JPH1141608A - Image transmitter, image coding method and image coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To transit data encoded by plural encoders within a limited frequency and while keeping image quality of data uniform. SOLUTION: A multiplexing part 10 takes in data existing in intermediate buffers 7, 8, 9 to a multiplex transmission buffer at each multiplexing time and outputs to a transmission line at a prescribed rate. In this case, a buffer occupied amount (transmission delay) is fed to each encoder. Each encoder controls the rate and the transmission amount to the multiplexing part 10 so as to avoid overflow/underflow at a receiver side. The configuration with distributed encoders is available and VBR transmission on a network is realized.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital compression encoding apparatus for video signals and a transmission apparatus for efficiently transmitting encoded data, and more particularly to multi-channel transmission of images in satellite communication or LAN. The present invention relates to a technique for compressing and encoding a plurality of image data within a limited band such as video transmission on a simple network and transmitting the image data.

[0002]

2. Description of the Related Art As a conventional technique for transmitting a plurality of image data, a fixed band is assigned to each image, and each coding apparatus performs coding in a predetermined band and multiplexes the data. was there. However, since the complexity of the image changes over time, in order to always encode and transmit so that the deterioration is not noticeable, even if an image whose input band is the most difficult to allocate is input, the image quality deteriorates. It is necessary to reduce the transmission efficiency, resulting in poor transmission efficiency. Therefore, the bandwidth is changed according to the coding difficulty of the image being coded, and when the image is simple, the bandwidth is reduced and transmitted, so that efficient transmission is performed and more channels are transmitted with the same transmission bandwidth. There have been attempts to do so. Examples of the prior art include, for example, WO
95/32565 has that description. FIG. 13 is a configuration diagram of a conventional image transmission device. In FIG. 13, 101, 102, 10
Reference numeral 3 denotes a video encoding unit, and 104 denotes a multiplexing unit.

The operation of the above configuration will be described. The video encoding units 101, 102, and 103 include a multiplexing unit 104
The input video is coded according to the target rate specified by. The multiplexing unit 104 includes the video encoding units 101 and 10
It multiplexes the coded data from 2, 103 and outputs the multiplexed data to the transmission path, and instructs each video coding unit on the target rate. The target rate is set so that the total value matches the transmission rate. At this time, each video encoding unit encodes the amount of encoding distortion caused by encoding at every fixed period by the multiplexing unit 10.
Send to 4. The multiplexing unit 104 raises the target rate to be set for a video with much distortion according to the amount of distortion of each encoder while maintaining the transmission rate for the total of the total target rates, Lower. Therefore, a video encoded by each encoding unit is assigned a bit rate according to the difficulty of encoding.

[0004]

However, in the above-described configuration, since the rate is corrected so as to correct the distortion after the distortion has occurred, it is difficult to change the scene to a scene where encoding is difficult. There was a problem that image quality deteriorated.

Further, since it is necessary for the multiplexing unit to calculate a target rate to be assigned to each encoding unit based on distortion information in all the encoding units, the number of encoding units connected to the multiplexing unit is large. In such a case, there is a problem that the amount of calculation for distributing the rate increases, and also, when each encoding unit exists in a distributed manner on the network and transmits video data over the network, the distortion information is collected. There is a problem that it is difficult to set the rate of each encoding unit.

The present invention solves the above-mentioned problems, and maintains the uniformity of the image quality of the video encoded by each encoding unit even when changing scenes. It is an object of the present invention to provide a video transmission method and apparatus in which a rate is assigned according to each input video within a band allowed by the whole, and the image quality is made uniform.

[0007]

In order to achieve the above object, an image transmission apparatus according to the present invention comprises a plurality of image encoding units and a transmission processing unit, wherein the transmission processing unit is provided for each multiplex timing. The coded data specified by the image coding unit is multiplexed and transmitted to a predetermined transmission path, and calculated based on the amount of coded data of each image coding unit, and the amount of delay before transmission to the decoder is calculated. Sends multiplexed information that can be calculated to each of the image encoding units, and the plurality of image encoding units determine a quantization width at the time of quantization in image encoding based on the multiplexed information from the transmission processing unit. And determining the image, encoding the image, and transmitting the encoded data to the transmission processing unit at each multiplex timing to instruct the transmission processing unit.

Further, the transmission processing unit of the present invention includes a transmission buffer, and fetches the coded data specified by each image coding unit into the transmission buffer at each multiplex timing, and according to a predetermined transmission rate in the fetching order. And outputting, as multiplexed information, the occupation amount of the encoded data in the transmission buffer.

The image encoding unit according to the present invention includes a quantization width determining unit and a basic encoding processing unit that encodes an image by performing quantization with the quantization width specified by the quantization width determining unit. The quantization width determination means calculates the decoder buffer occupancy of the plurality of decoders that receive the encoded data transmitted from each of the image encoding units through the transmission processing unit from the multiplexing information, and calculates the calculated decoder buffer. The quantization width is set based on the occupation amount and the amount of coded data that has not been transmitted after coding.

