DE2402513A1 - Binary coded video signal compression method - with analogue signals sampled quantised and coded into words of length N - Google Patents

Abstract

Codewords are transmitted and reconverted in the receiver in luminance signals. Adjacent scanned values are compared at the transmitter, and the richness of picture details determined and a criterion derived. Depending on this criterion, a first scanning frequency and a coarse quantisation are used, so that at least two adjacent scanned values are transmitted in one codeword. Picture parts with few details are scanned at a second frequency, and a very fine quantisation is used, so that only one scanned value and one code signal derived from the criterion can be transmitted in one codeword. When the code signal is missing, the two coarsely quantised scanned values are reconverted by the first frequency, and when the code signal is present, the finely quantised scanned value is reconverted by the second frequency.

Description

"Method for compressing the message flow binary-coded Video signals "The invention relates to a method for compressing the flow of messages binary-coded video signals, in which the analog signals are transmitted by an image recording device generated video signals are sampled, quantized and binary coded as code words of the Length n are transmitted and on the receiving side the sampled values after decoding of the transmitted code words in an image display unit in brightness signals be transformed back, The transmission of analog video signals requires a bandwidth of approx. 5 NHz. For portraits when telephoning if you are satisfied with a quality that has a bandwidth of 1 NEz for the analog Video signals required.

There is hardly any deterioration in digital communication the image quality when 7 to 8 bits are used per sample. this means taking into account the sampling theorem bit rates of 70 bits for television quality and 14 bit / s for portrait video call quality.

There is a reduction in the flow of messages during image transmissions therefore desirable for economic reasons.

To reduce the flow of messages, one uses instead of a transmission with pulse code modulation (PCM transmission) a transmission using differential pulse code modulation (DPCM). Approx. 3 bits per sample value are then sufficient (T. Stenger, "A method for effective intraframe DPCM coding ", message from the FTZ of the German Federal Post Office, Darmstadt (1971)). However, such DPCM methods have the disadvantage that transmission errors as a result of error propagation clearly visible picture interference lead in the form of light or dark horizontal stripes.

To reduce the. Transmission error influences are therefore at Some proposals use DPCM encoders that have one at the sender and receiver "Integrator with leak" included.

This measure reduces the influence of transmission errors by the image disturbances no longer propagate to the end of the line, but in the course from ten to thirty pixels are degraded. However, arises in the picture plane eiremer or less clearly visible periodic stripe structure parallel to the image contours.

You can only get a completely redundancy-free flow of messages through an optimal coding (BS2J, J. 27 (1948), pp. 379-423 and pp. 623-656). These However, coding is technically very complex and requires precise knowledge of the source statistics as well as a stationary signal source.

The known coding methods are particularly in the but bearing of video signals over relatively strongly disturbed but transmission channels with error rates up to 10 3, as they occur, for example, in local network lines, almost unusable.

The invention is based on the object of a method for compression of the message flow of the type mentioned at the beginning, in which the mentioned Disadvantages, especially the error propagation when transmitting video signals Avoided as far as possible on relatively badly disturbed transmission channels will.

According to the invention, the object is achieved in that on the transmission side the richness of detail of what is to be transmitted by comparing neighboring samples Part of the image is determined and a criterion is derived therefrom that according to the measure of the criterion for detailed image parts a first sampling frequency is used and such a coarse quantization is used that at least two adjacent samples can be transmitted in a single code word of length n and for little detail Image parts used a second sampling frequency and applied such a fine quantization becomes that in a code word of length n only one sample and one from the Criterion-derived identifier can be transmitted that roughly quantized Samples for the code word of length n, including finely quantized samples an identifier derived from the criterion to form the code word of length n and are transmitted that on the receiving side, if the identifier is missing, the two roughly quantized samples are converted back with the first sampling frequency and if the identifier is present, the finely quantized sample value with the second Sampling frequency is converted back and missing, between two transmitted finely quantized Samples lying values by interpolation of neighboring transmitted finely quantized Samples are derived.

This P-double mode system has the advantage that the false and on 2 is limited to 3 samples. Very little propagation / transmission errors are therefore much less disruptive to the viewer than with the known DPCM procedure. Furthermore, through comparatively simple additional measures, a effective subsequent error correction can be achieved. With an undisturbed transmission channel is the achievable image quality for Codings that have an output data rate of four or more bits per pixel have a DPCM coding of the same bit rate approximately equivalent, with relatively heavily disturbed transmission channels with error rates up to 10 3 the image quality is much better. The invention will now be based on of exemplary embodiments explained in more detail. They show: FIG. 1 analog video signal with samples.

