CN1166174C - Method for correcting timer reference of programs in multiplexing - Google Patents

Abstract

A kind of MPEG-2 transport stream packet flow in multiplexing, the correction method of program clock reference, for each input channel of multiplexing system, determine the fixed delay time Ts that a transport stream packet passes through this channel; During the multiplexing process, the actual dwell time Ta of the transport stream data packet is measured; the difference Td between the actual dwell time Ta and the fixed delay time Ts is used as a correction value to adjust the program clock reference in the transport stream data packet. By adopting the method, the function that multiple uncorrelated MPEG-2 transmission code streams can be multiplexed into one MPEG-2 transmission code stream can be achieved, and the introduced error is very small.

Description

Correction method of program clock reference in multiplexing

The present invention relates to the field of media synchronization technology in multimedia communications, in particular to the program clock reference (hereinafter also referred to as PCR ) method to correct when multiplexing the output.

The importance of media synchronization technology in multimedia communication is self-evident, especially the video/audio synchronization processing of the same program. Video/audio synchronization includes two aspects: encoding/decoding synchronization and audio/video lip synchronization. If any aspect is not done well, it will cause great damage to the video/audio quality during the encoding and decoding process. The specific synchronization process includes acquisition of synchronization information in the encoding stage and encoding according to the MPEG-2 synchronization information format to join the MPEG-2 transport stream packet stream (hereinafter also referred to as TS stream or TSP stream). Due to delay changes during multiplexing, it must Make necessary modifications to the program synchronization information, recover the code stream synchronization information in the decoding stage and control the video/audio synchronization decoding and display. The synchronous processing techniques are introduced one by one below.

Figure 1 shows the synchronization model defined by MPEG-2. In order to ensure that the decoder recovers the quality of the program video/audio signal, the MPEG-2 protocol defines a time model, which is used to synchronize the decoded video and audio signal output and control the video and audio signals are reproduced at the rate at which they were encoded.

Please refer to Figure 1, at the encoding end, the encoder marks the display time (PTS) and decoding time (DTS) of each video and audio display unit, and at the same time encodes the display time label and decoding time label in the transport stream according to certain rules sent in. On the other hand, the system clock (STC) must be encoded and added to the code stream and sent at the same time. The encoding method of the system clock is the program clock reference (PCR).

At the decoding end, the decoder uses the program clock reference (PCR) in the code stream to restore the system clock (STC), and arranges the video/audio access unit decoding sequence according to the video/audio decoding time tag (PTS). Similarly, according to the display The time stamp (PTS) arranges the display order of the display units so that the synchronized audio/video can be resumed.

Please refer to Figure 1, MPEG-2 requires end-to-end fixed delay during the entire process of program encoding, storage, transmission and decoding. Under the guarantee of fixed delay, the decoding end can use PCR to accurately recover the system clock of the encoding end. If jitter is caused during the stream multiplexing process, the PCR must be adjusted so that the decoder can accurately recover the system clock.

Encoding synchronization information is to inject correct PCR, PTS/DTS data into MPEG-2 encoding output code stream. MPEG-2 stipulates that the system clock (STC) is 27MHZ, and the precision is 30PPM. PCR is the sampling of the 42bits count value of the system clock. It consists of 33bits PCR-BASE and 9bits PCR-EXTERN. The accuracy of PCR is required to have a maximum jitter range of 500ns. The calculation formula of PCR is as follows:

PCR-BASE=STC/300

PCR-EXTERN=STC%300

Among them, PCR-BASE is 90kHz sampling of STC, and PCR-EXTERN is 27MHz sampling of STC.

PTS/DTS are both STC's 90kHz sampling.

The difficulty of synchronous information encoding is that the accuracy of PCR is particularly high, and the other is that PTS/DTS must maintain the correct relationship with PCR. Too low accuracy of the encoded PCR will affect the recovery of the system clock in the decoding stage, so that it will affect the correct decoding of video/audio, and will also affect the number of times to participate in multiplexing. PTS/DTS must be processed, otherwise it may cause errors in the buffer operation of the decoder, resulting in buffer overflow or abnormal underflow, which are not allowed.