Further, the image encoding unit of the present invention further has an encoded data storage unit for temporarily storing encoded data,
The coded data storage unit temporarily stores untransmitted data that has not been sent to the transmission processing unit among the coded data for each transmission timing, and an upper limit value previously set in the entire image coding unit. Or transmitting the encoded data to the transmission processing unit with a value obtained by subtracting the decoder occupancy calculated based on the multiplexing information from the decoder buffer size assumed on the receiving side as an upper limit value. is there.

Further, the image transmission apparatus of the present invention comprises a band management unit for performing band management on a transmission line and one or more image encoding units, and the band management unit includes the one or more image encoding units on the transmission line. Determine the total bandwidth of the transmission rate of the image encoding unit,
The transmission of coded data requested to be transmitted by a certain image encoding unit at each transmission timing is permitted, and transmission information relating to a transmission delay that changes according to the permission is transmitted to another image encoding unit on a transmission path. The one or more image encoding units each encodes an image by quantizing the transmission information and the untransmitted encoded data and a quantization width based on an assumed buffer size of a receiving side. Is what you do.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to FIGS.

(Embodiment 1) FIG. 1 is a configuration diagram of an image transmission apparatus according to an embodiment of the present invention. In FIG. 1, reference numerals 1, 2, and 3 denote video coding units for coding an input image, each including coding units 4, 5, and 6 and intermediate buffers 7, 8, and 9, and 10 a multiplexing unit. Yes, it is composed of a reading and sending control unit 10a and a sending buffer 10b.

The above operation is performed in the above configuration.
The video encoders 1, 2, and 3 respectively compress and encode the input video signals. Encoding unit 4 in each video encoding unit,
Reference numerals 5 and 6 denote MPEG2 encoders that convert video signals into I / O signals.
M of SO / IEC13818-2 (commonly known as MPEG2 video)
The encoded data is encoded into a video signal conforming to P ＠ ML, and encoded data to be multiplexed in the next multiplexing cycle is output to the intermediate output buffers 7, 8, and 9 at each multiplexing timing. The multiplexing unit 10 includes a multiplex transmission buffer 10b, and controls the coded data stored in the intermediate buffers 7, 8, and 9 of the video encoding units 1, 2, and 3 for each multiplex cycle under the control of the read transmission control unit 10a. Are all fetched into the transmission buffer, and the data is transmitted to the transmission path according to the transmission rate according to the fetching order in the transmission buffer. At this time, the order of taking in the data from the buffer in each video encoding unit is not particularly limited as long as the encoded data in this period is taken into the transmission buffer. Then, the buffer occupancy value Bmux_oc of the encoded data in the transmission buffer 10b at the time when the capturing is further completed is sent to each video encoding unit. This Bmux
Assuming that the output transmission rate is R from the _oc value, each encoding unit can estimate the relative transmission delay time of the multiplexed encoded data by Bmux_oc / R, and using this, each video encoding unit Encode the video. Hereinafter, the encoding process in each video encoding unit will be described.

The video encoder compresses and encodes the video signal into MPEG2 video data. In encoding in MPEG2, a video signal is subjected to predictive encoding between frames,
The prediction error component is DCT (discrete cosine transform) coded. The code amount of the encoded data is performed by increasing or decreasing the quantization width when quantizing the DCT coefficient. When the quantization width is small, the generated code amount increases, but the image is encoded into a compressed image closer to the original image. On the other hand, when the quantization width is increased, the generated code amount decreases, but the image becomes noticeably deteriorated due to distortion due to a quantization error. Therefore, in order to encode the image with a uniform image quality closer to the original image, it is necessary to control the amount of change so that the quantization width is as small as possible and as constant as possible. In this embodiment, since the image quality and the generated code amount are controlled by the quantization width, details of the encoding unit other than the quantization width control are described in a known technique (for example, “the latest MPEG textbook”).
(ASCII Publishing, 1995), pp. 103-105), the control of the quantization width will be described in detail below.

FIG. 2 is a configuration diagram of an encoding unit in the video encoding unit. In FIG. 2, reference numeral 11 denotes a basic encoding unit,
Difference processing unit 11a, DCT processing unit 11b, quantization processing unit 11
c, variable-length coding unit 11d, and inter-frame prediction unit 1
1e, 12 is a bit counter, 13 is a buffer for temporarily storing and transmitting encoded data, 14
Is a control unit.

