RU2201654C2 - Low-noise coding and decoding method - Google Patents

Abstract

FIELD: image compression to reduce requirements to digital video decoder bandwidth. SUBSTANCE: digital-image adaptive processor precedes MPEG2 coder and receives high-definition signal (1920x1080) pixels per image) designed for relaying or storage and adaptively filters off signal with respect to low frequencies. Signal is subjected to two-dimensional filtering of low frequencies to eliminate coding distortions and associated noise; then it is horizontally decimated to produce hybrid signal of lower definition (1280x1080 pixels per image) which is unpacked and decoded by receiver, whereupon hybrid-signal sampling frequency is raised to its source definition. Proposed method enables transmission of one highdefinition program with several standard-definition programs over 6- MHz ground relaying channel. EFFECT: improved image recovery for high-definition reproduction. 30 cl, 5 dwg, 1 tbl

Description

FIELD OF THE INVENTION
This invention relates to image compression to reduce the bandwidth requirements of a digital video encoder.

State of the art
The United States Federal Communications Commission (FCC) has approved the Digital High Definition Television (HDTV) standard proposed by Grand Alliance (GA), paving the way for terrestrial digital television broadcasting in the United States. GA's HDTV system uses the MPEG2 Moving Image Expert Group's international video compression and transmission standard. See "Information Technology - General Coding of Motion Picture and Interconnected Audio Information: Video" for details, ISO / IEC 13818-2: 1996 (E). Using modern and sophisticated video compression methods, such as source processing, motion estimation and compensation, conversion task, and statistical coding, the MPEG compression system can reduce the transmission bit rate by a factor of 50 or more. Prior to compression, a full high-definition signal in one second requires approximately one billion bits. As suggested in the GA specification, images of 1920 • 1080 pixels (image elements) at a frequency of 60 fields per second are compressed to 18 megabits per second for digital broadcasting.

 GA's video compression system typically contains two main subsystems, a preprocessor and an MPEG2 video encoder, followed by an output buffer. The input signal to the preprocessor is an analog video signal in the RGB format (red-green-blue, red - green - blue). The preprocessor digitizes the input signals and performs gamma correction of each color component to compensate for the non-linear characteristics that form the image of the camera. Gamma correction reduces the visibility of the quantization noise contained in the compressed image, especially in dark areas of the image. The preprocessor then linearly converts the digitized and gamma-corrected sampled RGB signals into the 240M YC1C2 Society of motion picture and television engineers color space. In conclusion, the resulting color signal components are subjected to downsampling to create a 4: 2: 0 digital video input signal. In addition to the tasks just described, the preprocessor can perform image conversion. For example, in a digital satellite broadcasting system, a video signal is horizontally thinned 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 redundancy between frames and some spatial redundancy within frames. Typically, compression is achieved by using a number of different technologies in sequential order, as described above. Regulation of quantization accuracy allows the encoder to generate a compressed bitstream of any frequency specified by a particular application. Quantization in MPEG2 systems is performed on the discrete cosine transform (DCT) coefficients of a data block, which can be source image information or difference information from motion estimation. Using quantization matrices together with scalable quantization step sizes, the quantizer selects and quantizes only a small fraction of the discrete cosine transform coefficients from each discrete cosine transform block intended for transmission, which leads to a significant reduction in the amount of data. The quantization matrices can be changed on a frame-by-frame basis according to the statistical distribution of the coefficients of the discrete cosine transform and the content of the video signal. For different zones within the frame, quantization can be finely tuned from the macroblock to the macroblock by scaling the size of the quantization step based on the complexity of the macroblock. For a given output bit frequency, the output buffer will provide control signals used by the encoder to adjust the quantization step size for a particular frame to maximize the quantization resolution within the available bandwidth.

 Ideally, a video compression system removes high-frequency components that viewers will not perceive as missing when the image is restored and reproduced. The remaining low-frequency components are quantized to fit within the available bandwidth. The quantization noise introduced into the signal should also be invisible to viewers when restoring the image. However, in a real system, the optimal ratio between the transmitted information and the quantization step size is selected for the available bandwidth. If the system does not provide a sufficient number of coefficients for quantization, the system increases the size of the quantization step, which leads to block distortion in the reconstructed image. If the image loses too much high-frequency information during the compression process, the reconstructed image will contain other noticeable distortion of the outlines.

 Moreover, differences in quantization between frames cause the frames in the group of pictures to contain different high-frequency components. An I frame, for example, may lose a significant number of high-frequency coefficients during encoding, while P and B frames retain high-frequency components corresponding to those lost in the I frame. The reconstructed group of pictures will now contain distortions, since the high-frequency information differs between the frames used to restore each other.