Figure 2 code word rich in detail image parts.

Figure 3 code words of low-detail image parts.

Figure 4 is a block diagram of an encoder and decoder for implementation of the method according to the invention.

Figure 5 is a block diagram of an encoder with a device for analogous derivation of the decision criterion according to. the invention.

The principle of the coding process is to save Bits of the code words to be transmitted create an exchange between contrast and area Resolution in the displayed image. In image areas with little detail, a high contrast resolution, but a reduced sampling frequency is used, in rich detail Image parts, on the other hand, have a high sampling frequency and a reduced contrast resolution.

The sampling theorem prescribes that the sampling frequency should be at least twice as high as the highest frequency occurring in the signal must be selected. For 5 llHz broadband television signals, a sampling rate of 10 11Hz is required. However, this high sampling frequency does not have to be this maximum in the entire image area Have rate. It only has to be ensured that the sampling rate for the respective highest frequency contained in the video signal is sufficient.

Experience shows that a reduction in the transmission bandwidth on-half in poorly detailed areas no perceptible image impairment caused. In such image areas, one is sufficient to avoid contour effects Contrast resolution of 128 quantization levels.

In image areas with a great deal of detail and great differences in contrast on the other hand is a reduction in the number of stages to sixteen or eight quantization stages permissible.

However, the full sampling frequency must be used in detailed image parts will.

Be appropriate to the poor or detailed image area two PCM modes used here. The currently used mode of the transmitter will be the Recipient communicated with the appropriate code word.

In FIG. 1, there is an analog signal VS with four sample values a1 to a4 shown as a function of time t. The samples represent a1 to a3 an image part rich in detail, the samples a2 to a4 an image part poor in detail.

The following is an example of a coder that uses a Data rate of 5 bits per sample generated. According to the proposal, two 5-bit words are used here replaced by a 10-bit word.

10-bit words allow 1024 different code words to be agreed. In this example, 900 words are used for the transmission of the sample values detailed image parts.

uses, the remaining 124 are used for low-detail image areas.

Each of the two samples is sampled in the detailed image areas quantized in 30 level steps. Assuming a 246-step reference signal, so here the level levels 0, 1, 2, 3, ... 30 correspond to the representative values 4, 12, 20, 28 to 244. Each representative value comprises of the 246-level reference signal an amplitude range of 8 steps, the step height is thus 3.2% o of the dynamic range In the image areas with little detail, only every second sample is transmitted. The number of steps is 124, the lintel height is two steps from that 246-step reference signal. The lowest amplitude is 0, the highest includes 246 steps.

A particularly simple and advantageous method consists in forming the 10-bit 8 transmission code in accordance with the scheme described below: FIG. 2 shows a transmission code word CW for detailed image parts. The digits of the values 21 to 27 are referred to there in a simplified manner with b to h. The level levels 0 to 29 are represented as 5-bit code words. The code words 11110 and 11111 are not used here. Both 5-bit code words are then combined according to Table 1 to form a 10-bit transmission code word. The first 5-bit word occupies the places 1 to 5 according to the decreasing significance, the second 5-bit word occupies the places 6 to 10, also according to the decreasing significance. Table 1 shows this assignment. Places 1 2 3 4 5 6 7 8 9 10 Value 1 27 26 25 24 23 27 26 25 24 5-bit word A of 5-bit word B of sample a1 sample a2 Table 1 In this case, the receiver supplements the non-transmitted digits in such a way that digits 22, 21 and 2 ° contain digits 11, 0 and 0, respectively .

In Figure 3, the transmission code words CW2 and CW3 are for little detail Image areas shown schematically.

The level steps 0, 2, 4, 6 ... to 126 are represented by the 10-bit transmission code word CW2 according to Table 2 (7-bit word C, for example of the sample a4). Places 1 3 4 5 6 7 8 9 10 Value 1 1 1 1 2² 2¹ 26 25 24 2³ Table 2 Places 1 to 4 are permanently assigned 1, places 5 to 10 contain the values 21 to 26 of the sample a4.

The level stages 128, 130, 132 ... to 246 are determined by the 10-bit transmission code word CW3 according to Table 3 (7-bit word D, for example of the sample a4). Places 1 2 3 4 5 6 7 8 9 10 Value 26 25 24 2³ 2² 1 1 1 1 2¹ Table 3 The pattern 1111 in positions 6 to 9 means that position 7 has the value 1.

