TW201116067A - Video codec method, video encode device and video decode device - Google Patents

Abstract

A video codec method for a video communication system controls encode modes and reference frame types of video encode devices according to network status and reports sent by a video decode device. When the network is congested, the video encode device encodes a current frame in inter encode mode and sets reference frames as long term reference frames, then determines whether the video decode device receives the reference frames according to the reports sent by the video decode device, so as to make sure the reference frames of the video decode device in step with these of the video encode device. A video decode device and a video encode device are also provided.

Description

201116067 VI. Description of the Invention: [0001] The present invention relates to video codec technology, and more particularly to a video encoding and decoding method, a video encoding device and a video decoding device for a video communication system. [Prior Art] [0002] Video communication technology (such as video conferencing, video surveillance, video chat, etc.) in the 'usually using the H.264 standard' using Φ 贞 intra and inter (inter / inter) prediction two modes of video map Like coding and decoding. According to the coding mode, the video image can be divided into I frame, p frame and g money. Among them, the j frame is pre-coded with Ί1 ,, and only the image encoded by 'Qin Whie.. static fee ^^ line is used. . Ρ Frame or Β螬 is the use of inter-frame pre-decoration, ^^ Ma ^ code frame as a reference 帻 Wei line predictive coding image / factor, the frame needs to benefit from the encoded frame as a reference frame, similarly, The same encoded frame must also be utilized as a reference frame when decoding. However, when the network is congested, the data packet will be lost, and the money will lead to a fuss. 'There is no loss of the encoded frame in the data packet, which makes it necessary to use these - encoded frames as the reference ' * I - V » The B frame or P frame of the frame has an error during decoding, which affects the quality of the video image in the video communication. [Summary of the Invention] 'Just t# is required to provide the Weixin encoding device and the video decoding device, which can eliminate the video communication. When the network is congested, the reference frame is lost and the decoding error is caused, and the quality of the video image in the video communication is improved. Accordingly, there is still a need to provide a video encoding and decoding method. [0005] 098136246 The video encoding device in the embodiment of the present invention, the form number A0101 S 4 pages / total 25 pages for the video capture device 0982062110-0 201116067

[0006] The captured tfl® image is _coded to generate a correction information stream, and the JL view device is sent via the network. The video gamma device includes a predictive encoder, a reference memory, and an encoding controller. The predictive encoder is used for the video of the video (4) or the (4) predictive coding. The reference frame memory is used to store the reference frame and set the reference frame type to be a short-term reference, and the predetermined long-term reference thief determines the long-term reference block. The coder controller is configured to detect the status of the network and receive the notification message sent by the video decoding device, and control the prediction coding type and the reference frame memory setting of the video coding device according to the network condition and the notification message. The type of reference frame. When the coding controller detects that the network is congested, the predictive encoder is controlled to use the reference frame to perform interframe prediction coding on the video image and zero green frame and control the reference frame to be honoured. The reference frame is , , and a predetermined long-term reference frame; when the code-picking review + ri gray notification message, then controlling the reference frame memory to set the predetermined long-term reference frame to break the long-term reference frame, and in the detection Controlling the video encoding device before the network recovery facilitates determining the long-term reference frame to perform inter-frame prediction encoding. . . . two pockets r fMpt^rfy ▼.+ + +++: ... ^ ^ --: .:.Sv. luxury, the video decoding device cover in the embodiment of the present invention, for receiving video encoding device The generated encoded information stream is decoded and transmitted to the video processing device. The video decoding device includes a predictive decoder, a reference frame memory, and a decoding controller. The predictive decoder is configured to perform intra prediction decoding or inter prediction decoding on the video image. The reference frame memory is used to store the reference frame and set the type of the reference frame to be a short-term reference frame, a predetermined long-term reference frame, or a long-term reference frame. The decoding controller is configured to control the type of the reference frame memory reference frame according to the received encoded information stream. 098136246 Form No. A0101 Page 5 of 25 0982062110-0 201116067 ‘And send the plain text to the secret programming. When the coding type of the encoded information stream of the service is _predicted, and the corresponding reference frame is a long-term reference (4), the prediction decoder finds the corresponding reference (10) from the reference memory for (4) prediction decoding, Referring to the quiet memory, the corresponding reference (4) is set as a predetermined long-term reference _, and