The decoding stage must control and recover the synchronous audio/video signal according to the synchronous information provided in the code stream. MPEG-2 decoder restores synchronization information including recovering system clock and arranging decoding and displaying technologies according to synchronization information.

The decoder has a local 27MHZ clock and a phase-locked loop (PLL). The working principle of the phase-locked loop is that when the first PCR of the program is received, the local STC is set to the PCR value, and the subsequent received The difference between the PCR value and the current STC count value adjusts the local 27MHZ clock to synchronize it with the encoding system clock.

The video frame structure of MPEG-2 includes I, P, and B frames. The I frame is a video frame without prediction, the P frame is a coded video frame predicted according to the previous frame, and the B frame is based on the previous I frame (or P frame). The video frame obtained by predictive coding with the subsequent I frame (or P frame), because the B frame needs to be predicted according to the subsequent frame, the display order and encoding/decoding order will be different, for example, the display order is IBBP frame sequence, its The encoding/decoding sequence should be IPBB, which requires a frame rearrangement process during decoding. MPEG-2 uses DTS to indicate the decoding time, and PTS to indicate the display time. When decoding, the decoding order is arranged according to the DTS, and the display order is arranged according to the PTS. For the case of different PTS/DTS, the decoder should cache the frame and use it as the prediction reference frame. At this time, PTS/DTS abnormal advance or lag may cause buffer operation error, resulting in overflow.

The decoding software detects the synchronous information of the code stream, extracts the synchronous information, and drives the decoder according to the synchronous information to complete the audio/video synchronous decoding and display control. It also includes two parts, that is to extract the program PCR, drive the phase-locked loop, complete the system clock recovery, on the other hand, extract the PTS/DTS, and arrange decoding and display according to the local STC situation.

In the case of multi-program multiplexing, assuming that the stay time of each transport stream data packet TSP in multi-program multiplexing is the same, that is, has a fixed delay, the PCR does not need any modification. However, due to the different processing time of each TSP in the multiplexing process, the existence of input/output FIFO concurrency (bus contention), and the existence of factors such as different input/output code rates, each TSP is multiplexed. The processing delay time (dwell time) of different boards is different, so it is necessary to adjust the PCR of each program, so that the decoder can still recover the accurate encoding system clock according to the PCR value.

The object of the present invention is to provide a kind of PCR correction method in the MPEG-2 transport code stream multiplexing process, can be multiplexed into one way MPEG-2 transport code stream to complete the MPEG-2 that is independent of each other. 2 The function of transmitting the code stream, and at the same time achieve the purpose of reducing the PCR jitter introduced by PCR correction.

Based on the above purpose, the technical solution of the present invention is: a method for correcting the program clock reference in the multiplexing of MPEG-2 transport stream data packets, wherein, for each input channel of the multiplexing system, a transmission The fixed delay time Ts of the flow data packet passing through the channel; the actual residence time Ta of the transmission flow data packet is measured during the multiplexing process; the difference Td between the actual residence time Ta and the fixed delay time Ts is used as the correction value to adjust the transmission flow data The program clock reference in the package.

Preferably, in the multiplexing process, only the actual residence time Ta of the transport stream data packet containing the program clock reference is measured, and the program clock reference in the transport stream data packet is adjusted. In this scheme, it is required Each transport stream data packet entering the input channel is judged to determine whether it contains the program time reference.

Preferably, during the multiplexing process, the actual dwell time Ta of all transport stream data packets is measured, but during the adjustment process, only the program clock reference of the transport stream data packets containing the program clock reference is adjusted.

The method of the present invention is driven by the transmission data packet when calculating the fixed delay and the actual delay of the transmission data packet, and the input time label and the output time label of the transmission data packet can be input and output by the hardware logic circuit according to the actual input and output time of the transmission data packet. The output is precisely locked at all times. This error mainly comes from the 27MHZ clock precision error of the local multiplexing system, and the delay error when the local clock counter is locked by the input/output signal trigger. Since the precision of the 27MHZ clock can be achieved by using a high-precision crystal oscillator, it generally requires more than 50ppm. However, the maximum delay error when triggering locking is one clock cycle, so the method of the present invention is used to correct the PCR, and the introduced error is very small.