 These problems arise in the GA system that currently exists. Compressing a high-definition image signal only further reduces the quality of the reproduced image. Satellite broadcast service providers do not want to transmit high-definition signals, since only one program can be transmitted through a transponder at a time. Compression of the high-definition program to meet the requirements, sufficient to place two programs on one satellite channel (for example, a 24 MHz channel with four-position phase shift keying) at the same time, leads to unacceptable picture quality for the viewer . Therefore, satellite broadcast service providers are hesitant to broadcast HDTV due to inefficient use of the channel. Similarly, terrestrial broadcast service providers are reluctant to provide full high definition programs when one program fully occupies a channel that can accommodate several standard definition programs (SD).

SUMMARY OF THE INVENTION
According to the principles of the present invention, the digital image processor identifies the type of video signal and selectively converts the format of the original signal to another format as necessary. The converted signal is filtered and converted back to the original format as needed. The filtered signal is converted to a lower resolution and compressed to the intended bit frequency. In conclusion, the compressed signal is supplied to the output data channel.

Brief Description of the Drawings
In FIG. 1 shows one configuration of a video compression device according to the present invention.

 Figure 2 shows in detail the block 22 shown in figure 1.

 In FIG. 3 shows one of the possible characteristics of an adaptive filter included in block 22.

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

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

Description of a preferred embodiment of the invention
An MPEG2 encoder including a device that complies with the principles of the present invention comprises a two-dimensional (eg, vertical and horizontal) filter in front of the encoder. An encoder, an output buffer, and a filter each generate information that can be used by other units to increase overall efficiency. Such information concerns, for example, image motion, image contrast, quantization matrix selection, scaling factor selection, bit frequency at the output of each block, image texture. The exchange of this information between blocks is carried out using a controller that monitors the coding process, or using individual controllers located in each block.

 The controller evaluates the incoming information and determines uniformity by a group of pictures, a frame or part of a frame, the use of which can be useful for modifying the filter and / or encoder to effectively encode a group, frame or part of a frame to the planned bit frequency. Typically, a filter is adjustable because filter control produces less noise than encoder control. Also, a filter is actually a group of filters, which makes it possible to have the greatest flexibility in adjusting the coefficients of individual filters as required. These filters are a horizontal low-pass filter to eliminate spectral overlapping effects / a vertical low-pass filter and a two-dimensional low-pass filter, usually located in the sequence just indicated. The controller evaluates the received information in terms of tuning the current filter and the encoder and adjusts one or more filters and / or encoder in accordance with one or more dominant unifications. The end result is that the input signal is filtered by a low-pass filter by a method that basically allows the encoder to encode the image uniformly over a group of pictures, a frame, or part of a frame relative to the dominant uniformity of uniformly encoded data.

 The encoded signal can be transmitted in an available bandwidth and then restored and reproduced without distortion that would otherwise be present. For high-definition signals having an image frame size of 1920 • 1080 pixels, after filtering and before encoding, the horizontal resolution is reduced to 1280 pixels per line to further reduce the bandwidth of the transmitted signal. The result is a hybrid image resolution that high-definition receivers can receive, decode, and reproduce with minor changes in software.

 An exemplary configuration of a video compression system according to the present invention is shown in FIG. The input video signal is received by the movie detector 20, which determines whether the signal is a movie signal (film), which is reformatted by television conversion methods from 24 frames per second to 30 frames per second. The reformatted movie signal is sent to the corresponding section of the adaptive image processor 22, as will be described later. If the input signal is not a reformatted movie signal, the signal passes to another section of the adaptive image processor 22. Identification of the movie signals that have undergone a television conversion takes place using known methods.

 The processor 22 receives control information from the output buffer 26 and from the MPEG2 encoder 24 through the controller 28 and filters the image frames so that the encoder 24 can efficiently encode the frame so that the frame is within the available bit frequency and is basically free from noticeable distortion . The processor 22 filters the signal in two directions (2-D) (for example, horizontal and vertical), which is required to improve the quality of the reconstructed image obtained from the MPEG2 encoded bit stream compressed to an average bit frequency. The objective is to modify the local two-dimensional frequency content of the source to increase the efficiency of MPEG2 encoding by the method that damages the image recovered from the MPEG2 format to the least extent from the point of view of image clarity and distortion during encoding. Signal filtering can be applied to predefined data, such as a group of pictures or frames, a single frame, or pixel by pixel.

 A two-dimensional filter filters low-frequency images. In the best case scenario, high-frequency information that is deleted is either redundant or invisible to the viewer. In practice, in order to achieve the desired bit frequency, some of the high-frequency information that is visible to the viewer can be deleted. However, a system that includes a processor 22 before MPEG2 encoding generates an image that is better than a system without a processor 22, which will be discussed later.

