GB/T 14857-1993 Specification for studio digital television coding parameters
Some standard content:
National Standard of the People's Republic of China
GB/T 14857—1993
Specifications of encoding parameters of digital television for studio
Issued on 1993-12-30
Implemented on 1994-09-01
Issued by the State Administration of Technical Supervision
National Standard of the People's Republic of China
Specifications of encoding parameters of digital television for studio
The specifications of encoding parameters of digital television for studioGB/T14857—1993
This standard is equivalent to Recommendation 601-3 (1992 edition) of the Radio Consultative Committee (CCIR) of the International Telecommunication Union (ITU). 1 Subject content and scope of application
This standard specifies the digital encoding method and parameters of 625-line/50-field studio color television component signals (Y, RY, BY signals or R, G, B signals).
This standard is applicable to 625 lines/50 fields digital TV studios and can be used as a technical basis for the design, production and maintenance of digital color TV systems and their equipment.
2 Reference standards
GB3174 Color TV broadcasting
2 Terminology
3.1 Coded signals
Color TV component signals for digital coding3.2 Sampling frequency samplingfrequency The frequency at which the instantaneous value of the video signal is obtained.
3.3 Number of samples per total line The number of samples in one line cycle.
3.4 Sampling structure samplingstructure The position of the sampling point in one frame of the image.
3.5 Form of coding
The digital coding method adopted.
3.6 Digital active line digital active line The positive period of the digital line.
4 Technical Parameters
This specification contains two sets of parameters, which can be easily converted to each other. The basic parameters are applicable to digital signal connection between digital studio equipment and international program exchange. The sampling frequency ratio of brightness and color difference signals should be 4:2:2 mode. The other set of parameters is applicable to digital TV signal source equipment and high-quality video processing. The sampling frequency ratio of brightness and color difference signals (or R, G, B) is 4:4:4 mode. The encoding parameters of 4:2:2 mode are shown in Table 1. Approved by the State Bureau of Technical Supervision on December 30, 1993 and implemented on September 1, 1994
1. Coding signal
Y, CR, CB
2. Number of samples for the entire line
Brightness signal (Y)
Each color difference signal (CR, CB)
3. Sampling structure
4. Sampling frequency
Brightness signal
Each color difference signal
5. Coding method
6. Number of samples for each digital effective line
Brightness signal
Each color difference signal
7. Time relationship between analog signal and digital signal line: from the end of the digital effective line to 0m
8. Corresponding quantization level range between video signal level and quantization level||tt ||Brightness signal
Each color difference signal
9. Code word usage
GB/T14857—1993
Formed by BY, -BY, B—E pre-corrected by y (see Appendix A2) 864
Orthogonal structure, that is, the sampling points are repeated by line, field and frame, and the C and C sampling points in each line are at the same position as the odd (1, 3, 5) sampling points of Y
The sampling frequency tolerance should be consistent with Article 2.5 (line frequency tolerance) of GB3174. The brightness signal and each color difference signal use linear quantized PCM, and each sample value is quantized by 8 (optionally 10) bits
12 brightness clock cycles
A total of 220 quantization levels, the black level corresponds to the quantization level 16, and the peak white level corresponds to the quantization level 235. The signal level may sometimes exceed quantization level 235, occupying 225 quantization levels in the middle part of the quantization level range. The codewords corresponding to quantization levels 1280 and 255 with zero signal level are dedicated to synchronization. The coding parameters for quantization levels 1 to 254 for video signals in 4:4:4 mode are shown in Table 2. Table 2
1. Coding signal
Y, CR, CB or R, G, B
2. Number of whole line samples of each signal
3. Sampling structure
4. Sampling frequency of each signal
