JPS6249784B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a signal processing method for compressing the bandwidth of an analog signal. In recent years, VTR (Video Tape Recorder) is used to record and transmit time-series signals obtained from image information.
This is done using a facsimile machine. Facsimile devices use telephone lines to transmit image information, but the frequency band of telephone lines starts from 300Hz.
Since the frequency is 3.4KHz, it takes about 5 minutes to electronically transmit a signal obtained from an A4 document at a scanning line density of 4 lines/mm. On the other hand, there is a desire to shorten the transmission time by compressing the signal bandwidth and to enable faster and cheaper transmission. Furthermore, in a VTR, video signals are recorded on magnetic tape, but since the signal frequency bandwidth is approximately 4.5 MHz, the magnetic head for recording and playback rotates at a relative speed of approximately 6 m/sec with respect to the magnetic tape. There is a need. In order to rotate the magnetic head at such high speed, mechanical mechanisms such as motors become large. On the other hand, there is a demand for miniaturization of home-use VCRs, but if the frequency bandwidth of the recording signal is compressed and the maximum signal frequency is lowered, the relative speed required of the magnetic head can be reduced, and the mechanical mechanism can also be made smaller. can be converted into As described above, techniques for band compression of time-series signals are required in various fields. Conventional band compression technology was developed on the premise of transmitting digital signals with binary amplitude information.In this case, after signal processing that focuses on the correlation characteristics of the original signal, appropriate encoding , the original signal is subjected to PCM (Pulse Code Modulation) to obtain a band-compressed digital signal with fewer bits than the signal obtained. For example, a facsimile device transmits document information electronically, but there is a correlation between pixels in a document such that there is a high probability that a pixel adjacent to a black pixel is black, and a high probability that a pixel adjacent to a white pixel is white. After converting the signal obtained from the document into a run-length code signal that represents the connection between black pixels and white pixels, highly efficient encoding is performed in which short codes are assigned to run-length states with a high probability of occurrence. There is. One such highly efficient encoding method is
There is a Modified Huffman encoding method, which performs encoding as shown in Table 1.