 The filtered signal is encoded by the MPEG2 encoder 24, which receives image parameters from the processor 22 and the output buffer 26 through the controller 28 and adjusts the compression in MPEG2 format so that it matches the available bit frequency. Compression occurs in the same manner as described in the GA specification. The encoder 24 sends the compressed data to the output buffer 26. The buffer 26 provides the compressed data at a predetermined frequency for encoding for transport, modulation and transmission over the transmission channel using known signal processing techniques. Before modulation, the compressed signal can be sent to a statistical multiplexer for multiplexing with several programs for transmission on a single channel. The signal processing units after the buffer 26 are well known and therefore not shown in FIG. 1 to simplify the drawing.

 The video compression system can be configured to perceive any type of video signal. The system of FIG. 1 is configured to receive both television programs (camera) and film programs (film) formatted in well-known industry standards. For example, one general configuration would be used for the system of FIG. 1 to receive an output signal from a preprocessor, as previously discussed in the Background section. The system can be configured to accept other types of video signals by adding appropriate hardware and / or software.

 These configurations are not shown to simplify FIG. 1.

 The movie detector 20 recognizes the presence of certain ratios in the input signal that can be used to increase coding efficiency: (Type 1) An interlaced source of 60 fields per second, (Type 2) A movie image of 30 frames per second with interlaced scanning of 60 fields per second, (Type 3) Motion picture 24 frames per second interlaced 60 fields per second, (Type 4) Progressive scan source, (Type 5) Movie 30 frames per second progressive 60 frames per second, and (Type 6) Movie image 24 frames per second with progressive scan of 60 frames per second. Detection occurs in response to an external control signal (not shown) or using well-known technologies, for example, those used in modern standard definition MPEG2 encoders. Information about the signal format is supplied together with the signal itself to the adaptive image processor 22, as described below. The movie detector 20 also recognizes whether the signal is interlaced or interlaced and sends this information to processor 22. These types of scans are exemplary and specify the parameters by which signals are routed through processor 22. Alternatively, other embodiments may be used. field and frame frequencies.

 The adaptive image processor 22 performs several programmable functions that reduce the amount of data that needs to be compressed using encoder 24. The processor 22 mainly processes each frame so that the processed frame can be optimally encoded to eliminate or significantly reduce the noise that is noticeable to the viewer. The processor 22 can mainly be considered as a spatial changing two-dimensional low-pass filter, as it decimates each image frame, spatially and adaptively filters the selected two-dimensional high-frequency components from the signal. To create a processed frame, adaptive filtering can be adjusted by a series of frames, for a single frame or pixel by pixel.

 The processor 22 may facilitate encoding for any type of signal. However, for this embodiment of the present invention, the processor 22 is programmed to operate with high definition data as specified by the GA specification. It can be either 1920 • 1080 pixels per image, or 1280 • 720 pixels per image. According to the GA specification, each high definition format requires approximately 18 megabits per second for broadcast. To simplify the discussion, only the 1920 • 1080 format will be discussed in detail below. This consideration is equally applicable for the format 1280 • 720 or any other format.

 Figure 2 shows in detail the adaptive image processor 22. Depending on the information on the signal format received from the detector 20, the image signal is sent by the controller 28 (Fig. 1) to the interlaced to line-by-line converter 221 (Type 1), the inverse television conversion unit 222 (Type 2, 3) or passes unmodified (Type 4-6) into the spatial low-pass filter with a limited frequency band 223. Filter 223 receives the output signal from blocks 221 and 222 after these blocks have processed the signal.

 Converter 221 receives a signal if its format contains interlaced fields with a frequency of 60 Hz, and converts the signal into progressive frames with a frequency of 60 frames per second. A progressive frame includes all image information in each frame. Filtering a signal with progressive scan in a typical case does not introduce distortion, as it can happen when filtering information fields of an interlaced signal. Converter 221 employs known methods for converting interlaced fields to a progressive frame.

 The inverse television conversion unit 222 removes the excess fields from the interlaced film at a frequency of 60 Hz and restores the original film with a progressive scan. The line-by-line format enables subsequent vertical low-pass filtering without distortion of movement. If the input signal of the movie source (Type 2 or Type 3) was processed as a Type 1 source, vertical low-pass filtering will reduce the ability of the MPEG2 encoder to detect and properly process the material of the movie source. Coding efficiency will suffer. Block 222 converts the signal to a line-by-line format and deletes redundant fields / frames before filtering, since filtering can filter this redundant information differently. If redundant information is not deleted before filtering, the information after filtering may not be identical, and the encoder may not recognize the signal as a Type 2/3 signal. Then the encoder encodes information that would otherwise be deleted due to redundancy.

 Also, the structure of processor 22 is simplified by providing a single output clock signal from block 222. If block 222 provides output progressive motion pictures with a frequency of 24 frames per second and 30 frames per second, two output clock signals and a circuit supporting them are required.