5. Coding method
6. Length of digital effective line represented by number of samples 7. Correspondence between video signal level and 8 most significant bits (MSB) of quantization level
Quantization level range
R, G, B or brightness signal Y
Each color difference signal
Forms 864
orthogonal structure by pre-correction BY, BR-BY, BB-BY or BR, Ea, B, that is, the sampling points are repeated by line, field and frame, and the sampling point structures of the three signals overlap with each other, which is also consistent with the brightness of 4·212 mode The sampling points of the luminance signal coincide with 13.5MHz
Linear quantized PCM,
Each sample value is quantized by 8 (optionally 10) bits 720
Total 220 quantization levels, black level corresponds to quantization level 16, peak white level corresponds to quantization level 235, signal level may sometimes exceed 235
Occupy 225 quantization levels in the middle of the quantization level range, zero signal corresponds to quantization level 128GB/T14857-1993
Appendix A
Definition of signals used in digital coding
(Supplement)
A1 Relationship between digital effective line and analog synchronization reference The relationship between the 720 sample values of the digital effective line luminance signal and the 625-line system analog synchronization reference is shown in Table A1 and Figure A1. The sampling of the luminance signal is co-located with the analog line reference point O. Table A1
Line Sync Front Half Width
Value Reference
Digital Active Line Period
The number of color difference signal samples corresponding to the luminance sampling clock period (rated value is 74ns) can be obtained by dividing the number of luminance signal samples by 2. 12T
Definition of the digital signals Y, Ca, C obtained from the original (analog) signals Er, E, and E2
The Y, C, C signals are constructed according to A2.1, A2.2, and A2.3. An example is given in A2.4. The digital signals Y, Ca, and C are defined and controlled by the control signal.
Figure A1 Relationship between image signal sampling and analog line synchronization On next line
GB/T14857—1993
A2.1 Structure of brightness signal (E'), color difference signal (Er-EY) and (EEY) The brightness signal and color difference signal are constructed according to the following formula: Ex-0.299ER+0.587E+0.114E
(Er-E)=0.701Er-0.587E—0.114E(EE)=—0.299Er-0.587Ec+0.886EB After normalizing the signal value (taking 1.0V as the highest level), the black, white and saturated primary colors and their complementary color values are listed in Table A2. Table A2
A2.2 Structure of the renormalized color difference signal (Ec and Ec)
It can be seen from Table A2 that the value range of Ey is 1.0~0, while the value ranges of the color difference signals EE and E-Ey are +0.701~0.701 and +0.886~0.886 respectively. In order to restore the offset value of the color difference signal to 1 (that is, +0.5~-0.5), the coefficient: 0.5
is introduced to obtain: Ec-0.713(ER-E)-0.500ER—0.419E—-0.081Eg
Ec=-0.564(EE)=
-0.169ER-0.331EG+0.500EB
Ec and Ec are the renormalized red and blue difference signals respectively, that is, they have the same peak-to-peak value as Ey. When the component signal is not rated to the range of 1 to 0, for example, when converting analog component signals with different brightness and color difference signal amplitudes, additional gain factors are required, and the corresponding gain factors K and K should be changed.
A2.3 Quantization
In the case of linear quantization 8-bit binary coding, 28 or 256 equally spaced quantization levels are specified, so the range of binary numbers is from 00000000 to 11111111, and its equivalent decimal number is from 0 to 255. In order to avoid confusion between 8-bit and 10-bit representations, the 8 most significant bits are regarded as the integer part and the other two bits are regarded as the fractional part. For example, the bit form of 10010001 is represented as 145a or 91, while 1001000101 is represented as 145.25a or 91.4. When the fractional part is not shown, it should be regarded as having a binary value of 00. In 4:2:2 mode, levels 0 and 255 are used for synchronization data, and levels 1 to 254 are used for video signals. It is known that the brightness signal occupies 220 levels, the black level is at quantization level 16, and the decimal number of the brightness signal quantization is: =219(Ey)+16