The assignment rules of the exemplary embodiment are then always unambiguous if it is ensured that the input 8-bit PCM code word is only 246 levels having. The bit sequence 1111 in positions 1 to 4 or 6 to 9 also indicates that the transmitted 10-bit word belongs to the image area with little detail. Missing 1111 group, this indicates that the transmitted 10-bit code word has two samples from the detailed picture part.

In parts of the image with little detail, only every second sample is sent. However, the omitted sampled values are made from the sampled values at the receiving end of the neighboring pixels and interpolated at the places of the missing samples reinserted.

The encoder can basically be implemented in many versions. In the following, two arrangements are to be described for the examples mentioned, the first of which is 8 bits.

PCM signals, the second from analog signals.

In the upper half, FIG. 4 shows an encoder for 8-bit PCM signals and in the lower half a decoder for the recovery of encoded data on the transmission side PCM signals.

The 8 bit PCM video signal is first fed into an 8 bit shift register 1 cached. The following limiter 2 reduces the Dynamic range in this example from 256 to 246 Qiiantization levels. That means, that of the total of 256 gray levels of a television picture, the last 10 light gray levels be suppressed. Of course, the analog video signal can also be dimensioned in this way that after the analog / digital conversion only a maximum of 246 quantization levels can occur, whereby the downstream limiter can be omitted.

The now following shift registers 3 and 4 delay the limited 8-bit signal by one sampling interval of duration 1 / f1, where f1 is the usual sampling frequency is.

The undelayed signal of the sample value a1 and the doubly delayed signal of the sample a3 is added in the Intexpolator 5 and multiplied by the factor 0.5.

This procedure corresponds to averaging between the sampled values a1 and a3. This mean value formed in this way is now taken from the simply delayed signal a2 is subtracted in an adder 6, one input of which is inverted. The difference signal is used as a measure of the level of detail.

If the difference signal is small, then the fine quantization is with 120 levels selected, otherwise the 30-level coding.

The decision about the working mode to be selected is made in the threshold level 7 likes. It consists, for example, of two comparators that each check whether the signal exceeds a predetermined positive or negative threshold. Of the The determined state is valid for two samples. Therefore it becomes the following 1 bit register 8, which contains the information of the decision, only with half Sampling frequency f2 = 0.5. f1 clocked.

The code converter 9, to which the undelayed signal of the sample a1 at the input, forms code words of the form CW2 (Figure 3a) and CW3 (Figure 3b). It encodes every other sample and forms those intended for this purpose Output code words. The treated exemplary embodiment is corresponding in each case one of 120 code words is formed, each of which has one of the 120 amplitude levels corresponds to the step height 2/256.

A code converter 10 receives the undelayed signals of the sample a1 and the simply delayed signals of the sample a2 are supplied. This code converter 10 encodes each sample. The two 8-bit samples of a code word are quantized in 30 steps each, with each step height being 8/256. The binary values these sample values are combined to form a code word of the form CW1 (FIG. 2).

The effort for the code converter 10, which binary coded the two Samples a1 and a2 combined to form a code word is very small, since it is this composition is a simple bit shift.

A changeover switch 11 controlled by the 1-bit register 8 selects depending on the operating mode the outputs of the code converters 9 and 10 provided. The one at the output of the encoder supplied code words are with the sampling frequency f2 = 0.5. 1 sent.

The decoder shown in FIG. 4, lower half, takes over the transmitted Code words in a storage register, for example a shift register 20 and leads they to the code converter 21. This converts the code words according to their meaning in 8-bit PCM codewords. The code words are defined by the presence or absence of the 1111 bit pattern always clearly decodable. Also controls the. Code converter 21 also includes the mode selection switch 22 at.

For the interpolation mode for processing the finely quantized Samples, for example a 4, become the switch position q for processing of the coarsely quantized samples, the position p of the mode selection switch 22 is set.

The code converter 21 has three outputs, namely r, s and t.

The first 5 bits of a code word can be removed at output r, am Output s the remaining 5 bits of a code word. Let through switch 23 the outputs r and s of the code converter 21 with a first contact p of the mode selection switch 23 connect.

The output t of the code converter 21 gives the code words to the first one to one / input of an interpolator 24, on the other hand to the one changeover contact u of a changeover switch 25. The other changeover contact v of the changeover switch 25 is with connected to the output of the interpolator 24, which is reduced by a factor of 1/2 Provides the sum signal of its input signals. The second entry of the interpolator is fed by a shift register 26, the content of which is from common changeover contact of the mode selection switch 22 is removed. This Changeover contact of switch 22 is also connected to a shift register 27, its 8 bit PCM code words at its output the 8 bit PCM code words at its input of the encoder correspond and those for further processing, for example via a digital / analog converter, into video signals for subsequent image reproduction can be converted back.