In order to make the purpose, solution, and advantages of the present invention clearer, the following examples are given in conjunction with the accompanying drawings to further describe the present invention in detail. in

Figure 1 shows an MPEG-2 synchronization model;

Fig. 2 shows an example diagram of MPEG-2 TS stream multiplexing in the second preferred embodiment of the present invention;

Fig. 3 shows a flow chart of the realization of the TS stream multiplexing and PCR correction method in the second preferred embodiment of the present invention.

The principle of PCR correction in the present invention can be explained as, assuming that the residence time of each TSP in the multiplexing board is known and constant, and is set as a fixed delay time Ts. During the multiplexing process, the actual residence time Ta of each TSP can be measured to obtain the difference Td of the residence time. PCR was adjusted according to Td. The formula is as follows:

Td＝Ta-Ts (Formula 1)

The fixed delay time can be the average dwell time, or a value ahead or behind the average time. According to the above formula 1, the difference Td between the actual residence time Ta of each TSP and the fixed delay time Ts can be calculated, and the PCR can be adjusted based on Td.

Preferably, the fixed delay time Ts is the actual delay time of the first transport stream packet containing the program clock reference of the input channel; and when the code rate of the input channel changes, the fixed delay of the input channel The time Ts should be recalculated. When the code rate of the output channel changes, the fixed delay time Ts of all input channels should be recalculated.

Preferably, a 27MHZ system clock is set in the multiplexing system, and N system clock counters are set according to the number N of input and output ports, and each counter is fixedly assigned to an input or output port for recording various input and output ports. output moment.

Preferably, during the multiplexing process, said measuring the actual residence time Ta of the transport stream data packet further includes the following steps: when the transport stream data packet enters the multiplexing system, record its input time tag; when the transport stream data packet When it is output, record its output time tag; use the output time tag and the input time tag to calculate the actual stay time Ta of the transport stream data packet in the multiplexing system.

In the first preferred embodiment of the present invention, when realizing the PCR correction in the multiplexing process of the MPEG-2 transport code stream, there are some following technical characteristics:

The system sets a 27MHZ system clock. According to the number N of input and output ports, N system clock counters are set. Each counter is fixedly assigned to an input or output port. These counters are used to record various input and output moments.

When each transmission data packet enters the system, its input time label must be recorded, and when the transmission data packet at this time contains PCR, the input time label must be saved until the data packet is output before being discarded.

When a transmission data packet needs to be output, it is necessary to record its output time tag. When the output transmission data packet contains PCR at this time, it is necessary to use the output time label and the input time label of the transmission data packet to calculate the transmission time Latency in multiplexed systems.

Using the delay time value calculated by the output time label and input time label of the transmitted data packet, subtracting a fixed value of the delay time, the adjustment value that should be adjusted by the PCR in the transmission packet can be obtained, and the PCR value can be corrected by using this adjustment value. Then the corrected PCR value can be obtained.

In this embodiment, the fixed value of the delay time is set like this: for each input channel, a fixed delay time must be set, and this fixed delay time is the actual delay of the first transmission data packet containing PCR of the input channel time. The general method is to use an empirical value as the fixed value of the delay time. The disadvantage of this approach is that the bit rate range of each input channel is very large, and the delay time of transmitting data packets is determined by the input and output channels and the current input. Due to factors such as the number of stream channels, it is not easy to obtain an experience value for each situation, or the error of the obtained experience value is too large. In this embodiment, using the actual delay time of the first transmission data packet containing PCR of the input channel as the fixed value of the delay time can bring the following advantages: it is closer to the actual fixed delay time, so that the accuracy error caused by the correction is reduced, and at the same time It is also easier to do.

When the code rate of the input channel changes, the fixed delay time of the channel needs to be recalculated, and when the code rate of the output channel changes, the fixed delay time of all input channels needs to be recalculated. To keep the fixed delay time as close as possible to the time delay time, reduce the PCR accuracy error introduced by the correction.