Similarly, it is known that the color difference signal occupies 225 levels in the middle of the quantization level range, and the zero level is equivalent to level 128. Therefore, the decimal numbers CR and C of the color difference signal quantization are:
CR=224[0.713(ER—EY)]+128
C=224[0.564(E—E)+128
Get: CR=160(ER—Ey）+128
CB=126(E-EY）+128
GB/T14857—1993
After quantization, the nearest integer values of Y, CR, and CB are taken, which are expressed as Y, CR, and CB respectively. A2.4 Constructing Y, CR, CB by quantizing Er, EG, E In the case where Y, C, CB are directly generated from the Y pre-corrected component signals E, E, E or directly generated in digital form, the quantization and encoding are equivalent to:
Er (digital form) int (219Er) + 16E (digital form) = int (219E.) + 16E (digital form) = int (219EB) + 16 So:
256BR+
The above three formulas are all taken to the nearest integer value. 21E+128
En'+128
In order to obtain the 4:2:2 component signals Y, CR, CB, it is necessary to perform low-pass filtering and sub-sampling on the above 4:4:4 CR, CB signals. It should be pointed out that the CR, C component signals obtained in this way may be slightly different from the CR, C component signals obtained directly by analog filtering. A2.5 Limitation of Y, Ce, C signals
Digital encoding in the form of Y, CR, CB may provide signal values larger than the signal range supported by the corresponding R, G, B. Therefore, when the Y, CR, CB signals are generated by electronic image generation or signal processing, although each value is accurate, when converted to R, G, B, it may exceed these signal value ranges. To prevent this from happening, a more convenient and effective method is to limit the Y, C, C signal values. It is also possible to minimize subjective damage by maintaining brightness and hue values and only sacrificing saturation. 5
GB/T148571993
Appendix B
Filter characteristics
(Supplement)
a Insertion loss/frequency characteristic template
b·Ripple tolerance within the passband
c Group delay tolerance within the passband
New rate: MHa!
Figure B1 is used for the characteristics of the brightness or R, G, B signal filter when the sampling frequency is 13.5MHz0
GB/T14857—1993
a Insertion loss/frequency characteristic template
b Ripple tolerance within the passband
c Group delay tolerance within the passband
Ss2024m
Figure B2 is used for the characteristics of the color difference signal filter when the sampling frequency is 6.75MHz Note: The lowest value marked in b and c is at 1kHz (not at 0MHz). dB
GB/T14857—1993
a Insertion loss/frequency characteristic template
Rate hiIri
b Fluctuation tolerance within passband
n.2, 6.ns
Figure B3 Characteristics of digital filter when the sampling rate of color difference signal is changed from 4:4:4 to 4:2:2 Note: ① The characteristic values of fluctuation and group delay are relative to their values at 1kHz. The solid line represents the actual limit, and the dotted line gives the limit recommended by theoretical design. ② In the digital filter, the actual limit is the same as the designed limit. According to the design, the delay distortion is 0. ③ In the digital filter (Figure B3), the amplitude-frequency characteristic (on the linear scale) should be skew-symmetrical at the half-amplitude point (as shown in the figure). ④ In the recommendation of the filter used in the encoding and decoding process, it is envisaged that the sinx/a characteristic of the sampling and holding circuit is corrected in the post-filter after the digital-to-analog conversion.
Appendix C
Some guidance on the implementation of the filters recommended in Annex B (reference)
The recommendations for filters used in the coding and decoding process have envisaged that correction for the sina/a characteristic is provided in the post-filter after the digital-to-analogue conversion. The combined bandpass tolerance of the filter plus the sina/a corrector plus the theoretical sina/α characteristic should be the same as the tolerance given to the filter alone. 8
GB/T14857-1993
This requirement is easily met if the filter, (sn/3) corrector and delay equalizer are considered as a single unit during the design process.
