perimeter of the image is less important and therefore less important to the viewer. The filter coefficients of the processor 22 are set by the controller 28 via parameters that are a function of the spatial position of the pixel, and all image information is processed uniformly.

The adaptation option splits the image into several regions using texture model parameters, local video variation, colorimetry, or other picture complexity measures based on the source image. The filtering characteristics of the processor 22 are appropriately changed for different regions.

Another solution is to suitably change the filtering characteristics of the processor 22 as a function of the difference between the actual bit rate and the target bit rate. In this case, a single parameter controls the change in filter coefficients for the 2-D frequency response.

Another method is to design the 2-D frequency response of the filtering provided by processor 22 to be compatible with the quantization matrix used by encoder 24. The quantization matrix can be considered as a lowpass filter with a 2-D form. For this method, the value of the filter coefficient will be a function of the quantization matrix step size. Since the step size changes in accordance with known encoder operation, corresponding changes will occur for the corresponding filter coefficients.

The above mentioned options describe the flexibility of the system using the principles of the present invention. Such a system preferably operates within an MPEG2 rate control situation to extend MPEG2 compression capability by reducing encoding defects and other noise. The flexibility and economics of HDTV deployment are improved through the use of the present invention. The number of HD programs per transponder transmitted to a direct broadcast satellite system (ie 24 MHz 4-PSK) increases from one program to two programs, or one HD program has several SD programs. The ability to transmit one HD program with several SD programs on a 6 MHz terrestrial broadcast channel can be obtained according to the principles of the present invention. Previously, broadcast stations were limited to transmitting one HD program on one channel or multiple SD programs on one channel.

Although the present invention describes a system for transmitting and receiving HD signals, the principles of the present invention can be applied to other devices such as data storage systems. In a system such as a digital video disc (DVD), video data is encoded and stored for later replay. The storage medium has a limited amount of effective storage space. If an encoded program, movie, or other video sequence has exceeded the effective amount of space in the medium, additional encoding / compression to make the program fit can cause unacceptable defects. The present invention described above can be used to efficiently encode a program at a lower bit rate that allows the program to fit on a disc. Or, several programs can now fit on one disk. Digital storage on tape may also be advantageous as described above.

As described above, the present invention provides an advantageous effect for image compression or the like to reduce the bandwidth requirement of a digital video encoder.

Claims (15)

A method of processing a non-telecine sequential scan video signal, Adaptively filtering the detected signal using two-dimensional filtering to produce a filtered signal, Converting the filtered signal to a lower resolution to produce a lower resolution signal; MPEG encoding the lower resolution signal to produce an encoded signal, and Delivering the encoded signal to an output channel And a non-telecine sequential scanning video signal. The method of claim 1, wherein the filtering step is low-pass filtering and the encoding step is MPEG2 encoding. A method of processing a non-telecine sequential scan video signal, Filtering the detected signal using two-dimensional filtering to produce a filtered signal, Converting the filtered signal to a lower resolution to produce a lower resolution signal having a resolution of 1280 × 1080 samples per frame; Encoding the lower resolution to produce an encoded signal, and Delivering the encoded signal to an output channel And a non-telecine sequential scanning video signal. In a high definition video signal processing system, a method of processing a received digital video signal representing at least two video resolutions, comprising a resolution of 1280 × 1080 data samples per frame, Adaptively filtering the signal using two-dimensional filtering; Decoding the signal to produce a decoded signal; Determining the image resolution of the decoded signal; If the decoded signal has a horizontal image resolution of 1280 samples per line, converting the horizontal information from the decoded signal to another resolution to produce a converted signal, and Transmitting the converted signal to an output device Including a digital video signal. 5. The method of claim 4, wherein the conversion is up-conversion and the other resolution is 1920 horizontal samples per line. 5. The method of claim 4, wherein the conversion is down-conversion and the other resolution is a lower resolution. 5. The method of claim 4, wherein the received digital video signal is MPEG2-compatible. 2. The method of claim 1, wherein the adaptive filtering is independent of signal subsampling by the transforming step. 4. The method of claim 1, wherein the adaptive filtering is a function of image signal parameters prior to filtering. The method of claim 1, wherein the method comprising the adaptive filtering step to the forwarding step operates on a fixed image frame format. The method of claim 1, wherein the adaptive filtering is adaptive within a video frame. 12. The method of claim 11, wherein the adaptive filtering is performed based on one pixel. delete delete delete

Images