The basic encoding unit 11 is a basic encoding unit that encodes an image into an MPEG2 video format, and calculates a prediction error of a prediction value from the inter-frame prediction unit 11e by a difference processing unit 1.
It is calculated in 1a, subjected to DCT processing by the DCT processing unit 11b, quantized by the quantization processing unit 11c, and then encoded by the variable length encoding unit 11d. At this time, the basic encoding unit 11 receives the quantization width from the control unit 14 in the quantization processing unit 11c, encodes the input video signal according to the quantization width, and transmits the generated data to the buffer 13. The bit counter 12 counts the number of bits of data generated by the basic encoding unit 11 and sends the number of generated bits to the control unit 14 at each multiplex timing. The control unit 14 determines the quantization width based on the buffer occupancy value Bmux_oc from the multiplexing unit 10 and the bit generation amount value from the bit counter 12, and sends the number of bits to be sent to the basic encoding unit 11 and to the intermediate buffer. And instructs it to the buffer 13. The buffer 13 outputs the data from the basic encoding unit 11 in accordance with the number of transmission buffers indicated by the control unit 14 and sends the data to the intermediate buffers 7, 8, and 9.

FIG. 3 is an explanatory diagram of the calculation process of the quantization width in the control unit 14. In FIG. 3, reference numeral 201 denotes a start process; 202, a process for initializing the reference quantization width, the buffer occupancy in the encoding unit, the intermediate buffer occupancy, and the parameter Di; 203, an encoding end detection process; , 205 is a quantization width calculation process, 206 is a bit generation amount detection process, 207 is a transmission bit number calculation process, 20
8 is a parameter Bmux_oc receiving process, 209 is a parameter Di calculating process, 210 is an underflow avoiding process,
211 is an encoding skip process. The group of encoding units in FIG. 1 performs the same operation on an input video, and here, as an example, will be described as the processing of the i-th video encoding unit. The control unit 14 sets the buffer size of the assumed decoder on the receiving side that reproduces the data encoded by the i-th video encoding unit as Bdec_size until the encoding is completed after the initialization process, and sets Bdec_size for each multiplexing cycle. The quantization width is calculated by the following equation.

[0019]

(Equation 1)

Here, q_st is set in advance at the stage of initialization with a reference quantization width, and is set to a quantization width sufficient to obtain an image quality to be realized. Di is a variable that changes for each multiplex cycle, and its calculation method will be described. The encoding unit stores the encoded data in the intermediate buffer Bmi for each multiplexing cycle.
To send to. At this time, the control unit 14 detects the bit generation amount in the period from the bit counter 12, and when the sum of the bit number REMi of the data that has not been transmitted before that is smaller than the size of the intermediate buffer, It sends to the buffer 13 that all encoded data is sent to the intermediate buffer within the range not exceeding the upper limit rate specified in the standard (for example, 15 Mbps for MP @ ML) (the sum of REMi and the number of generated bits). An instruction is given as the instruction amount Bmi. If the size is larger than the intermediate buffer size, the capacity of the intermediate buffer is designated as the value of Bmi, and the rest is held as untransmitted data.
Also, when multiplexing is performed at this multiplexing cycle that can be calculated from Bmux_oc, the amount of data that has been encoded by this encoding unit at the time of arrival on the receiving side and is in an unreproduced state (for example, Bmux_oc is equal to Bmux_si
In the case of ze, if the time stamp of the playback time is attached so that playback is performed without delay time, (Bmux_size−Bmux
x_oc) / R is obtained by the total value of bT generated within the time. ) Is equal to or larger than Bdec_size, Bmi is set so that the value obtained by subtracting Bdec_size from the unplayed data amount becomes REMi.
(That is, the amount of unreproduced data arriving at the receiving end is set to Bdec_size to prevent overflow). That is, the upper limit of the encoded data is obtained by subtracting the decoder occupancy from the decoder buffer size. Then, the number of bits of the untransmitted data is calculated by the following equation.

[0021]

(Equation 2)

Where bT is the number of bits generated in the multiplex period. Next, the current data occupancy Bmux_oc of the transmission buffer of the multiplexing unit 10 is received, and Di is calculated by the following equation.

[0023]

(Equation 3)

However,

[0025]

(Equation 4)

It is assumed that From the Di calculated as described above, a quantization width is calculated according to (Equation 1) and coding is performed. In addition, Bdec_size-Di is not reduced to 0 or less, so that Bdec_size
When the size-Di becomes equal to or smaller than the threshold K, the basic encoding unit 11 is instructed to perform a skip process, and the basic encoding unit 11 performs the skip process without encoding the data at that time.

The multiplexing unit 10 fetches the data in the intermediate buffer of each video encoding unit at each multiplexing timing.
It is packetized and multiplexed to IEC13818-1 (commonly called MPEG2 TS stream) and output to the transmission path. At this time, the time stamp indicating the playback timing of each video shall be attached according to the occupation amount of the transmission buffer, and the time stamp will be based on the timing at which the data multiplexed when the transmission buffer capacity is full is reproduced without delay time. Add a stamp and send.