The total delay caused by filtering and encoding the luminance and colour difference components should be the same. The delay in the colour difference filter (Annex B, Figure B2) is twice the delay in the luminance filter (Annex B, Figure B1). Because of the difficulty of using analog delay networks and equalizing these delays within the passband tolerance, it is recommended that the majority of the delay differences (integer multiples of the sampling period) should be equalized in the digital domain. In correcting the remainder, it should be noted that the sample and hold circuits in the decoder introduce a flat delay of half a sampling period. The passband tolerances for amplitude fluctuation and group delay are very tight. Current research has shown that it is necessary to ensure that a large number of codecs can be cascaded without compromising the high quality of the 4:2:2 coding standard. Due to the performance limitations of currently available test equipment, it may be economically difficult for the factory to meet the tolerances of each filter in production. However, it is still possible to design filters that meet the specified characteristics. The factory is required to make efforts in production to adjust each filter to the specified characteristic template. The specifications given in Appendix B are intended to maintain the spectral content of the Y, Cr, and Cs signals as much as possible throughout the component signal chain. However, it is obvious that the spectral characteristics of the color difference signals in the picture monitor or at the end of the component signal chain will inevitably be affected by the slow roll-off filters inserted. Additional Notes:
This standard was proposed by the Ministry of Radio, Film and Television. The Standardization Planning Institute of the Ministry of Radio, Film and Television is responsible for the technical coordination of this standard. The Science and Technology Department of the Ministry of Radio, Film and Television is responsible for drafting this standard. The drafter of this standard is Kang Songshi.4. Y, CR, CB are formed by quantization of Er, EG, E. When Y, C, CB are directly generated by the component signals E, E, E of Y pre-correction or directly generated in digital form, the quantization and encoding are equivalent to:
Er (digital form) int (219Er) + 16E (digital form) = int (219E.) + 16E (digital form) = int (219EB) + 16. The above three formulas are all taken to the nearest integer value. 21E+128
En'+128
To obtain the component signals Y, CR, CB of 4:2:2, it is necessary to perform low-pass filtering and sub-sampling on the above CR, CB signals of 4:4:4. It should be pointed out that the CR, C component signals obtained by this method may be slightly different from the CR, C component signals obtained directly by analog filtering. A2.5 Limitation of Y, Ce, C signals
Digital encoding in the form of Y, CR, CB may provide signal values larger than the signal range supported by the corresponding R, G, B. Therefore, when the Y, CR, CB signals are generated by electronic image generation or signal processing, although each value is accurate, when converted to R, G, B, it may exceed these signal value ranges. To prevent this from happening, a more convenient and effective method is to limit the Y, C, C signal values. It is also possible to minimize subjective damage by maintaining brightness and hue values and only sacrificing saturation. 5
GB/T148571993
Appendix B
Filter characteristics
(Supplement)
a Insertion loss/frequency characteristic template
b·Ripple tolerance within the passband
c Group delay tolerance within the passband
New rate: MHa!
Figure B1 is used for the characteristics of the brightness or R, G, B signal filter when the sampling frequency is 13.5MHz0
GB/T14857—1993
a Insertion loss/frequency characteristic template
b Ripple tolerance within the passband
c Group delay tolerance within the passband
Ss2024mbZxz.net
Figure B2 is used for the characteristics of the color difference signal filter when the sampling frequency is 6.75MHz Note: The lowest value marked in b and c is at 1kHz (not at 0MHz). dB
GB/T14857—1993
a Insertion loss/frequency characteristic template
Rate hiIri
b Fluctuation tolerance within passband
n.2, 6.ns
Figure B3 Characteristics of digital filter when the sampling rate of color difference signal is changed from 4:4:4 to 4:2:2 Note: ① The characteristic values of fluctuation and group delay are relative to their values at 1kHz. The solid line represents the actual limit, and the dotted line gives the limit recommended by theoretical design. ② In the digital filter, the actual limit is the same as the designed limit. According to the design, the delay distortion is 0. ③ In the digital filter (Figure B3), the amplitude-frequency characteristic (on the linear scale) should be skew-symmetrical at the half-amplitude point (as shown in the figure). ④ In the recommendation of the filter used in the encoding and decoding process, it is envisaged that the sinx/a characteristic of the sampling and holding circuit is corrected in the post-filter after the digital-to-analog conversion.
Appendix C
Some guidance on the implementation of the filters recommended in Annex B (reference)
The recommendations for filters used in the coding and decoding process have envisaged that correction for the sina/a characteristic is provided in the post-filter after the digital-to-analogue conversion. The combined bandpass tolerance of the filter plus the sina/a corrector plus the theoretical sina/α characteristic should be the same as the tolerance given to the filter alone. 8
GB/T14857-1993
This requirement is easily met if the filter, (sn/3) corrector and delay equalizer are considered as a single unit during the design process.