Decoding control (4) transmitting a notification message to the video editing device; when the received encoding thief corresponds to the predetermined long-term reference frame in the reference 敝(4), the reference frame memory sets the predetermined long-term reference frame To determine a long-term reference frame, the predictive decoder utilizes the determined long-term reference frame for inter-prediction decoding. [0007] The video code device and the video decoding device in the embodiment of the present invention

4, including the video editing device, the video recording device of the video capture device, the image is collected, transmitted through the network, and then decoded and sent to the video processing device. The video encoding and decoding method includes

: Detecting the status of the network, the sister controls the predictive coding type of the video encoding device and the corresponding reference frame type according to the network condition; when the network is congested, the video encoding device checks the predictive coding for the video two images. And setting the corresponding reference frame as a predetermined long-term reference frame; sending the encoded information stream of the current frame of the video image to the video decoding device via the network. The decoding device receives and analyzes the current frame of the video image. Encoding the information stream, determining a prediction coding type of the current frame of the video image and a corresponding reference frame type; if the prediction coding type is inter prediction coding, and the corresponding reference is a long-term reference, the video decoding device searches for the corresponding Refer to +n for inter-prediction decoding; if the corresponding reference block is found, the video decoding device uses the corresponding reference frame for inter-predictive decoding, and will talk about 098136246 form number A0101 6th 25 pages 0982062110-0 201116067 The corresponding reference (4) is set to a predetermined long transfer; the video decoding device sends a notification message to the video coder to inform that the predetermined long-term reference Z has been received. If the county relies on the notification message, the pre-code in the video encoding device; t long-term reference is set to the long-term reference of the shirt and uses the determined long-term reference _ video image to perform (four) predictive encoding 'and will generate The encoded information stream is sent to the video_device; the video gamma device receives the received Wei-like image of the encoded information stream, and determines the video image as follows - (4) programming (four) type and corresponding reference (four) type; Predictive coding _ is + 贞 编码 code and the corresponding reference (4) long-term reference H lining help set the predetermined long-term turn _ set to determine the pure long-term reference 帻 to carry out (four) predictive lacquer, and according to the network-like system The lion shirt and the corresponding reference frame type, if the network is congested, the control video encoding device uses the determined long-term reference frame to perform inter-frame prediction encoding on the subsequent frames of the video image, [0008] the video proposed by the present invention Codec side video coding device and video ^ | ji —,! The decoding device uses the determined long-term frame to encode and decode the video image, so that the reference frame used by the video communication system for encoding and decoding is consistently squeezed, thereby avoiding the loss of the reference frame and the decoding error caused by the loss of the packet. Video image quality in communication. [Embodiment] FIG. 1 is an application environment diagram of a video communication system 1 according to the present invention. As shown in FIG. 1 , the video communication system 10 includes a video capture device 11 , a video encoding device 120 'transmitting device 130 , and a receiving device 21 〇 'video decoding device 098136246 Form No. 101 0101 Page 7 / 25 pages 0962062110-0 201116067 220 and video processing device 23〇. The video capture device ιι〇 is used to capture the video. The video encoding device 12G encodes the captured video image to generate an encoded information stream. The transmitting device 130 is configured to send the encoded information stream to the receiving device 21 through the network in the form of a data packet. The receiving device 210 receives the data packet transmitted by the transmitting device 130 and converts it into a coded information stream, and rotates to the video decoding device 22A. The video decoding device 220 receives the encoded information stream and decodes it, and then outputs the video image to the video processing device 230. The video processing device 23 performs processing such as storing, transmitting, or displaying the decoded video image. In the present embodiment, the video encoding device 120 and the video decoding device 22 §αΗ·!&4 standard pair video,