Signals that were originally created in a 30-fps line-by-line format pass directly to filter 223. Filter 223 waits for video information presented as finished image frames. The spatial low pass filter 223 is actually a group of filters. For example, the first filter is a horizontal low-pass filter to eliminate the effects of aliasing. The second filter is a vertical low-pass filter. The latter filter is a two-dimensional low-pass filter, as described previously. The coefficients of each output of the filter can be adaptively set in accordance with the control information from the encoder 24 and the buffer 26, as can be seen in figure 1. The progressive signal is horizontally filtered at low frequencies to avoid overlapping spectra during subsequent decimation in the sampling rate converter 226. The final horizontal result from 1920 pixels per line will be 1280 pixels per line, as will be discussed later. To eliminate noise from overlapping spectra in the resulting signal, the low-pass filter 223 has a cutoff frequency of 640 cycles per line. The horizontal filter for eliminating the effects of superposition of spectra, included in block 223, can be a filter with a finite impulse response (FIR) with 17 pins with the following coefficients for pins:
[f0, f1, ..., f15, f16] = [- 4, 10, 0, -30, 48, 0, -128, 276, 680, 276, -128, 0, 48, -30, 0, 10, -4] / 1024.

Encoding high-definition video signals at a reduced bit frequency usually requires additional vertical low-pass filtering to further reduce the video signal bandwidth. The removal of vertical high-frequency energy before MPEG encoding is necessary to achieve an acceptable overall picture quality. The vertical frequency regions of the highest phase sensitivity are attenuated. The vertical cutoff frequency is set equal to a fraction of the Nyquist frequency. For example, a cutoff frequency of approximately half the line frequency of a high definition input signal may be suitable for some video material. For a high-definition signal with 1080 lines per image height, this would correspond to a cut-off of 540 lines per image height. This frequency may be programmable, and the programmable cutoff frequency will then be determined by the controller 28 from the parameters available from the encoder 24 and the buffer 26 in FIG. 1 (i.e., the desired bit frequency, quantization matrices, etc.). The vertical low-pass filter included in block 223 may be a FIR filter with 17 pins with the following coefficients for the pins:
[f0, f1, ..., f15, f16] = [- 4, -7, 14, 28, -27, -81, 37, 316, 472, 316, 37, -81, -27, 28, 14 , -7, -4] / 1024.

 Alternatively, the cutoff frequency may be equal to the double line frequency of the standard definition signal. Typically, a vertical low-pass filter follows a horizontal filter designed to eliminate the effects of aliasing.

 The processor 22 performs vertical filtering rather than vertical decimation, as a result of which a constant vertical string resolution is maintained. Currently, filtering is preferable for interlaced video signals than decimation. Converting a vertical string resolution to an interlaced image sequence requires sophisticated hardware and software, resulting in high receiver costs. Vertical conversion of the sampling frequency negatively affects the vertical high-frequency response due to the increasing complexity of the conclusions in combination with Nyquist sampling (i.e., the absence of super-sampling). Cost considerations for the receiver currently make it unpromising to reduce vertical resolution to reduce distortion and bit rate resulting from encoding. The reproduced picture would be significantly degraded if existing technology was used in the vertical sampling rate converters instead of the vertical low-pass filter described above. However, efficient and cost-effective vertical sampling rate converters can replace the vertical filter described herein without violating the principles of the present invention.

 The coefficients for both horizontal and vertical low-pass filters can be modified programmatically and, if necessary, can be applied at the pixel level to achieve the planned bit frequency without creating distortions in the reconstructed image. Usually, it is sufficient to modify the coefficients on a frame-by-frame basis. An alternative for slower processors is pre-programming a number of different sets of coefficients for filters and choosing the set that is most suitable for the processed image information. The greater flexibility of adaptive filters enables the entire system as a whole to create a data stream with less distortion compared to a system without adaptive filters.

 After the signal is filtered at low frequencies in the horizontal and vertical directions by block 223, the controller 28 determines whether the signal can be uniformly encoded by encoder 24 on a frame-by-frame basis without introducing significant quantization noise. If so, then the signal enters any of the blocks 224, 225 or 226 depending on its format, as will be discussed later. If, however, the encoding process is likely to introduce noise and / or distortion into the signal, the signal is sent to a two-dimensional low-pass filter in block 223 for further adaptive filtering. The control parameters from processor 22, encoder 24, and output buffer 26 (FIG. 1) allow the controller 28 (or the individual block controller in block 223) to determine whether further filtering is required. The control parameters used to make this determination are, for example, motion and contrast measurements, available quantization tables, coding efficiency, and the current planned bit rate.

 The two-dimensional filter in block 223 reduces the amount of high-frequency information coming from the image frame, mainly in the diagonal direction instead of the horizontal or vertical directions individually. The human eye is very sensitive to high-frequency noise vertically and horizontally in diagonal directions. Removing enough diagonal high-frequency information to enable uniform quantization by encoder 24 typically results in a higher quality signal with less observable noise. The diagonal filter, like all previous filtering, works with the entire image frame and is programmable.