The total delay caused by filtering and encoding the luminance and colour difference components should be the same. The delay in the colour difference filter (Annex B, Figure B2) is twice the delay in the luminance filter (Annex B, Figure B1). Because of the difficulty of using analog delay networks and equalizing these delays within the passband tolerance, it is recommended that the majority of the delay differences (integer multiples of the sampling period) should be equalized in the digital domain. In correcting the remainder, it should be noted that the sample and hold circuits in the decoder introduce a flat delay of half a sampling period. The passband tolerances for amplitude fluctuation and group delay are very tight. Current research has shown that it is necessary to ensure that a large number of codecs can be cascaded without compromising the high quality of the 4:2:2 coding standard. Due to the performance limitations of currently available test equipment, it may be economically difficult for the factory to meet the tolerances of each filter in production. However, it is still possible to design filters that meet the specified characteristics. The factory is required to make efforts in production to adjust each filter to the specified characteristic template. The specifications given in Appendix B are intended to maintain the spectral content of the Y, Cr, and Cs signals as much as possible throughout the component signal chain. However, it is obvious that the spectral characteristics of the color difference signals in the picture monitor or at the end of the component signal chain will inevitably be affected by the slow roll-off filters inserted. Additional Notes:
This standard was proposed by the Ministry of Radio, Film and Television. The Standardization Planning Institute of the Ministry of Radio, Film and Television is responsible for the technical coordination of this standard. The Science and Technology Department of the Ministry of Radio, Film and Television is responsible for drafting this standard. The drafter of this standard is Kang Songshi.4. Y, CR, CB are formed by quantization of Er, EG, E. When Y, C, CB are directly generated by the component signals E, E, E of Y pre-correction or directly generated in digital form, the quantization and encoding are equivalent to:
Er (digital form) int (219Er) + 16E (digital form) = int (219E.) + 16E (digital form) = int (219EB) + 16. The above three formulas are all taken to the nearest integer value. 21E+128
En'+128
To obtain the component signals Y, CR, CB of 4:2:2, it is necessary to perform low-pass filtering and sub-sampling on the above CR, CB signals of 4:4:4. It should be pointed out that the CR, C component signals obtained by this method may be slightly different from the CR, C component signals obtained directly by analog filtering. A2.5 Limitation of Y, Ce, C signals
Digital encoding in the form of Y, CR, CB may provide signal values larger than the signal range supported by the corresponding R, G, B. Therefore, when the Y, CR, CB signals are generated by electronic image generation or signal processing, although each value is accurate, when converted to R, G, B, it may exceed these signal value ranges. To prevent this from happening, a more convenient and effective method is to limit the Y, C, C signal values. It is also possible to minimize subjective damage by maintaining brightness and hue values and only sacrificing saturation. 5
GB/T148571993
Appendix B
Filter characteristics
(Supplement)
a Insertion loss/frequency characteristic template
b·Ripple tolerance within the passband
c Group delay tolerance within the passband
New rate: MHa!
Figure B1 is used for the characteristics of the brightness or R, G, B signal filter when the sampling frequency is 13.5MHz0
GB/T14857—1993
a Insertion loss/frequency characteristic template
b Ripple tolerance within the passband
c Group delay tolerance within the passband
Ss2024m
Figure B2 is used for the characteristics of the color difference signal filter when the sampling frequency is 6.75MHz Note: The lowest value marked in b and c is at 1kHz (not at 0MHz). dB
GB/T14857—1993
a Insertion loss/frequency characteristic template
Rate hiIri
b Fluctuation tolerance within passband
n.2, 6.ns
Figure B3 Characteristics of digital filter when the sampling rate of color difference signal is changed from 4:4:4 to 4:2:2 Note: ① The characteristic values of fluctuation and group delay are relative to their values at 1kHz. The solid line represents the actual limit, and the dotted line gives the limit recommended by theoretical design. ② In the digital filter, the actual limit is the same as the designed limit. According to the design, the delay distortion is 0. ③ In the digital filter (Figure B3), the amplitude-frequency characteristic (on the linear scale) should be skew-symmetrical at the half-amplitude point (as shown in the figure). ④ In the recommendation of the filter used in the encoding and decoding process, it is envisaged that the sinx/a characteristic of the sampling and holding circuit is corrected in the post-filter after the digital-to-analog conversion.