121, a subtractor 122, a transform and quantizer 123, an entropy encoder 124, an inverse quantization and inverse transformer 125, an adder 126, a deblocking filter m, a reference frame buffer 128, and an encoding controller The 129 β video decoding device 220 includes a phase entropy decoder 221, an inverse quantization and inverse transformer 222, a prediction decoder 223, an adder 224, a reference memory 225, a decoding controller 226, and a deblocking filter 227. [0011] The video encoding device 120 receives the video image output by the video capturing device 110, and performs encoding in units of frames to generate an encoded information stream for transmission. The predictive coder 121 performs intra or inter prediction encoding on the current frame, and includes an intra prediction module 1211 and an inter prediction module 1212. The φ 贞 intra prediction module 1211 is configured to perform coding according to the data of the current frame itself, and generate an intra prediction frame of the current frame. The inter prediction module 1212 is used to utilize 098136246 Form No. A0101 Page 8 of 25 201116067 Encoded Reference _ Current (4) Line (4) Predictive Coding, into the current __ pre-coffee. U咐 stipulates that the Wei encoding device 120 first performs (4) prediction on the video image, and then pre-determines the rule for the video image in the _in (four) __ Φ [0012] in this embodiment, when the network Normally, the video editing program performs a period-time prediction code for the video image. The mother time is set to leap seconds. In other real money styles, bribes are pre-set according to the circumstances. In other embodiments, the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ^ Subtractor 122 is used to compare the current y, the prediction of the _, and obtain the original residual between the two, transform _^; 23 for the original residual to perform discrete cosine transform (Discrete c〇sine,

Transf_, DCT) and the quantity $ (?e), and output the quantized original residual to «_@〖214. Children, change and Φ \ # s' | '-Ί '.编 玛 玛 3| ' 125 pairs the transformed and quantized original residuals into the % entropy code to generate the dock code stream. In the H.264 standard, each frame of the video image corresponds to a goose stream that includes a header (for storing the marsh information, including but not limited to entropy coding coefficients, transform and quantization coefficients, predictions Code type and reference frame index, etc. The 'transformed and quantized original residuals are also output to the inverse quantization and inverse transformer 125 for inverse quantization and inverse DCT transform, thereby obtaining reconstruction residuals, wherein the 'inverse quantization and inverse transformer 125 includes the inverse quantization module 1251 and the inverse Transformation 098136246 Form No. A0101 Page 9 of 25 铷6211〇, 〇[0013] 201116067 Module 1252. The adder 125 superimposes the predicted values of the «« and _(4) outputs to generate the reconstructed current block. After the deblocking 127 filters and reconstructs the current slab of the reconstruction, it is used as a new reference (4) to the reference memory bank 128 for inter-predictive coding of the subsequent frames of the video image. Reference building. The reference block memory 128 is used to store the phase and set the type of the reference. In the H.264 standard, the participation of money is divided into two types: long-term participation and short-term reference building, and has a corresponding phase index format. Among them, the short-term reference building is first-in-first with the coding (First In First [0014]