[Table] In Table 1, for example, white pixels are 3 (run length)
This means that if there are 2 black pixels (run length), it will be encoded as ``11''. On the other hand, in the case of television signals, there is a correlation characteristic that the difference in brightness between adjacent pixels is small, so after converting the television signal into a signal consisting of the difference in brightness between pixels, for example, the amplitude value of the converted signal Pulse coded modulation (hereinafter referred to as PCM) is performed after assigning a short code to a location where the absolute value is small.
There is a method to do this, and it is called the differential PCM (DPCM) method. These band compression processes can reproduce the original signal almost faithfully by performing inverse processing such as decoding after reception. Hereinafter, the above-described band compression processing will be referred to as digital band compression processing, which also includes processing using the ΔM method and orthogonal transform method, which will be described later. Normally, when performing PCM on an analog signal, the input signal is first approximated to a staircase wave with appropriate steps (quantization), then sampled, and then this sample value is
The encoding operation is performed using binary codes. For example, for a television signal, the quantization level is set to about 6 to 8 bits, and the sampling frequency is set to about 10.7MHz, which is approximately equal to the Nyquist frequency. 54~72Mega bit/sec found by product
(For simplicity, it is assumed to be 60 Mb/S). In differential PCM using correlation within one screen,
It can be compressed to about 1/2 compared to a standard PCM signal, resulting in a signal with a bandwidth of about 31 Mb/S (equivalent to about 15 MHz). However, since this signal has a band wider than the 4.5 MHz bandwidth of the original analog signal, it cannot be said that band compression has been achieved for the original analog signal. In addition, in order to compress a signal, it is sufficient to reduce the redundancy (entropy) in the signal and then convert it to a narrow band encoded signal, but this signal processing cannot be performed within the scope of analog signal processing. Have difficulty. In this way, band compression of analog signals is generally not as easy as band compression of digital signals. The signal processing method of the present invention provides a method of compressing the frequency bandwidth of an analog signal having amplitude information of two or more values, and converts the original analog signal into a digital signal with reduced entropy through the digital band compression processing. By converting the signal to an analog signal, band compression of the analog signal is realized. Embodiments of the present invention will be described below with reference to the drawings. In Figure 1a, 1 is the input terminal for the original analog signal, 2 is the digital band compression processing circuit, and 3 is the D/
A converter (converter that converts a digital signal into an analog signal), 4 is a modulator, 5 is an output terminal for compressed analog signals, 6 is an input terminal for compressed analog signals, 7 is a demodulator, 8 is A/D converter (converter that converts analog signals to digital signals),
9 is a decoder for digital band compression processing, and 10 is an output terminal. FIG. 1b shows a differential PCM showing an example of the internal configuration of the digital band compression processing circuit 2 shown in FIG. 1a.
FIG. 3 is a circuit configuration diagram of an encoder. 12 is a differential amplifier, 13 is a nonlinear quantizer, 14 is a decoder, 15
is an adder, 17 is a register, and 18 is a D/A converter. Figure 1c shows the decoder 9 shown in Figure 1a.
A circuit configuration diagram showing an example of the internal configuration (difference)
PCM decoder). 20 is an input terminal, 21
is a decoder, 22 is an adder, 24 is a register, 2
5 is a D/A converter, and 23 is an output terminal. In FIG. 1a, a time-series analog video signal input to an input terminal 1 is processed by a digital band compression processing circuit 2 to obtain a digital signal whose band is compressed compared to a digital signal directly subjected to PCM. Existing digital band compression processing methods include the △M method and the orthogonal exchange method (e.g. Hadamard transform method).
The block diagram shown in FIG. 1b is a differential PCM encoder that performs intra-frame correlation processing and uses the value of the previous pixel as a prediction function. When an NTSC signal is band-compressed by the differential PCM processing circuit, the sampling frequency fs is usually about three times the color carrier frequency (10.7 MHz), and the quantization level is 8 bits. The time-series video signal input to the input terminal 1 of the digital band compression processing circuit 2 is converted into a difference signal by taking the difference from the signal one pixel (1/fs second) before by the differential amplifier 12. Ru. The probability of occurrence of the amplitude value of the difference signal shows a distribution concentrated around the zero value, and takes a quantization level from -256 to 256 when directly digitized. The nonlinear quantizer 13 includes a plurality of level comparators 11
an encoder 16 (for example,
RAM). Level comparators 11a-11
d performs fine quantization where the absolute value of the quantization level is small, and coarse quantization where the absolute value is large. Furthermore, the code configuration of the encoder 16 is such that a shorter code is assigned to a signal with a high probability of occurrence (Modified code as shown in Table 1).
Huffman codes can also be used). The encoded signal is output from a terminal 19 and is also supplied to a decoder 14 to obtain a decoded signal. The register 17 stores the signal of the previous pixel, and the adder 15 adds the difference signal decoded by the copier 14 and the signal of the previous pixel from the register 17. This signal is approximately equal to the original signal of the pixel currently being processed and is stored in register 17 for processing the next pixel. Then, when processing the next pixel, the signal is converted into an analog signal by the D/A converter 18 and input to the differential amplifier 12. The digital band compression processing performed by the circuit shown in FIG. 1b compresses the band to about 1/2 compared to the PCM of the video signal. For example, convert this to NTSC
When used for signal band compression, the transmission speed is approximately 43 Mb/S. The signal compressed in the digital band compression processing circuit 2 of FIG. 1a in this manner is converted into an analog signal by the D/A converter 3. When the input of D/A converter 3 is 8 bits
The average sampling frequency for a 43Mb/S signal is approximately
It becomes 5.3MHz. This value is approximately 1/2 of the sampling frequency of 10.7MHz at which pixel signals are extracted from video signals. As can be seen when the sampling frequency is set equal to the Nyquist sampling frequency, the original signal at input terminal 1 is reduced in bandwidth to approximately 1/2 by digital band compression processing and D/A conversion processing. Compressed. The thus analogized signal is modulated into a signal suitable for recording by a modulator 4 that performs frequency modulation, for example, and is outputted from an output terminal 5. In FIG. 1a, a reproduced signal taken out from an input terminal 6 is demodulated by a demodulator 7 corresponding to a modulator 4, and becomes a band-compressed analog signal. The A/D converter 8 corresponds to the D/A converter 3, and converts an analog signal into a digital band compressed signal. This signal can be reproduced into an analog original signal by a decoder 9 corresponding to the digital band compression processing circuit 2. FIG. 1c is an example of the decoder 9, and FIG.
This is a decoder of the differential PCM encoder (digital band compression processing circuit 2) shown in FIG. Since the decoder 21, adder 22, register 24, and D/A converter 25 have the same functions as the decoder 14, adder 15, register 17, and D/A converter 18, respectively, FIG. With the configuration shown in c, the digital signal input from the input terminal 20 is transmitted to the output terminal 2.
3, it is converted into an analog signal and output. The signal processing method of the present invention as described above is widely used as a transmission/reception system for color televisions.
It can also be applied to the NTSC system. In the NTSC system, when adapting to a signal that has been multiplexed from a luminance signal (Y signal) and a chrominance signal (I signal and Q signal), it is necessary to apply it to some signals or all of each signal. Therefore, more suitable band compression processing can be performed. Figure 2 shows a case where an NTSC signal is separated into the Y signal, I signal and Q signal and then subjected to band compression processing.
This is an example. 27 is an input terminal for the NTSC signal, 28 is a circuit that separates a luminance signal (Y signal) and a color signal, 29 is a circuit that separates a color signal into an I signal and a Q signal, 30, 31
and 32 are circuits that perform digital band compression processing on the Y signal, I signal, and Q signal, respectively; 33, 34, and 35 are D/A converters; and 36, 37, and 38 are band compressed Y signal, I signal, and Q signal, respectively.
A circuit for modulating the signal, 39 an adder for multiplexing the modulated signal, 40 an amplifier, 41 an output terminal, 42 for inputting a signal current to the magnetic head 26 and recording it on a magnetic tape (not shown). terminal,
43 is an input terminal for inputting a signal reproduced from the magnetic tape (not shown) by the magnetic head 26, 44 is an amplifier, 45 is a modulated Y signal, and I
Demodulators 49, 50, 49, 50, for demodulating the band-compressed Y signal, I signal and Q signal; 51 is an A/D converter corresponding to the D/A converters 33, 34, and 35, respectively; 52, 53, and 54 are decoders for the digital band compression processing circuits 30, 31, and 32;
55 is a color signal synthesis circuit, 56 is a color signal and luminance signal synthesis circuit (55 and 56 are circuits for producing an NTSC signal from the Y signal, I signal and Q signal), and 57 is an output terminal. In Figure 2,
The time-series NTSC signal input to the input terminal 27 is sent to a circuit 28 that separates the luminance signal (Y signal) and color signal.
is separated into a luminance signal and a color signal. The color signal is separated into an I signal and a Q signal by a color signal separation circuit 29. The separated Y signal, I signal, and Q signal are baseband
These are analog signals with 4.2MHz, 1.5MHz, and 0.5MHz, respectively, and the digital band compression processing circuit 30,
31 and 32, respectively. Here, the sampling frequency and quantization level of the digital band compression processing circuits 30, 31, and 32 are set as appropriate for each signal. Each signal subjected to digital band compression processing is input to D/A converters 33, 34, and 35, and each is converted into an analog signal. Next, the outputs of the D/A converters 33, 34, and 35 are inputted to modulators 36, 37, and 38 to be modulated, and further, in the next adder 39, the modulated Y signal, I signal, and Q The signals are combined into one time-series signal. An example of such a synthesis method is an FM-AM signal obtained by converting a color signal in a VTR signal to low frequency. In this case, the modulators 37 and 38 amplitude-modulate the I signal and the Q signal so that they have the same carrier wave and their phases are orthogonal to each other. Further, the modulator 36 frequency modulates the Y signal. The signal multiplied and time-seriesized by the adder 39 has an amplitude-modulated color signal in a low frequency region and a frequency-modulated luminance signal in a high frequency region adjacent to the frequency region. Note that the time-series Y signal, I signal, and Q signal output from the adder 39 are band-compressed. The time-series signal outputted from the adder 39 is converted into a signal having a level suitable for recording by an amplifier 40, and is applied to the magnetic head 26 via terminals 41 and 42, and is recorded on the magnetic tape shown below. In the reproduction system, each band-compressed signal must be returned to the original signal. A signal reproduced from the magnetic tape (not shown) via the magnetic head 26 passes through an input terminal 43, an amplifier 44, and is sent to a separation circuit 45. The separation circuit 45 separates the time series signal into modulated Y signal, I signal, and Q signal. Next, demodulators 4 corresponding to the modulators 36, 37, and 38, respectively.
6, 47, and 48 demodulate the modulated Y signal, I signal, and Q signal. The demodulated signals are band compression processed Y signal, I signal and Q signal, and these signals are passed through A/D converters 49, 50,
51 and decoders 52, 53, and 54, the signal is reproduced into the original Y signal, I signal, and Q signal (the basic principle is explained in FIGS. 1a to 1c). In order to display these signals as images on a TV monitor, they are multiplexed by a color signal synthesis circuit 55 and a superposition circuit 56 that superimposes the color signal and the luminance signal, and are converted into time-series NTSC signals and sent to the output terminal 5.
Output from 7. The method explained in the above embodiment is a method in which each multiplexed signal is band-compressed, but only the luminance signal with a wide bandwidth is band-compressed, and the chrominance signal is not band-compressed. You can also make it easier. Further, the signal processing method of the present invention can be applied not only to video signals but also to band compression processing of audio signals, for example, as long as the signals are time-series signals that have redundancy (entropy).