Appendix C
Some guidance on the implementation of the filters recommended in Annex B (reference)
The recommendations for filters used in the coding and decoding process have envisaged that correction for the sina/a characteristic is provided in the post-filter after the digital-to-analogue conversion. The combined bandpass tolerance of the filter plus the sina/a corrector plus the theoretical sina/α characteristic should be the same as the tolerance given to the filter alone. 8
GB/T14857-1993
This requirement is easily met if the filter, (sn/3) corrector and delay equalizer are considered as a single unit during the design process.
The total delay caused by filtering and encoding the luminance and colour difference components should be the same. The delay in the colour difference filter (Annex B, Figure B2) is twice the delay in the luminance filter (Annex B, Figure B1). Because of the difficulty of using analog delay networks and equalizing these delays within the passband tolerance, it is recommended that the majority of the delay differences (integer multiples of the sampling period) should be equalized in the digital domain. In correcting the remainder, it should be noted that the sample and hold circuits in the decoder introduce a flat delay of half a sampling period. The passband tolerances for amplitude fluctuation and group delay are very tight. Current research has shown that it is necessary to ensure that a large number of codecs can be cascaded without compromising the high quality of the 4:2:2 coding standard. Due to the performance limitations of currently available test equipment, it may be economically difficult for the factory to meet the tolerances of each filter in production. However, it is still possible to design filters that meet the specified characteristics. The factory is required to make efforts in production to adjust each filter to the specified characteristic template. The specifications given in Appendix B are intended to maintain the spectral content of the Y, Cr, and Cs signals as much as possible throughout the component signal chain. However, it is obvious that the spectral characteristics of the color difference signals in the picture monitor or at the end of the component signal chain will inevitably be affected by the slow roll-off filters inserted. Additional Notes:
This standard was proposed by the Ministry of Radio, Film and Television. The Standardization Planning Institute of the Ministry of Radio, Film and Television is responsible for the technical coordination of this standard. This standard was drafted by the Science and Technology Department of the Ministry of Radio, Film and Television. The drafter of this standard is Kang Songshi.ns
Figure B3 Characteristics of a digital filter for converting the sampling rate of a colour difference signal from 4:4:4 to 4:2:2 Note: ① The characteristic values of the ripple and group delay are relative to their values at 1kHz. The solid line represents the actual limit and the dotted line gives the limit recommended by the theoretical design. ② In the digital filter, the actual and designed limits are the same. By design, the delay distortion is zero. ③ In the digital filter (Figure B3), the amplitude-frequency characteristic (on a linear scale) should be skew-symmetrical at the half-amplitude point (as shown in the figure). ④ In the recommendation for filters used in the coding and decoding process, it is envisaged that the sinx/a characteristic of the sample and hold circuit is corrected in the post-filter after the digital-to-analog conversion.