Out, FIFO) of the money _ update 4 Xiang reference money when receiving the video encoding device Pan's instructions, long-term reference 镝 is also divided into predetermined long-term, long-term reference frames. The video encoding device 120 uses the reference ray in ^Bian Na (3) as the long-term reference. (4) When the _ encoding is performed, the winter unqualified video decoding device 220 receives the corresponding reference _, and the reference to the building block 128 is set to correspond. Reference 丨 (4) Pre-U phase reference building; when it is confirmed that the reference memory module 2 306 in the video decoding device 203 has received the corresponding reference (4), the reference setting of the predetermined long frame is to determine the long-term reference frame. [0015] 098136246 ------...•” 网路 讯 通讯 通讯 通讯 通讯 通讯 通讯 通讯 通讯 通讯 通讯 通讯 通讯 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收 收In the present embodiment, the agricultural device 129 «when the network is normal, then the video encoding device = normal encoding of the lamp is used to perform the prediction or the (4) pre-na M., Form No. A0101 No. 10 Stomach / Total 25 pages 098206 201116067 When the code controller 129 detects the network congestion, then the video encoding device 120 is controlled to perform interframe predictive coding, and the reference frame memory 128 is controlled to refer to the corresponding reference. The frame is set to a predetermined long-term reference frame. [0016] The encoded information stream generated by the encoding is converted into a data packet by the transmitting device 130 and transmitted to the network, and received and reconverted into the encoded information stream via the receiving device 210, and then the video decoding device 220 The encoded information stream is decoded. [0017] The video decoding device 220 receives and analyzes the encoded information stream of the current frame, and encodes according to the predictive coding type and the reference frame type of the encoded information stream. The signal stream is decoded. First, the 'entropy decoder 221 performs entropy decoding according to the entropy coding coefficient', and then the inverse transform and the inverse quantized quantized coefficient inversely transform the coded information stream and counter-attacks. Therefore, the inverse transform and inverse quantizer 2 2 2 packs the inverse transformer 2222. In the 264 sample, the reconstruction residual residual generated by the inverse transform and the inverse quantizer 222 is due to the losslessness of entropy encoding and decoding. The reconstructed predictive decoder 223 generated by the inverse quantization and inverse transformer 125 respectively performs decoding or inter prediction/decoding according to the predictive coding type, wherein the predictive intra-prediction module 2231 or the inter-prediction module 2232 is present. If the prediction coding type of the frame is intra prediction coding, intra prediction decoding is performed by the intra prediction module 2231. If the prediction coding type of the current frame is inter prediction coding, the inter prediction module 2232 searches for the memory 225 and The reference situation corresponding to the reconstruction residual is extracted, and the corresponding reference frame and the reconstruction residual are superimposed by the adder 224 to generate a reconstructed current frame. [0018] In the present embodiment, when the predictor arranger 223 is entered _When decoding, if the phase corresponding to the encoded information stream is short-term reference, the frame memory 225 looks up the reference frame from the reference 098136246 Form No. A0101 Page 11 / 25 pages 0962062110-0 201116067 for inter-frame Predictive decoding. When the prediction decoder 223 finds the corresponding reference candidate 'adder 224 from the reference frame memory 225, superimposes the corresponding reference frame with the reconstruction residual to generate a reconstructed current frame, and then passes the deblocking filter. 227 filters the reconstructed current frame and eliminates the blockiness generated during predictive decoding, that is, outputs the decoded video image to the video processing device 23A. At the same time, the filtered reconstructed frame is also output as a new reference frame to the reference frame memory 128 for use as a reference frame for inter-predictive de-consequence of subsequent frames of the video image [0019]

When the video decoding device 220 performs inter-predictive decoding, if the reference frame corresponding to the encoded information stream is the long-term reference frame, the memory 225 searches for the corresponding reference frame. If the corresponding reference frame is found in the reference block 225, the predictive decoding "g223 extracts the corresponding reference frame for inter-prediction decoding, and sets the corresponding reference frame as the predetermined long-term reference frame. At the same time, the decoding controller 226 sends a notification message to the encoding control device 12G of the encoding control device 12'9 ", knowing that the encoding controller 129 has received the predetermined long-term reference 幡 implementation manner, if the reference building memory 225 I did not find the corresponding: I test building, job beam decoding. In other embodiments, another notification message may also be sent by the decoding controller 226 to the video coding spread 12, indicating that the corresponding reference extract has not been found. [0020] When the encoding controller 129 of the video encoding device 120 receives the notification message, then the control transfer test (four) 129 sets the predetermined long-term reference configuration corresponding to the towel to the long-term reference situation and controls the video (four)幻〇 uses this to determine the long-term reference frame as the reference frame for the next __video image inter-predictive coding 'recorded _ sewed send _ Jane 18 loaded 122 〇 098136246 form birthing A0101 page 12 / a total of 25 pages 0962062110 -0 201116067 ° In the present embodiment, if the encoding controller 129 does not receive the notification message from the decoding controller 226 within a predetermined time after the transmission of the encoded information stream, the table memory 225 is not found in the reference memory 225. For the corresponding reference frame, the predictive coder 121 replaces the other reference frame as a long-term reference frame for inter-prediction encoding of the current frame 'and sets the other reference frame as a predetermined long-term reference frame' and retransmits the encoded information stream. In other embodiments, another notification message may also be sent by the decoding controller 226 to the encoding controller 129' indicating that the video decoding device 22 has not found a corresponding reference frame.