The effectiveness of the embodiment operating as described above will be described below. The initial transmission start time of the multiplexing unit 10, the transmission buffer size Bmux_size, and the like should be changed according to the number N of encoding units connected to the multiplexing unit 10, the transmission rate, and the like. Here, Bmux_size
= N × Bdec_size, when data immediately after encoding is multiplexed when the occupancy of the sending buffer is 0 and reaches the decoder
It is assumed that data is reproduced after X pictures (that is, data for a maximum of X pictures may accumulate in the decoder). Since the relative delay time of the data multiplexed with the occupation amount Bmux_oc of the transmission buffer with respect to when Bmux_oc is 0 is calculated by Bmux_oc / R, the data is in the decoder when the data reaches the decoder. The number of picture data is x · (Bmux_size−Bmux_oc (t)) / Bmux_siz
e (however, when the untransmitted data is 0). At this time, it will be described that each video data arrives without underflow on the receiving side.

First, when REMi = 0, (Equation 1) (Equation 3)
Than

[0030]

(Equation 5)

When Bmux_oc approaches Bmux_size, the quantization width increases, the amount of bits generated is suppressed, and no underflow occurs. In addition, the skip processing shown in FIG. 3 works even for extremely complicated video input,
Since _oc is controlled to be smaller than Bmux_size, no underflow occurs.

Next, when REMi ≠ 0,

[0033]

(Equation 6)

Then, REMi is Bdec_size × (Bmux_size−
The encoder is controlled to be smaller than (Bmux_oc) / Bmux_size. In FIG. 3, skip processing is forcibly performed even when an extremely complicated video is input (step 211).
To reduce the bit generation, and REMi is Bdec_size × (Bmu
x_size−Bmux_oc) / Bmux_size. On the other hand, even if all the other encoding units request multiplexing of the maximum data amount, each encoding unit can perform multiplexing at an average rate obtained by dividing at least the transmission rate by the number of encoding units. Bd
ec_size × (Bmux_size−Bmux_oc) / Bmux_size is (Bmx
ux_size−Bmux_oc) / R Since this data corresponds to data that can be sent at the average rate within the time, no underflow occurs on the receiving side.

Data that exceeds the buffer capacity of the decoder is untransmitted data, so that overflow does not occur and the data encoded by each video encoding unit can be correctly reproduced.

In the above configuration, since each video encoder determines the quantization width according to the same standard, if there is no untransmitted data, the quantization width of all encoders is the same regardless of the input video. Therefore, each input video can be encoded with almost the same image quality. Also,
Even when a scene change or the like occurs in a specific input image, since it is controlled by the quantization width, the image quality does not extremely deteriorate at that time, and bits are allocated according to the increase in the difficulty, The bit increment is controlled by the quantization width according to the bit generation amount in the entire video encoding unit, so that the influence is small and the change in image quality is small.
Each video encoding unit controls the bit generation amount and the multiplexed data amount so that the encoded data does not overflow and underflow on the receiving side.
A value of 0 only multiplexes information on all video data regularly without processing the information as in the related art, and the processing is dispersed and the device configuration can be simplified.

In the present embodiment, the formula for calculating the quantization width is (Equation 1), but any other formula may be used as long as Bdec_size-Di> 0 can be maintained.

Further, f (REMi) = REMi may be a different function, and in order to maintain the image quality of a specific video signal higher, the characteristic of determining the quantization width of a specific encoding unit or f (REMi) is intentionally set. (REMi) may be changed. For example, when the capacity of the intermediate buffer is multiplied by M for a specific j-th encoding unit, f (REMi) = REMi / M may be set. By doing so, unmultiplexed data can be increased to an amount close to about M times REMi allowed in the present embodiment. Normally, when a video that is more complex than the average of the entirety is input, x · (Bmux_size−Bmux_oc) that has already been encoded when the data multiplexed at the time of Bmux_oc reaches the decoder.
(T)) / Bmux_size Since the data size of a picture is limited by Bdec_size + REMi, more bits should be allocated to an image that is difficult to encode due to a large amount of untransmitted data. Becomes possible.

The multiplexing cycle in the present embodiment can be controlled more accurately by shortening the cycle. However, several macroblocks in MPEG2 data or
A few slices (several line periods of the video signal) are appropriate, but control may be performed at a longer period such as one picture.

Further, in the present embodiment, the encoding method and the data transmission method of all the video encoding units are the same, but only the processing of a specific basic encoding unit is different from the other standard ones in image quality control. It is also possible to do. For example,
The function of determining the quantization width in (Equation 1) may be any function that increases the quantization width as Di increases, and it is possible to reduce the rate of change of the quantization width to achieve more uniform image quality. However, if most of the video coding units are operated as in the embodiment, the specific video coding unit can fix the quantization width to make the image quality uniform, or perform coding at a fixed rate to convert bits. Even if it occurs, the tendency that the bit generation amount decreases as Di increases becomes dominant, and the entire band can be within the transmission band.