To summarize the PCR correction calculation method of this embodiment is as follows: for a transmission data packet containing PCR, mark its channel input time label as T(in)j, where j represents the jth transmission data containing PCR input by the channel package, after multiplexing, the output time label is T(out)j, the multiplexing fixed delay time of this channel is ConstDelta, the multiplexing actual delay time of each transmitted data packet is ActualDelta, and the original PCR value is recorded as OrgPCR, after correction The PCR value of is recorded as NewPCR. The following is the calculation formula of NewPCR:

(1) Fixed delay time: ConstDelta=T(out)1-T(in)1

(2) Actual delay time: ActualDeltaj=T(out)j-T(in)j

(3) PCR correction formula: NewPCRj=OrgPCRj+(ActualDeltaj-ConstDelta)

Fig. 2 is a diagram showing an example of MPEG-2 TS stream multiplexing in the second preferred embodiment of the present invention. Among them, IN_FIFO1 receives and buffers the first input TS stream data, IN_FIFO2 receives and buffers the second input TS stream data, and IN_FIFOn receives and buffers the nth input TS stream data. Multiple channels of input TS stream data are multiplexed by the MPU and transmitted to the buffer OUT_FIFO of the output TS through the bus in time division, and are driven by the output clock to form an output TS stream. The TS stream is composed of TS data packets transmitted in sequence. Each TS data packet has a fixed length of 188 bytes. The TS packet header has 4 bytes. The first byte must be the synchronization byte 0X47, the synchronization byte As the delimitation mark of the TS data packet in the TS stream, the correct TS data packet can be obtained from the TS stream by demarcating the synchronization byte. It should be noted that the MPU processes the input TS stream in units of 188-byte TS data packets, and the unit of output to OUT_FIFO is also in units of 188-byte TS data packets. For the MPU, the processing during multiplexing includes PCR correction, PID reallocation, PSI reconstruction and other aspects. The present invention relates to the processing of PCR correction.

Fig. 3 shows a flow chart of the realization of the TS stream multiplexing and PCR correction method in the second preferred embodiment of the present invention. Each step will be described in detail below in this embodiment.

Step 202: Record the time stamp when the TS data packet arrives at the system. When the TS data packet arrives at the multiplexing system, the system demarcates the TS data packet according to the synchronization byte of the TS data packet. After the delimitation is completed, the input time of each subsequent input TS data packet can be obtained through the synchronization byte and the current value of the system clock counter, that is, when the multiplexing system receives the synchronization byte of the TS data packet, it reads The current value of the system clock counter, this value is the input time label Tinj of the TS data packet, and j represents the jth input TS data packet from the first TS data packet input to the FIFO.

Step 203: TS data packets are sent to FIFO. For a bus-type time-division multiplexing system, it is possible for multiple inputs to request the bus at the same time. The system needs a scheduling process for multiple input requests, so the input data needs to be cached.

Step 204: The MPU reads the input TS data packet into the processing buffer. For multiple input requests, the MPU can only process one of them at any time, and the MPU selects one of the input requests for processing according to the input scheduling policy (such as first-come-first-served, etc.). When the MPU processes the TS data packet, it needs to read the TS data packet from the input FIFO into the data buffer of the MPU.

Step 205: The MPU performs multiplexing processing on the TS data packets. From the multiplexing of multiple MPEG-2 TS streams into one MPEG-2 TS stream, the input TS data packets need to be processed accordingly to ensure that the multiplexed TS streams still conform to the MPEG-2 standard. Regulation. These processes include reassignment and modification of PID, etc.

Step 206: When the output FIFO request is made, the system records the output time label of the next data packet according to the output FIFO request signal and the current value of the system clock counter, that is, when the output FIFO generates a request signal, a TS needs to be sent at this time data packet, so when the output FIFO generates the request signal, it is the time when the next TS data packet is sent out. At this time, the current value of the system clock counter is the time label when the next TS data packet is output. The output time label of the j-th packet from which 1 TS packet is output.

Step 207: If there is a PCR in the TS data packet, calculate the correction value of the PCR according to the input time tag and the output time tag, and correct the PCR in the TS data packet. The method of calculation is as described in Section 5g.