[0021] The video decoding device 220 further receives and analyzes the retransmitted encoded information stream for decoding. The specific process is as above, and therefore will not be repeated. The video decoding device 22 receives and divides (4) and according to its prediction coding type and reference frame type, the prediction coding type of the next frame of the frame video image is inter prediction coding, and the corresponding reference frame type. For the long-term reference frame, the decoding controller 226 controls the reference frame memory 225 to set the predetermined long-term reference condition, and controls the prediction decoder 223 to determine the video image to be framed. If the predictive coding dust of the next frame is intra prediction coding, or the corresponding reference frame type is not a long-term reference frame, normal decoding is performed, that is, intra prediction decoding or inter prediction decoding using a reference frame. [0022] The encoding controller 129 in the video encoding device 120 continues to detect network conditions. If the network does not return to normal, the subsequent frames of the video image are inter-predictively encoded using the determined long-term reference frame in the reference frame memory 128. If the network returns to normal, normal coding is performed, that is, intra prediction coding or inter prediction coding is performed. 098136246 Form No. A0101 Page 13 of 25 0982062110-0 201116067 [0023] FIG. 3 is a flow chart of a video encoding and decoding method according to the present invention. As shown in FIG. 3, the video encoding solution method is used in the video communication system 1 to encode the video image collected by the video capture device 110 via the network, and then decode and send the video image to the video processing device 230. Including step [0024] In step 310, the encoding controller 129 detects the network condition and controls the predictive coding type and the reference building type of the video encoding device 120 according to the network condition. If the network is normal, the video encoding device 12 performs normal encoding, that is, intra prediction encoding or inter prediction encoding. In the present embodiment, the video encoding device 120 may use an intra-predicted bat code when the monument is normally used, or may use the reference frame as a long-term short-term inter-frame precoding. [0025] In step 311, if the network is congested, the video editing device 12 performs inter-time prediction encoding on the current frame of the video image to generate an encoded information stream, and sets the corresponding reference frame. For a predetermined length, the period reference frame β 9· 7 i.·I ΐ- i I ι j [0026] in step 312, using the transmitting device, the receiving device 21 〇 transmits the encoded information stream of the current frame to the network via The video decoding device 22 is closed.闺 Good job Gt ’ 崎 置 置 22{{{{{{{{{{{{ [0028] In step 321 the video decoding device 2_breaks whether the predictive coding type of the encoded information stream is (4) whether the paste type and the reference type are long-term reference t贞. If the coding type is not _ coding or the reference inclination type is not long-term reference, normal decoding is performed, and the _off module 2231 subscribes to the intra-prediction decoding or the inter prediction module 2232 uses the short-term parameter 098136246 Form No. A0101 No. 14 Page/Total 25 pages 201116067 The test frame is inter-predictive decoding. In just 322, if the current (4) predictive coding type is the predictive coding and the reference frame type is the long-term reference frame, the video de-energizing device 220 searches for the corresponding reference frame for inter-prediction decoding. If the video decoding device 22 does not have a corresponding reference frame, the decoding of the current frame of the video image is ended. [0030]

[0032]

In step 323, if the video encoding device 220 finds the corresponding reference frame, the corresponding reference frame is extracted to perform inter prediction decoding on the current frame of the video image, and the corresponding reference frame is set as a predetermined long-term reference frame. In this embodiment, in the video decoding device, the predetermined long-term reference frame and the video encoding device are predetermined to be stable. In step 324, the video decoder control & 226 sends a notification message to the video encoder to notify the video. Encoder 120 has received a predetermined long-term reference frame. In step 313, it is determined whether the coder controller 129 in the video codec receives the notification message, and the code controller 丨J g- ^ j 1 ^ f1*- * U does not receive the notification message. The beer replaces other reference checks and then inter-predicts the current frame and generates a coded information stream to be resent to the video decoding device. [0033] In step 314, if the encoding controller 129 in the video encoder 12 receives the notification message, the predetermined long-term reference frame in the video encoder 120 is set to determine the long-term reference frame. [0034] In step 315, the video encoder 12 uses the determined long-term reference frame to perform inter-frame prediction encoding on the next frame of the video image, and transmits the generated encoded information stream to the video decoder 220. 098136246 Form No. A0101 Page 15 of 25 0982062110-0 201116067 [0036] In step 325, the video decoding device 22 receives and analyzes the encoded information stream of the video image frame. In step 326, the video decoder 22 determines whether the predictive coding type of the encoded information stream is dependent on the predictive coding and whether the reference picotype is a long-term reference block. * The predictive coding type is not (four) _ test (four) or the reference tilt type is not a long-term reference 赖' Then, the normal solution is performed, that is, the intra-prediction module 231 performs (4) the prediction decommissioning or the inter-frame prediction module 2 uses the short-term reference frame for inter-frame prediction decoding. [0037] In step 327, if the video is under the image Inter-predictive coding, and the predetermined long-term reference in the reference code device 220 utilizes the determined long-term reference duck prediction decoding