(Embodiment 2) Next, a second embodiment of the present invention will be described with reference to FIGS. FIG. 4 is an explanatory diagram of an image transmission device according to the second embodiment of the present invention.
In the figure, reference numeral 15 denotes a band management node that performs band control on the LAN, and 16, 17, 18, and 19 denote nodes connected to the LAN having an image encoding unit. Node 16,
Reference numerals 17, 18, and 19 each have an image reproducing unit together with the image encoding unit. The image encoding unit performs image encoding by assuming the decoder buffer size Bdec_size and the decoder buffer occupancy of the image reproducing unit of the destination node.

The operation of the above configuration will be described below. In the present embodiment, an image encoding unit is connected to nodes distributed and connected on a local area network (LAN), and transmits an encoded video signal to another node on the LAN. Things. On the network, packetized data is transmitted. The packet includes header information necessary for transmission and transmission data. FIG. 5 shows the header format. As header information, information necessary for correctly transmitting data, that is,
Although necessary information on the LAN considered to be publicly known is also included, here, the sender address, the receiver address, the video packet flag, the transmission cycle No. Will be described only. The network has a ring-like configuration in which one-way transmission paths are connected between the nodes. The generation of the packet is controlled by the band management node. The band is secured in response to a request from each node, and the transmission packet is generated.
FIG. 6 is an explanatory diagram of a communication method between the nodes. When communication is performed from the node 1 to the node 4, the bandwidth management node generates a packet in which 1 is written in the transmission side address and 4 is written in the reception side address, and outputs the packet to the network at an empty packet timing. Each node transmits a packet to an adjacent node without doing anything if the packet is not related to itself. When receiving the packet of the transmission side address 1, the node 1 puts the data to be transmitted in the transmission data and sends it to the adjacent node. Nodes 2 and 3 send packets in order without doing anything. When the node 4 detects its own address as the receiving side address, it receives the transmission data and releases the packet.

Next, communication of video data will be described with reference to FIGS. The bandwidth management node allocates the entire bandwidth of the video data to be transmitted in the network. For example, if there are four nodes transmitting video data in the network and the average rate is about 3 Mbps each, the bandwidth management node secures a 12 Mbps bandwidth as a video bandwidth and sends out 12 Mbps video transmission packets. I do. At this time, a multiplex cycle number is provided to the video transmission packet, and each node is made to declare the number of video data packets to be transmitted within the multiplex cycle number at each fixed time.

FIG. 7 shows a transmission example of the packet number report packet. Both the sending and receiving addresses of the packet are f
f, the video packet flag is set to 1, and each node is identified as a packet number report packet. Then, the number of packets to be transmitted during the period in which the transmission cycle number is displayed is displayed on each node. The bandwidth management node reads the packet that has made a round on the network.

Packets for transmitting the video data with the transmission cycle number are output by the number of declared values. FIG.
5 shows an example of video data transmission by a video transmission packet. The bandwidth management node sets the video packet flag to 1,
It is generated on the network by indicating the transmission cycle number. This packet can be used by any node that has declared video transmission, and if there is data to be sent in the transmission cycle, the sender / receiver address of the packet is written in the header, the video data is loaded, and the video data is sent to the target node. Send. FIG. 8 shows transmission from the node 1 to the node 4.
The node 4 releases the packet after receiving. When the node 1 closest to the bandwidth management node has transmitted all the packets to be transmitted within the transmission cycle indicated in the header, the subsequent packets with the same transmission cycle number are passed as they are. Then, the next closest node 2 transmits the data within the transmission cycle by the video packet in the same manner as the previous node 1. Data to be sent within the transmission cycle is transmitted by video packets in order from the node close to the bandwidth management node in accordance with the number of declared packets. Since the number of packets with the same transmission cycle number has been declared by each node, packets for transmitting data to be sent within the transmission cycle of each node are generated without excess or deficiency. Then, the next multi-period packet is generated according to the declaration from each node.

This multiplex cycle corresponds to the multiplex timing cycle of the first embodiment, and the delay time until the band management node actually generates a packet with the multiplex cycle number is calculated based on the declaration. . For example, if the multiple period is T
If the amount of delay data after k multiplex cycles is DD (kT), DD ((k + 1) T) is the sum of the declared packet number at that time is Pn, and the data amount per packet is P_DATA.

[0047]

(Equation 7)

Is calculated. R is a band for video transmission secured by the band management node. Then, the multiplexed virtual buffer size Bmux_size is calculated from the declared number of nodes N.

[0049]

(Equation 8)

The relative delay parameter delay_mux is

[0051]

(Equation 9)

In the video transmission packet calculated and generated in the above, the packet is placed according to the format shown in FIG.