[Brief explanation of the drawing]

FIG. 1a is a circuit configuration diagram showing an embodiment for realizing the signal processing method of the present invention, FIG. 1b is a configuration diagram of a digital band compression processing circuit, and FIG. 1c is a decoder circuit. FIG. 2 is a circuit diagram showing an embodiment of the signal processing method of the present invention applied to the NTSC system of a color television. In the figure, 1, 6, 20, 27, 43...input terminals, 5, 10, 19, 23, 41, 57...
Output terminal, 2, 30, 31, 32...Digital band compression processing circuit, 3, 18, 25, 33, 34,
35...D/A converter, 4, 36, 37, 38...
... Modulator, 7, 46, 47, 48 ... Demodulator,
8,49,50,51...A/D converter, 9,1
4, 21, 52, 53, 54...decoder, 12
... Differential amplifier, 13 ... Nonlinear quantizer, 11
a to 11d... Level comparator, 16... Encoder, 15, 22, 39... Adder, 17, 24...
...Resistor, 26...Magnetic head, 28,2
9, 45... separation circuit, 40, 44... amplifier,
55...color signal synthesis circuit, 56... synthesis circuit.

Claims (1)

[Claims] 1. A signal processing method that uses the correlation of input analog signals to compress the band of a digital signal corresponding to the input analog signal, converts the band-compressed digital signal into an analog signal, and then transmits the signal.