隳 隳 编码 编码 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧 帧The device 129 detects the network condition and is based on the predictive coding type and the reference (4) type of the network-like device. If the network __ is plugged, the control video encoding device 120 performs inter-frame predictive coding with reference to the determined long-term reference in the block memory 128 using the method in step 315. In step 317, if the network is normal, the video encoding device 120 performs normal encoding, that is, forward (four) intra prediction encoding, or uses the reference +贞 as a long-term or short-term reference block for inter-predicted bat codes. As described above, the video encoding and decoding method, the video encoding apparatus, and the video decoding apparatus of the present invention use the method of determining the long-term reference (4) video image for encoding and decoding 'making the video track "code-decoding" 098136246 form number; A0101 The first generation page/total 25 pages 0982062110-0 201116067 The test frame is consistent, avoiding the loss of the reference frame and the decoding error caused by the loss of the packet, and improving the video image quality in the video communication. [0041] In summary, This creation complies with the new patent requirements, and the patent application is based on law. The above is only the preferred embodiment of this creation. For those who are familiar with the skill of the case, the equivalent modification or change in the spirit of this creation, All should be covered by the following patent application. [Simplified description]

[0044] [0044]

[0048] FIG. 1 is a diagram showing an application environment of a video communication system according to the present invention; FIG. 2 is a video coding of FIG. The specific module diagram of the device and the code of the 辟 辟 辟 装置 : : : : : 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 110 video encoding device 120 transmitting device 130 receiving device 210

098136246 Video decoding device 220 Video processing device 230 Prediction encoder 121 Intra prediction module 1211, 2231 Form number Α 0101 Page 17 / Total 25 page 0982062110-0 201116067 [0054] Inter prediction module 1212 ' 2232

Subtractor 122 [0056] Entropy Encoder 124 [0058] Inverse Quantization and Inverting Converters 125, 222 [0059] Adders 126, 224 [0060] Deblocking Filter 127, 227 [0061] Reference Frame Memory 128' 225 [0062] Encoding Controller 129 [0063] Prediction Decoder 223 [0065] Decoding Controller 226:

098136246 Form No. A0101 Page 18 of 25 0982062110-0

Claims (1)