The delay_init is a delay time from when the transmitted data arrives to when it is reproduced when the delay data amount DD is 0, and is a value shared by each node in the network. The calculated value is immediately transmitted in the header of a subsequently generated packet.

FIG. 9 is a diagram illustrating a packet generated by the bandwidth management node. The packet number report packet is generated at a constant cycle. For video transmission packets, the rate R that secures the number of packets declared by the packet number declaration packet for the entire video secured by the bandwidth management node
Generated by Therefore, even when the number of declared packets cannot be generated within the transmission cycle, a packet with the transmission cycle number is generated as it is until the declared number is reached. However, if the delay data amount DD exceeds delay_init × R, d
Even if the number of declarations is less than the value when elay_init × R is matched, the multiplex number is updated to generate a packet. When the number of report packets has been generated within the multiplex cycle, a packet having an updated multiplex cycle number is generated after the next multiplex cycle. As described above, the video data of the number of packets declared by each node is transmitted within a range where the multiplex delay amount DD does not exceed delay_init × R.

Next, the processing in each node that performs video transmission will be described. FIG. 10 is a configuration diagram of a node. 20
Is a packet processing unit that performs packet input / output processing, 21
Denotes a control unit, 22 denotes an intermediate buffer, 23 denotes an encoding unit, and 24 denotes a video reproduction unit. The packet processing unit 20
It receives a packet sent from the network and detects its header. If the received address is its own address, it sends it to the control unit 21 and resets and sends out the parameters in the packet. If the sender address in the header of the received packet is its own, the control unit 21
Under the above control, the packet is rewritten and transmitted based on the information in the control unit or the intermediate buffer 22. If the header information indicates a packet number report packet, the control unit 21
And a detection signal together with the transmission cycle number. Control unit 21
Sends the report value of the transmission cycle to the packet processing unit 20. The intermediate buffer counts the amount of data sent to the intermediate buffer by the encoding unit 23 for each transmission cycle, and the control unit 21 uses the value as the declared value of the transmission cycle. The packet processing unit 20 writes the report value in a packet and transmits the packet.

When the received data is video data, the control unit 2
1 transmits the received data to the video reproducing unit 24 as it is to reproduce the video.

Next, when the received packet is a video transmission packet, the control unit 21 having received the detection signal from the packet processing unit 20 reads the transmission cycle signal and sets the number of packets transmitted during the transmission cycle to the declared value. If not, the instruction to write the video and the address of the receiver are sent to the packet processing unit 20. The packet processing unit 20 writes its own address in the transmission-side address and the destination address in the reception-side address, writes one packet of video data in the data portion of the packet captured in the intermediate buffer, and sends it out. When all the data within the transmission cycle has been sent out, the control unit 21 updates the transmission cycle number. If the data of the update cycle number of the transmitted packet has already been transmitted, the control unit 2
1 sends a transmission instruction signal to the packet processing unit 20 to cause the packet to be transmitted without any processing. Further, when receiving the video transmission packet, the control unit 21 receives the multiplex virtual buffer size and the relative delay parameter from the packet processing unit 20 and sends them to the encoding unit 23. This is always detected irrespective of the transmission timing, and is immediately sent if it differs from the previously read value.

Next, the operation of the encoding unit 23 will be described. The encoding unit 23 operates substantially the same as the encoding unit in FIG. 2 according to the first embodiment. In the configuration of FIG.
4 is the multiplex virtual buffer size from the control unit 21 of FIG.
Receives Bmux_size and relative delay parameter delay_mux. These operations are replaced with the buffer size Bmux_size of the multiplexing unit 10 and the buffer occupancy Bmux_oc in the multiplexing unit 10 in the first embodiment, respectively, and the same operation as in the first embodiment is performed. At that time, the bandwidth management node 15
Is equal to the transmission rate of the transmission line transmitted by the multiplexing unit 10 of the first embodiment, the data rate generated by the encoding unit on the network is R, and each encoding unit Is transmitted at a timing that can be reproduced without overflow and underflow on the receiving side.

As described above, according to the present embodiment, it is possible to transmit a video signal transmitted over a network with almost uniform image quality. FIG. 11 is a diagram illustrating a comparison of bandwidth utilization between a conventional transmission of variable-rate video data over a network and that according to the present embodiment. In the case of transmission of variable rate data on a conventional network, the image encoding unit secures the band of data encoded according to the difficulty of the image on a transmission path, as shown in FIG. There is a band that cannot be used effectively. According to the present embodiment, encoding is always performed with the entire band, so that the entire band can be used effectively, and each video data is also encoded using a large number of bits, which results in high coding. Image quality can be realized. Also, if the image quality is the same, more images can be transmitted on the network.