 The diagonal filter may be compatible with the quantization matrices in the encoder. Quantization matrices often use rhomboid matrices to quantize I frames. However, these matrices often generate noise, since B and P frames use other types of quantization matrices that store high-frequency components during the compression and motion compensation process that occurs in encoder 24. Filters from processor 22 remove high-frequency information from each image frame before how the MPEG2 encoder 24 creates frames in the motion estimation loop during data processing. Thus, high-frequency components are mainly removed from P and B frames, as well as from I frames. During recovery, as you know, the image is mostly free from distortion created by MPEG2 encoding.

 In practice, the controller 28 of FIG. 1 evaluates the signal parameters (eg, motion, contrast, etc.) before filtering a particular frame by the processor 22 and determines the coefficient settings for all filters in block 223, including the necessary diagonal filtering. During the filtering process of a particular frame, the controller 28 monitors the signal parameters from the processor 22, the encoder 24 and the buffer 26 and changes the coefficients as necessary to maintain the planned bit frequency with minimal distortion / noise. Each frame is filtered based on the most recent signal parameters and fed to encoder 24 for compression and then to buffer 26, while subsequent information is input to filter 223.

 If the signal was created as a movie signal with a frequency of 24 frames per second, the filtered signal is supplied to block 224, designed to stretch 3: 2. Block 224 duplicates the selected frames to provide an output signal with a frequency of 30 frames per second. This occurs using known methods. Then, from block 224, the signal enters the transducer with horizontal decimation 226.

 The field downsampling unit 225 converts the interlaced signals from the filter 223 from the interlaced format to the interlaced format. This conversion is carried out by known methods. Without the inverse conversion to interlace format, the signal would contain a double amount of data, since the frequency of progressive frames coming from block 221 is 60 Hz. An interlaced signal is provided to converter 226.

 Sample converter 226 receives progressive signals at a rate of 30 frames per second directly from filter 223. In addition, as described above, blocks 224 and 225 provide signals to converter 226. Converter 226 decimates the high-definition signals to a selected transmission format. This format may not be standard. This can be any image and frame size that is required. However, the non-standard format will require modification of the receiver.

 When converter 226 receives HDTV signals of GA 1920 • 1080 standard, converter 226 decimates horizontal information and outputs a 1280 × 1080 hybrid pixel frame format. GA compliant HDTV receivers are capable of receiving image frames containing 1920 • 1080 pixels and 1280 • 720 pixels. Therefore, GA-compatible receivers can be modified to support 1280 pixels of horizontal resolution and 1080 pixels of vertical resolution. The hardware of a compatible receiver increases the sampling rate from 1,280 horizontal pixels to 1,920 horizontal pixels, while increasing vertical resolution. However, GA-compliant receivers are not required and are not programmed to receive image frames with a resolution of 1280 • 1080 pixels (horizontal resolution for vertical resolution) as the specified format. Hardware for receiving and decoding this resolution is already available, but software must be added to decode and increase only horizontal resolution. Adding software is simpler and cheaper than changing the design and adding new hardware needed for other custom formats.

 The processor 22 provides a hybrid format of 1280 • 1080, since modern playback technology is not able to reproduce a resolution of 1920 pixels per line. Currently, the best television monitors can only reproduce a resolution of approximately 1200-1300 pixels per line. Therefore, limiting the output resolution to 1280 pixels per line in the horizontal direction has a small adverse effect on picture quality, if any. Providing resolution for playback of 1280 • 1080, which is supported by existing receiver hardware for decoding and decompression, will give receiver manufacturers a minimum of problems, since only a software change is necessary. For certain receivers, such as broadcast satellites, a software modification can be downloaded and installed remotely via a satellite link. For these receivers there is no need to engage service personnel.

 The hybrid format has advantages, as terrestrial and satellite program providers are reluctant to transmit high-definition programs. A satellite transponder transmits a bitstream at a frequency of approximately 24 Mbps (Mbps). Terrestrial HDTV broadcasts can be transmitted at a speed of up to 19 Mbit / s, including a high-definition program with 18 Mbit / s and other information (for example, an audio channel, program guide, access mechanisms, etc.). Each of the modern satellite transponders can carry at most one HDTV program, which satellite program providers consider insufficiently beneficial. Simply reducing the horizontal resolution of the frame from 1920 to 1280 is not enough to make it possible to simultaneously transmit two high-definition programs through one satellite transponder. The filtering provided by the processor 22, successfully allows such a dual transmission of high definition on a single channel.