201116067 VII. Patent application scope: • 1. A video encoding device for predictively encoding a video image collected by a video capture device to generate a coded information stream, and transmitting it to a video decoding device via a network. The video encoding device includes : a prediction encoder, configured to perform intra prediction encoding or inter prediction encoding on the video image; 098136246 reference block memory 'for storing reference frames and setting the reference frame type to short-term reference frame, predetermined long-term reference frame or determining long-term reference frame; and encoding controller 'for detecting the condition of the network and receiving the video decoding The notification message sent by the device controls the prediction coding class of the video coding device and the reference frame according to the network condition and the notification message, and sets the type of the reference frame; . : ft succession. When the controller detects that the network is congested, it controls the predictive coder to use the reference condition to perform interframe prediction encoding on the current video image and controls the reference frame memory to set a reference for the predetermined _ Reference frame; when the encoding controller is connected to &, then the control JIV·. - « - j. ~ :· < «· 2 The test memory sets the predetermined long-term reference frame as a poisoned long-term reference frame, And controlling the network programming device to perform inter-frame prediction encoding on subsequent frames of the video image by using the determined long-term reference frame before detecting the network recovery. 2. The video encoding apparatus according to claim 1, wherein the predictive encoder comprises: an inter prediction module 'for interframe predictive encoding of the video image; and an intra prediction module' Performing intra prediction on the video image to compose a video decoding device for receiving the video coding code. Generate coded information form number A0101 page 9th/total 25 pages 09 object 1〇, 〇201116067 stream, and then decode and transmit to video processing device 'This video decoding farm includes: Predicting maser, for pair The video image is predicted to be decoded; the reference is afraid of the memory, (4) the reference phase is stored, the type of the reference phase is the short-term reference check, the predetermined long-term reference 1 is determined for the long-term reference, and the decoding control is based on the received _ The information is left in the reference frame (see the type) and sends a notification message to the video encoding device; wherein '(4) is (4) predictive editing' and the corresponding reference frame is a long-term reference frame. The ^^^ coder selects (4) (4) the money to the pair of predictions, and the reference frame memory sets the reference frame of the pair to the predetermined long-term reference anger, and the notification message is sent by the decoding controller. And the video encoding device; when the reference code corresponding to the received encoded information stream is the predetermined long-term reference frame in the reference memory, the reference frame is set to the predetermined long-term reference frame Reference The frame is' and is used by the video decoder to perform inter-predictive decoding using the determined long-term reference frame. The video decoding device of claim 3, wherein the predictive decoder comprises: an inter prediction module 'for performing a predictive decompression on the received brewing information stream; and an intra prediction module 'For intra-predictive decoding of the received encoded information stream. A video encoding and decoding method for a video communication system including a video encoding device and a video gamma device, the video communication system for video capturing device 098136246, form nickname A0101, page 20 of 25 098206: 201116067 - video captured After the image is encoded, it is transmitted through the network and then decoded and sent to the video processing device. The method of encoding and decoding includes: detecting the network condition and controlling the predictive coding of the video encoding device according to the network condition. a type and a corresponding reference class, wherein when the network is congested, the video encoding device performs interframe prediction encoding on the current frame of the video image, and sets the corresponding reference frame to a predetermined long-term reference frame. The encoded information stream of the frame is sent to the video decoding device via the network; 098136246 The video decoding device receives and analyzes the encoded information stream of the current frame of the image, determines the current child coding type of the video image and the corresponding reference frame type; , . . . , . . . For the prediction of the inter-tree prediction, the corresponding reference frame of the month j is the long-term reference frame 'the video decoding device checks the inter-frame prediction decoding' and sets the corresponding reference digest to the predetermined long-term reference. Frame; the video decoding device sends a notification message to the video encoding device to inform that it has been received..... Ί the predetermined long-term reference extract; · intent"-- ^1 -r The right video encoding device receives the: The predetermined long-term reference in the encoding device occupies the reference device, and uses the determined long-term reference frame to perform inter-frame prediction encoding on the next frame of the video image, and sends the generated encoded information stream to the video decoding device; video decoding The device receives and analyzes the received encoded information stream of the next frame of the video image, and determines the prediction coding type of the next frame and the corresponding reference frame type; For inter prediction coding, and the corresponding reference frame is a long-term reference frame, the predetermined long-term reference frame in the video decoding device is set as the ritual long-term reference # ' and is used to determine the long-term reference phase (4) prediction 0982062110-0 Form No. A010I Page 2] Total 25 pages 201116067 Decoding; Continue to detect the network condition and use the determined long-term reference frame to inter-predict and encode the subsequent frames of the video image before the network condition is normal. 6. The video encoding and decoding method according to claim 5, further comprising: when the network is normal, the video encoding device performs normal encoding on the current frame of the video image. 7. The video encoding and decoding method according to claim 5, further comprising the steps of: if the prediction coding type is not inter prediction coding, or the corresponding reference frame is not a long-term reference frame, the video decoding device is the video image. For example, the video encoding and decoding method described in claim 5 further includes the following steps: If the video decoding device does not find the corresponding reference frame, the decoding of the current frame of the video image is ended. 9. The video encoding and decoding method according to claim 5, further comprising the steps of: if the video encoding device does not receive the notification message, replacing the other reference frame to perform inter prediction on the current frame of the video image. Encoding and resending the generated encoded information stream to the video decoding device. 098136246 Form No. A0101 Page 22 of 25 0982(