In the present embodiment, the form on the network in which the nodes are connected in a ring is shown. However, the rate assigned to the video on the transmission path is guaranteed, and the amount of delay in transmission is reduced by each node. If the network can transmit the data determined by the encoding unit, the same operation can be performed on a network of another shape or a network with a different transmission protocol. The delay time indicated by delay_init can be set arbitrarily within a range of Bmux_size− (delay_init × R) ≧ 0. In particular, when the equal sign holds, the most efficient variable length coding can be performed. Also, as in the first embodiment, by changing the setting of a specific encoding unit, only a specific video can be encoded with high image quality. Further, according to the method of the present invention, it is possible to secure an average rate at least by dividing the entire band by the number of entries by multiplexing the maximum data amount allowed for each transmission cycle. An encoded video signal can be transmitted.

In this embodiment, the case where the number of transmitted images has not changed has been described. However, the number of transmitted images may change in the middle. In that case, Bmux_size
And update the delay_mux value. Also delay
If the _mux value becomes negative due to a change in the number of video transmissions, the delay_mux value is set smaller than the value based on (Equation 9) before adding a new video, and a new video is set. May be added after reducing the amount of delay that does not become negative even if is added.

Each node in the present embodiment is
As shown in FIG. 12, it is also possible to implement the algorithm shown in the present embodiment as software on a general-purpose computer, and realize the image signal captured by a camera or the general-purpose computer. A configuration for encoding a video signal in a storage device such as a hard disk is also conceivable.

[0063]

As described above, according to the present invention, since each encoding unit controls the amount of code and the amount of data to be multiplexed based on the entire amount of encoding, the multiplexing unit uses simple processing. It is only necessary to multiplex the data in a regular manner, and the multiplexing process of video data at a variable rate is distributed to each encoding unit. Therefore, it is not necessary to prepare a processor with large processing capacity in the multiplexing unit. It will be easier. Since each encoding unit encodes based on the same multiplexing information, each encoding unit encodes with substantially the same quantization width, and each multiplexed video signal is divided according to its complexity. Data is evenly allocated, and the image quality of the multiplexed video can be equalized.

Further, even in an unexpected case such as a scene change, a code amount commensurate with the increase in complexity is allocated and the image quality can be maintained.

When the present invention is applied to video transmission over a network such as a LAN, the transmission can be performed by dividing the video data without wasting bandwidth, and the video can be transmitted on multiple channels with high image quality and without interruption. Transmission is possible, and the practical effect of the present invention is great.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an image transmission device according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of an encoding unit in a video encoding unit according to the first embodiment of the present invention.

FIG. 3 is a control unit 14 according to the first embodiment of the present invention.
Of the calculation of the quantization width in the system

FIG. 4 is an explanatory diagram of an image transmission device according to a second embodiment of the present invention.

FIG. 5 is an explanatory diagram of a header format of a packet according to the second embodiment of the present invention.

FIG. 6 is an explanatory diagram of a communication method between each node according to the second embodiment of the present invention.

FIG. 7 is an explanatory diagram of a reaction coefficient control unit according to the second embodiment of the present invention.

FIG. 8 is an explanatory diagram of an example of video data transmission using a video transmission packet in the second embodiment of the present invention.

FIG. 9 is a diagram illustrating a packet generated by a bandwidth management node according to the second embodiment of the present invention.

FIG. 10 is a configuration diagram of a node according to the second embodiment of the present invention;

FIG. 11 is an explanatory diagram of comparison of band utilization between the prior art and the second embodiment.

FIG. 12 is a configuration diagram when the present invention is installed in a general-purpose computer.

FIG. 13 is a configuration diagram of a conventional image transmission device.

[Explanation of symbols]

 1,2,3 Video encoding unit 4,5,6 Encoding unit 7,8,9 Intermediate output buffer 10 Multiplexing unit 10a Reading / transmission control unit 10b Transmission buffer 11 Basic encoding unit 11a Difference processing unit 11b DCT processing unit 11c Quantization processing unit 11d Variable length coding unit 11e Inter-frame prediction unit 12 Bit counter 13 Buffer 14 Control unit 15 Band management node 16, 17, 18, 19 node 20 Packet processing unit 21 Control unit 22 Intermediate buffer 23 Encoding unit 24 video reproducing unit 101, 102, 103 video encoding unit 104 multiplexing unit image rearrangement processing unit

Claims (13)