 The filtering characteristic provided by the processor 22 may take various forms, including a rhombus, a cross, and a hyperbola along the coordinate axes, where for each filter, the filtering is diagonal. One possible form, a two-dimensional hyperbole, has particular advantages for this application and has an amplitude-frequency response, which is shown in FIG. 3. The cutoff frequency of the adjustable filter is mainly set to allow uniform compression of the selected group of pictures, frame, or part of the frame by encoder 24. If necessary, additional horizontal and vertical high-frequency information can be filtered, but this is usually not required. As the complexity of the picture changes or the available bit rate increases, the amount of data filtered by the diagonal filter and other previous filters decreases. A two-dimensional filter can be described, for example, as a two-dimensional FIR filter with 13 pins in each direction (13 • 13) or as a two-dimensional filter with an infinite impulse response (IIR filter).

 The two-dimensional FIR filter included in block 223 may be a filter having 13 • 13 conclusions, with coefficients for the conclusions given in the table (see the end of the description).

 For these coefficients, the DC down-conversion factor is 1024. The coefficients exhibit octant symmetry, which gives 28 independent coefficients. Symmetric areas of the coefficients enable faster adjustment of the adjustable filter. It is possible, however, that each octant is different if, for example, the filtered image or region exhibits a different characteristic in one part of the image.

 The filter characteristic of the processor 22 may vary continuously from one set of coefficients to another on a per-pixel basis. Thus, the processor 22 can set various operating parameters to maintain good image quality while limiting the bit frequency, as will be discussed later.

 As mentioned previously, the processor 22 may be adaptively modified for adaptive filtering depending on the parameter (s) used to specify the adaptation of the filter. For example, variation in the image frame can be used to segment the image into regions for different processing. Contours are an important feature of the image, since the dominant contours mask coding errors in close proximity to them, and they can also define areas of the image. Colorimetry can be used to identify areas of low complexity, such as the body or the sky. Textures can also be identified and processed as an area. Textures are usually less important than outlines. Therefore, textures identify areas that can be more filtered out than other areas. Also, to capture important images or actions that require coding with higher efficiency and, therefore, less filtering, a kinematic composition can be used. The background is mainly softened by the depth of field of the camera optics and can be filtered more strongly. The panning and viewing information can be used / to determine the center of interest in the image for differentiated processing by the processor 22.

 The operation of encoder 24 is compatible with the MPEG2 standard. Encoder 24 may provide information through controller 28 that processor 22 may use to improve performance. Such information may include information on a bit frequency, for example. This bit rate information may comprise an average bit rate for a group of pictures, a frame bit rate, and a bit frequency of a macroblock or block. Other information that can improve the performance of processor 22 includes the complexity of the discrete cosine transform, the type of quantization matrix used, and the used size of the quantization matrix step. The processor 22 can also provide information through the controller 28 to the encoder 24 to regulate its operation in order to improve coding performance.

 After forming into the transport packet data stream using known technologies, the high-definition signal is transmitted in a known manner to the receiver, for example, as described in the specification of the Grand Alliance. With the exception of the required increase in sampling rate to full high definition pixel resolution at the receiver, the signal processing provided by processor 22 is transparent to the decoder in the receiver compatible with the Grand Alliance standard.

 At the receiver, the data stream is demodulated and the transport stream is processed to extract data packets and program information using well-known technologies. For high-definition programs in the hybrid format described above, the sample rate of the signal in the playback processor is increased in the horizontal direction if playback requires a full high-definition signal. The number of vertical lines in the image signal remains unchanged. This restores the full high-definition signal with a resolution of 1920 • 1080 for the device to reproduce the high-definition image. If the image reproducing apparatus does not require a full high definition signal, the signal is suitably punctured using known methods during image restoration before reproduction. Existing receivers that are compliant with the Grand Alliance standard require software modifications so that they can reconstruct the hybrid signal. Software modification allows you to independently select the algorithms for horizontal and vertical hardware and software processing, highlighted by the standardized mode of the Grand Alliance standard, as required for the incoming signal.

 Figure 4 is a block diagram of the passage of a video signal image through an encoding system. At step 30, a signal format is identified, and identification information is supplied along with the video signal. The format information may indicate, for example, whether the signal originally comes from a 24-fps movie and whether it is interlaced or interlaced. If the video signal is interlaced, at step 31 it is converted to a progressive signal at a frequency of 60 frames per second. If the signal is converted by stretching 3: 2, at step 32, the excess frames are deleted. If the video signal already has a progressive format of the video camera, it goes directly to the filter of step 34. At step 34, the video signal is spatially filtered at low frequencies. This includes vertical, horizontal and diagonal filtering, as described above. At step 35, the progressive conversion of the interlaced signal, which was formed from the interlaced signal at step 31, back to the interlaced signal, occurs. At step 36, the signal is stretched 3: 2 to return redundant frames previously deleted at step 32. The video signals received from step 30 directly to step 34, as described above, now flow from step 34 directly to step 38. At step 38, the video signal from one of steps 34, 35, and 36, it is sub-sampled to the hybrid resolution of the high-definition signal 1280 • 1080 defined above, or to another output format, as necessary. At step 40, the hybrid signal is encoded in MPEG2 format, as described previously. Step 42 processes the encoded signal for transportation, step 44 modulates the transport signal as required for transmission over an output channel, such as a high frequency terrestrial broadcast channel. Finally, step 46 transmits a modulated signal. Steps 42 to 46 are carried out by known methods.