The switches 23 and 25 are both in time with the sampling frequency f2 = f1 / 2 switched. The shift register 26 is also clocked with the clock f2, however, the shift register 27 with the original sampling clock 9 f1. These clock pulses are used on the transmit side and on the receive side to improve clarity not shown generated control units, the operation of which is not the subject of Invention and will therefore not be described in more detail, especially since these are common, in the coding and decoding technology common control devices. Hence will is limited here only to the control function of the control unit.

The interpolator 24 calculates the omitted, the finely quantized Sample value previous sample value te by averaging the neighboring sample values, thus for example a3 in Figure 1 from a2 and a4.

The output of the mode selection switch 22 is in the 8-bit shift register 27 buffered in order to eliminate the switching interference of switches 22, 23 and 25. Like shift registers 1 and 20, it is only used for signal improvement. FIG. 5 shows an analog version of the signal interpolation on the transmit side. At this Version becomes the decision on the working mode to be selected from the analog video signal derived. For this purpose, the signal is initially provided with two delay elements 30 and 31 delayed by a sampling interval of duration 1 / f1.

The input of the delay element 30 and the output of the delay element 31 are each connected to one of the two inputs of an interpolator 32, which feeds half the sum signal to the first input of a differential amplifier 33. The output of the delay element 30 is connected to the other input of the differential amplifier tied together.

The difference signal arrives at a symmetrical threshold circuit 34, which only occur when tZ is exceeded a specified positive or negative threshold a signal of one type, for example corresponding to a 1, to a subsequent one 1-bit register 8 outputs.

The analog / digital converter connected to the output of the delay element 31 35 feeds the code converters 9 and 10 already known from FIG. as well as those of the 1-bit register 8 and the shift register 11 are the same as in Figure 4 are.

Since the function of the encoder in FIG. 5 corresponds to that of the encoder in FIG corresponds to that shown, a further functional description of the Figure 5.

To avoid high-frequency image structures of low amplitude may be lost in coding if the difference between the interpolated value and the exact value smaller than the mode switchover threshold is, in a further improvement of the invention, a pre-emphasis is connected upstream of the encoder and the receiver is followed by a de-emphasis that compensates for this pre-emphasis.

It is advisable to choose the pre-emphasis so that the spectral components of the video signals near the upper limit frequency are increased by 6 to 9 dB. It is particularly advantageous to use phase-linear circuits as pre-emphasis, so that a symmetrical overshoot is achieved with steep jumps in the video signal will.

"Aperture equalizers" are often used in television cameras to improve image sharpness. used. They already represent the desired pre-emphasis. They are also below known by the name "Cosinud 'equalizer. For the method according to the invention only the setting of the camera aperture equalizer can be changed so that the described The high-frequency signal components are increased in the desired manner.

The receiver-side de-emphasis can be done with very little effort, for example realized by a low-pass filter with additional de-emphasis properties will.

Claims (2)

Claim Method for compressing the message flow of binary-coded video signals, the analog video signals generated by an image recording device on the transmission side are scanned, quantized and transmitted in binary code as code words of length n and at the receiving end, the sample values after the decoding of the transmitted code words are converted back into brightness signals in an image display unit, thereby characterized in that the richness of detail is determined on the transmission side by comparing neighboring samples of the image part to be transmitted is determined and a criterion is derived from this, that according to the criterion for detailed image parts a first sampling frequency is used and such a coarse quantization is applied that at least two adjacent Samples can be transmitted in a single code word of length n and a second sampling frequency is used for low-detail image parts and such a fine quantization is used that in a code word of length n only one sample and an identifier derived from the criterion can be transmitted, that coarsely quantized samples in addition to the code word of length n, finely quantized samples including an identifier derived from the criterion for the code word of the Length n are put together and transmitted that on the receiving end if there is no Identify the two roughly quantized sampling values with the first sampling frequency are converted back and, if the identifier is present, the finely quantized sample value is reconverted at the second sampling frequency. and missing, between two transmitted finely quantized sample values by interpolation of neighboring values transferring finely quantized samples can be derived. 2. The method according to claim 1, characterized in that the transmission side before comparing neighboring samples, high video frequencies are raised and received at the receiving end after decoding, high video frequencies to compensate for the frequency increase on the transmitter side be lowered. L e r s e i t e