[Claims] 1. A transmission processing unit comprising: a plurality of image encoding units; and a transmission processing unit, wherein the transmission processing unit multiplexes encoded data specified by each image encoding unit at each multiplex timing and transmits the multiplexed data to a predetermined transmission path. And multiplexing information, which is calculated based on the amount of encoded data of each of the image encoding units and can calculate the amount of delay until transmission to a decoder, is sent to each of the image encoding units. The encoding unit determines a quantization width at the time of quantization in image encoding based on the multiplexing information from the transmission processing unit, encodes the image, and encodes the encoded data for each of the multiplex timings. An image transmission device for transmitting and instructing a transmission processing unit. 2. A transmission processing unit includes a transmission buffer, fetches coded data designated by each image coding unit into a transmission buffer at each multiplex timing, and outputs the coded data according to a predetermined transmission rate in the fetching order. 2. The image transmission apparatus according to claim 1, wherein the occupation amount of the encoded data in the transmission buffer is output as multiplexing information. 3. Each of the image encoding units has a quantization width determination unit and a basic encoding processing unit that encodes an image by performing quantization with a quantization width indicated by the quantization width determination unit. The quantization width determination means calculates the decoder buffer occupancy of the plurality of decoders that receive the coded data transmitted from each of the image coding units through the transmission processing unit from the multiplexing information, and calculates the calculated decoder buffer occupancy. 2. The image transmission apparatus according to claim 1, wherein the quantization width is set based on the amount of encoded data that has not been transmitted after encoding. 4. Each of the image encoding units further includes an encoded data storage unit that temporarily stores encoded data, wherein the encoded data storage unit includes a transmission processing unit of the encoded data for each transmission timing. Temporarily stores the untransmitted data that has not been sent to the decoder, and the decoder occupancy calculated based on the multiplexing information from the upper limit value set in advance by the entire image encoding unit or the decoder buffer size assumed on the receiving side. 2. The image transmission apparatus according to claim 1, wherein the coded data is transmitted to a transmission processing unit using a value obtained by subtracting the amount as an upper limit. 5. A transmission processing device for inputting encoded data from a plurality of image encoding units, comprising a transmission buffer, and transmitting the encoded data designated by each image encoding unit for each multiplex timing. A transmission processing device for outputting the occupied amount of encoded data in a transmission buffer as multiplexed information while outputting the data in accordance with the transmission rate in the order of capture. 6. A quantization width determination unit, comprising: a basic encoding processing unit that encodes an image by performing quantization with a quantization width specified by the quantization width determination unit; Quantization based on multiplexing information that can calculate the amount of delay before transmission to the virtual decoder, virtual decoder buffer occupancy calculated from the multiplexing information, and the amount of encoded data that has not been transmitted after encoding. An image coding apparatus for determining a width. 7. An encoded data storage unit for temporarily storing encoded data, wherein the encoded data storage unit temporarily stores untransmitted data that has not been transmitted among the encoded data at each transmission timing. The upper limit value stored and subtracted from the virtual decoder buffer occupancy calculated based on the multiplexing information from the decoder buffer size assumed on the receiving side or the decoder buffer size assumed on the receiving side. 7. The image encoding apparatus according to claim 6, wherein the encoded data is output as a value. 8. A band management unit for performing band management on a transmission path, and one or more image encoding units, wherein the band management unit determines a transmission rate of the one or more image encoding units on the transmission path. A total bandwidth is determined, transmission of encoded data requested to be transmitted by one image encoding unit is permitted at each transmission timing, and transmission information on a transmission delay that changes according to the permission is transmitted to another image encoding unit on the transmission path. The one or more image encoding units each encode an image by quantizing with the quantization width based on the transmission information, the untransmitted encoded data, and the assumed buffer size of the receiving side. An image transmission apparatus characterized in that the image transmission apparatus comprises: 9. A band management device for managing transmission rates of a plurality of image encoding units, comprising: determining a total bandwidth of transmission rates of all image encoding units on a transmission path; A band management device that permits transmission of encoded data requested to be transmitted by an encoding unit, and transmits transmission information relating to a transmission delay that changes according to the permission to another image encoding unit on a transmission path. 10. Encoding an image by setting a quantization width to be quantized based on transmission information relating to transmission delay that changes according to a transmission permission amount, untransmitted encoded data, and an assumed buffer size on a receiving side. An image encoding device characterized by the above-mentioned. 11. A quantization width determination step, and a basic encoding processing step of encoding an image by quantizing with a quantization width indicated by the quantization width determination step, wherein the quantization width determination step is performed. The quantization width is calculated based on multiplexed information from which the amount of delay until transmission to the virtual decoder can be calculated, the virtual decoder buffer occupancy calculated from the multiplexed information, and the amount of coded data that has not been transmitted after encoding. An image encoding method characterized by setting. 12. At each transmission timing, untransmitted data that has not been transmitted among encoded data is temporarily stored, and an upper limit value previously set in the image encoding device or assumed on the receiving side. 12. The image encoding method according to claim 11, wherein the encoded data is output with a value obtained by subtracting a decoder occupancy calculated based on multiplexing information from a decoder buffer size as an upper limit value. 13. Encoding an image by setting a quantization width for quantization based on transmission information relating to a transmission delay that changes according to a transmission permission amount, untransmitted encoded data, and an assumed buffer size on a receiving side. An image encoding method characterized by the following.