 FIG. 5 is a block diagram of the passage of a transmitted image signal through a receiver. This flowchart assumes that the image resolution of the received signal is a 1280 x 1080 high definition hybrid signal as defined above. At step 50, the transmitted signal is received by the tuner and demodulated. The demodulated signal is decoded and decompressed from the MPEG2 format in step 52. Step 54 identifies the resolution of the video signal as 1280 x 1080 pixels per image using the control information sent along with the signal. The Grand Alliance MPEG2 protocol provides image resolution information along with the transmitted signal. A hybrid signal is usually identified by a unique code, like any other specified resolution. The hybrid signal may also be described in another way, for example, using information in user data contained in the transmitted data. At step 56, during playback processing, the sampling frequency of the hybrid video signal in the horizontal direction is increased to the full high-definition signal 1920 • 1080. The increase in the sampling frequency of the hybrid signal occurs using new software tools in conjunction with existing hardware and software available in the receiver, as described previously. Finally, the full high-definition video signal is displayed on a display with a resolution of 1920 • 1080 in step 58. Steps 50, 52 and 58 use known methods.

 The device and methods described above can be applied in a number of configurations in order to achieve improved image recovery for high fidelity playback. Adaptive and non-adaptive options may be used depending on the requirements of a particular system. Some of these options are discussed below.

 A non-adaptive strategy would be to filter the frames by the processor 22 to the planned bit rate and the resolution to process all the images uniformly. Another non-adaptive strategy should be based on the assumption that the center of the reproduced image is the most interesting area. It also suggests that the periphery of the image is less interesting and therefore less important to the viewer. The filter coefficients of the processor 22 are set by the controller 28 using parameters that are functions of the spatial position of the pixel, and all image information is processed uniformly.

 An adaptive option is to segment the image into regions using texture model parameters, local video variance, color measurements, or other measurements of image complexity based on the original image. The filtering characteristics of the processor 22 are adaptively modified for various fields.

 Another approach is the adaptive modification of the filtering characteristics of the processor 22 as a function of the difference between the real bit frequency and the planned bit frequency. In this case, one parameter controls the change of filter coefficients for a two-dimensional frequency conversion.

 Another strategy is to develop such a two-dimensional filtering frequency response provided by the processor 22, which is compatible with the quantization matrix used by the encoder 24. The quantization matrix can be considered as a low-pass filter that has a two-dimensional shape. For this strategy, the values of the filter coefficients should be a function of the step size of the quantization matrix. Since the step size changes in accordance with the known mode of operation of the encoder, a corresponding change would occur for the corresponding filter coefficients.

 The above options illustrate the flexibility of a system using the principles of the present invention. Such a system preferably operates in the context of controlling the frequency of the MPEG2 standard to expand the capabilities of MPEG2 compression by reducing coding distortion and other noise. The versatility and cost-effectiveness of introducing HDTV using the present invention is enhanced. The number of high-definition programs transmitted by one transponder in a satellite broadcast system (i.e., 24 MHz four-position phase shift keying) is increased from one to two, or one high-definition program with several standard definition programs. In accordance with the principles of the present invention, it is feasible to transmit one high definition program with several standard definition programs over a 6 MHz terrestrial broadcast channel. Previously, broadcast stations were limited to transmitting one high definition program on one channel or several standard definition programs on one channel.

 Although the present invention has been described in the context of systems transmitting and receiving a high definition signal, its principles apply to other equipment, such as data storage systems. In systems such as digital video disc (DVD), video data is encoded and stored for playback at a later time. The media has a limited amount of storage space. If the encoded program, movie or other video sequence exceeds the amount of available space on the medium, further encoding / compression in order to fit the program can create unacceptable distortions. The invention described above can be used to efficiently encode a program to a lower bit frequency, allowing you to place the program on disk. Or now several programs can fit on one drive. Digital film storage can also provide benefits as described above.

Claims (30)