Step 208: The TS data packet is output to the output FIFO. The TS data packets processed by MPU multiplexing and PCR correction are sent to the output FIFO, driven by the output clock, and the multiplexed MPEG-2TS code stream is generated from the output FIFO.

The method of the present invention is driven by the transmission data packet when calculating the fixed delay and the actual delay of the transmission data packet, and the input time label and the output time label of the transmission data packet can be input and output by the hardware logic circuit according to the actual input and output time of the transmission data packet. The output is precisely locked at all times. This error mainly comes from the 27MHZ clock precision error of the local multiplexing system, and the delay error when the local clock counter is locked by the input/output signal trigger. Since the precision of the 27MHZ clock can be achieved by using a high-precision crystal oscillator, it generally requires more than 50ppm. However, the maximum delay error when triggering locking is one clock cycle, so the method of the present invention is used to correct the PCR, and the introduced error is very small.

The technical features in the embodiments of the present invention can be further combined, modified, and replaced without departing from the spirit and scope of the present invention, so details are not repeated here. And it should be understood that the above descriptions are only preferred embodiments of the present invention, and should not be used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention, All should be included within the scope of the claims of the present invention.

Claims (10)

1, a kind of MPEG-2 transport stream packet flow in multiplexing, the correction method of program clock reference, it is characterized in that: For each input channel of the multiplexing system, determine a fixed delay time Ts for a transport stream packet to pass through the channel; Measuring the actual residence time Ta of transport stream packets during the multiplexing process; The difference Td between the actual dwell time Ta and the fixed delay time Ts is used as a correction value to adjust the program clock reference in the transport stream data packet. 2. The method according to claim 1, wherein said fixed delay time Ts is the actual delay time of the first transport stream data packet containing the program clock reference of the input channel; and when the input channel When the code rate changes, the fixed delay time Ts of the input channel should be recalculated, and when the code rate of the output channel changes, the fixed delay time Ts of all input channels should be recalculated. 3. The method according to claim 1, wherein a 27MHZ system clock is set in the multiplexing system, and N system clock counters are set according to the number N of input and output ports, and each counter is fixedly assigned to an input Or output ports for recording various input and output moments. 4. The method according to claim 1, characterized in that, in the multiplexing process, only measure the actual residence time Ta of the transport stream data packet containing the program clock reference, and measure the program clock in the transport stream data packet The benchmark is adjusted. 5. The method according to claim 1, characterized in that, during the multiplexing process, the actual residence time Ta of all transport stream data packets is measured, but during the adjustment process, only the transport stream data containing the program clock reference are adjusted Package's program clock reference. 6. The method according to claim 1, wherein, in the multiplexing process, said measuring the actual residence time Ta of the transport stream data packet further comprises the following steps: when the transport stream data packet enters the multiplexing system, Record its input time label; when the transport stream data packet is output, record its output time label; use the output time label and the input time label to calculate the actual residence time Ta of the transport stream data packet in the multiplexing system. 7. The method as claimed in claim 6, characterized in that, when the transport stream data packet contains the program clock reference, its input time tag can be discarded at least until the data packet is output; and the transport stream data packet When the program clock reference is included, it is necessary to use the output time tag and the input time tag to calculate the actual stay time Ta of the transport stream data packet in the multiplexing system. 8. The method of claim 1, wherein the second step further comprises: The transport stream data packet arrives at the multiplexing system, and the system records the input time label of the transport stream data packet; The transport stream packet is sent to the FIFO; The MPU reads the transport stream packet from the FIFO into the buffer; The MPU performs multiplexing processing on the transport stream data packet; Output FIFO request output, the multiplexing system records the current output time label; When the transport stream packet contains the program clock reference, the actual stay time Ta is calculated based on the input time tag and the output time tag. 9. The method according to claim 6, 7, or 8, wherein the input time label of the transport stream data packet is that when the multiplexing system receives the synchronization byte of the transport stream data packet, Current value of the system clock counter. 10. The method according to claim 6, 7, or 8, wherein the output time tag of the transport stream data packet is the current value of the system clock counter when the output FIFO sends a request signal.

Images