Appendix C
Some guidance on the implementation of the filters recommended in Annex B (reference)
In the recommendation for filters used in the coding and decoding process, it is envisaged that the sinx/a characteristic is corrected in the post-filter after the digital-to-analog conversion. The combined passband tolerance of the filter plus the sina/a corrector plus the theoretical sina/α characteristic should be the same as the tolerance given to the filter alone. This is easily achieved if the filter, (sn/3) corrector and delay equalizer are considered as a single unit during the design process. The total delay caused by filtering and encoding the luminance and color difference components should be the same. The delay in the color difference filter (Appendix B, Figure B2) is twice the delay in the luminance filter (Appendix B, Figure B1). Because it is difficult to use an analog delay network and equalize these delays within the passband tolerance, it is recommended that the majority of the delay difference (integer multiples of the sampling period) should be equalized in the digital domain. When correcting the remainder, it should be noted that the sample and hold circuit in the decoder will cause a flat delay of half a sampling period. The passband tolerance for amplitude fluctuation and group delay is very tight. Current research shows that it is necessary to ensure that a large number of codecs can be cascaded without compromising the high quality of the 4:2:2 coding standard. Due to the performance limitations of existing test equipment, it may be economically difficult for factories to meet the tolerances of each filter in production. However, it is still possible to design filters that meet the specified characteristics. Factories are required to make efforts in production to adjust each filter according to the specified characteristic template. The specifications given in Appendix B are intended to maintain the spectral content of Y, Cr, and Cs signals as much as possible throughout the component signal chain. However, it is obvious that the spectral characteristics of the color difference signal in the image monitor or at the end of the component signal chain will inevitably be affected by the inserted slow roll-off filter. Additional notes:
This standard was proposed by the Ministry of Radio, Film and Television. The Standardization Planning Institute of the Ministry of Radio, Film and Television is responsible for the technical coordination of this standard. This standard was drafted by the Science and Technology Department of the Ministry of Radio, Film and Television. The drafter of this standard is Kang Songshi.ns
Figure B3 Characteristics of a digital filter for converting the sampling rate of a colour difference signal from 4:4:4 to 4:2:2 Note: ① The characteristic values of the ripple and group delay are relative to their values at 1kHz. The solid line represents the actual limit and the dotted line gives the limit recommended by the theoretical design. ② In the digital filter, the actual and designed limits are the same. By design, the delay distortion is zero. ③ In the digital filter (Figure B3), the amplitude-frequency characteristic (on a linear scale) should be skew-symmetrical at the half-amplitude point (as shown in the figure). ④ In the recommendation for filters used in the coding and decoding process, it is envisaged that the sinx/a characteristic of the sample and hold circuit is corrected in the post-filter after the digital-to-analog conversion.
Appendix C
Some guidance on the implementation of the filters recommended in Annex B (reference)
In the recommendation for filters used in the coding and decoding process, it is envisaged that the sinx/a characteristic is corrected in the post-filter after the digital-to-analog conversion. The combined passband tolerance of the filter plus the sina/a corrector plus the theoretical sina/α characteristic should be the same as the tolerance given to the filter alone. This is easily achieved if the filter, (sn/3) corrector and delay equalizer are considered as a single unit during the design process. The total delay caused by filtering and encoding the luminance and color difference components should be the same. The delay in the color difference filter (Appendix B, Figure B2) is twice the delay in the luminance filter (Appendix B, Figure B1). Because it is difficult to use an analog delay network and equalize these delays within the passband tolerance, it is recommended that the majority of the delay difference (integer multiples of the sampling period) should be equalized in the digital domain. When correcting the remainder, it should be noted that the sample and hold circuit in the decoder will cause a flat delay of half a sampling period. The passband tolerance for amplitude fluctuation and group delay is very tight. Current research shows that it is necessary to ensure that a large number of codecs can be cascaded without compromising the high quality of the 4:2:2 coding standard. Due to the performance limitations of existing test equipment, it may be economically difficult for factories to meet the tolerances of each filter in production. However, it is still possible to design filters that meet the specified characteristics. Factories are required to make efforts in production to adjust each filter according to the specified characteristic template. The specifications given in Appendix B are intended to maintain the spectral content of Y, Cr, and Cs signals as much as possible throughout the component signal chain. However, it is obvious that the spectral characteristics of the color difference signal in the image monitor or at the end of the component signal chain will inevitably be affected by the inserted slow roll-off filter. Additional notes:
This standard was proposed by the Ministry of Radio, Film and Television. The Standardization Planning Institute of the Ministry of Radio, Film and Television is responsible for the technical coordination of this standard. This standard was drafted by the Science and Technology Department of the Ministry of Radio, Film and Television. The drafter of this standard is Kang Songshi.
Tip: This standard content only shows part of the intercepted content of the complete standard. If you need the complete standard, please go to the top to download the complete standard document for free.