 1. A method of processing the first and second video signals, respectively, showing the first and second dissimilar image formats, comprising detecting the presence of the first or second video signal, according to which A. upon detection of the first video signal (a) convert the first video signal to another format to create a converted signal, (b ) filter the transformed signal to create a filtered signal, (c) re-convert the filtered signal to the original format of the first signal to re-convert signal, (d) convert the reconverted signal to a lower resolution to create a lower resolution signal, (e) encode a lower resolution signal to create an encoded signal, and (f) supply an encoded signal to the output channel, and B. detecting the second video signal (g) filter the second video signal to create a filtered signal, (h) convert the filtered signal to a lower resolution to create a lower resolution signal, (i) encode the lower resolution signal to create an encoded signal, (j) apply the encoded signal to the output channel.  2. The method of claim 1, wherein the first video signal is an interlaced signal and said interlaced signal is converted to an interlaced signal in step (a).  3. The method according to claim 1, wherein the first video signal is a signal of a telecinema conversion of a film and said signal of a television transmission of a film is converted into a signal of a reverse film conversion of a film in step (a).  4. The method of claim 1, wherein the second video signal is a progressive scan signal.  5. The method according to p. 1, in which the filtering stages provide filtering of low frequencies.  6. The method according to p. 5, in which at the stages of filtering provide two-dimensional filtering.  7. The method of claim 1, wherein the filtering steps provide adaptive filtering adjusted for frames of a group of pictures, or a single frame, or part of a frame.  8. The method according to p. 1, in which the filtering stages provide temporary low-pass filtering and adaptively change the filtering characteristics in response to the characteristics of the signal.  9. The method of claim 1, wherein the filtering steps provide spatial low-pass filtering and adaptively change the filtering characteristics in response to the characteristics of the signal.  10. The method according to claim 1, in which the encoding steps are compatible according to the international standard for video compression and transmission by an expert group in the field of MPEG2 moving image.  11. The method of claim 1, wherein the lower resolution signal has a resolution of 1280 x 1080 discrete data elements per frame.  12. The method according to claim 1, in which the first and second video signals are high-definition signals having a resolution of 1920 • 1080 discrete data elements per frame.  13. A method of processing one of the following: an interlaced video signal and a signal in the format of a tele-converted film, according to which the presence of one of the following is detected: an interlaced video signal and a signal in the format of a tele-converted film, convert the detected signal to one of the following: progressive scan and a signal in the format of the film that underwent reverse television film conversion, respectively, to create a converted signal, to the filter the converted signal is used to create a filtered signal, the filtered signal is converted again into one of the following: an interlaced signal and a signal in the format of a tele-film conversion film, respectively, to create a converted signal, the converted signal is converted to a lower resolution to create a signal with a higher low resolution, encode a signal with a lower resolution to create an encoded signal and apply the encoded signal to the output channel.  14. The method of claim 13, wherein the filtering step is low-pass filtering and the encoding step is encoding according to the international standard for compressing and transmitting a video signal by an expert group in the field of MPEG2 moving image.  15. The method of claim 13, wherein the lower resolution signal has a resolution of 1280 x 1080 discrete data elements per frame.  16. A method of processing a non-telecast video signal with progressive scan, according to which the detected signal is adaptively filtered to create a filtered signal, the filtered signal is converted to a lower resolution to create a lower resolution signal, a lower resolution signal is encoded according to the international compression and transmission standard a video signal by an expert group in the field of MPEG2 moving image to create an encoded signal and apply the encoded signal to the one channel.  17. The method of claim 16, wherein the filtering step is low pass filtering and the encoding step is encoding according to the international MPEG2 standard.  18. A method of processing a non-telecast video signal with progressive scan, according to which the detected signal is filtered to create a filtered signal, the filtered signal is converted to a lower resolution to create a lower resolution signal having a resolution of 1280 • 1080 discrete elements per frame, the signal is encoded lower resolution to create an encoded signal and apply the encoded signal to the output channel.  19. A method of processing a received digital video signal in a high-definition video signal processing system, which can have more than one image resolution, including a resolution of 1280 • 1080 discrete data elements per frame, according to which a signal is decoded to create a decoded signal, the image resolution of the decoded signal is determined, converted horizontal information from the decoded signal to a different resolution if the decoded signal has a horizontal image resolution of 1280 discrete ENTOV on line to create a transformed signal and the converted signal is supplied to an output device.  20. The method of claim 19, wherein the transform is a boost transform and the other resolution is 1920 horizontal discrete elements per row.  21. The method of claim 19, wherein the transform is a down transform and the other resolution is a lower resolution.  22. The method according to p. 19, in which the received digital video signal is compatible according to the international standard for compression and transmission of video signal by the expert group in the field of moving image MPEG2.  23. The method according to p. 16, in which adaptive filtering is independent of the downsampling of the signal, carried out at the stage of conversion.  24. The method according to p. 16, in which adaptive filtering is a function of the parameters of the image signal before filtering.  25. The method according to p. 16, which, starting from the stage of adaptive filtering according to the filing stage, provides for processing with respect to a fixed format of the image frame.  26. The method of claim 16, wherein the adaptive filtering is adaptive within an image frame.  27. The method according to p. 26, in which adaptive filtering is performed on a per-pixel basis.  28. A method of supplying video information using a signal format, wherein said format is set by 1280 picture elements to 1080 picture elements.  29. The method of claim 28, wherein said 1280 picture elements represent horizontal information and said 1080 picture elements represent vertical information.  30. The method of claim 29, wherein the video information is broadcast satellite information.

Images