Image coding and transmitting apparatus

Claim 2

An average value predicting motion compensation operator, wherein input digital image signals are divided into blocks of prescribed size, and coding operation is performed to differential signals between the block of input signals and interframe predicting signals produced block by block, a circuit for producing a block of predicting signals is constituted by input signal blocks and previous frame blocks disposed on the same position on the previous frame as the input signal blocks, the blocks on the previous frame are given according to L sorts (L: positive integer) of displacement (hereinafter referred to as "motion vector"), and L blocks of the previous frame and average value of the input block is quantized into one of the M levels (M: positive integer), further are quantized average value is formed as the value of picture element within the block and total (L + 1) blocks are defined as predicting signals blocks (hereinafter referred to as "reference block"), whereby the block pattern matching between the input signal block and the reference blocks is estimated by prescribed measurement method and the block being most matched with the input signal block is picked up from the reference blocks, and the picked up block signal is outputted as the predicting signal block, and if the picked up block is selected from L blocks of the previous frame blocks, the index of the picked up block corresponding to the motion vector is outputted, and if the average value block is selected the index of the picked up block corresponding to the quantized level of the average value is outputted.

Claim 3

An average value predicting motion compensation operator as set forth in claim 1, comprising: a differential absolute value operator which performs pixel by pixel differential absolute value process between a block of input signals and a block of signals disposed on the same position or near the position of the previous frame or a block of samples with the quantized average value as hereinafter described; an accumulator which performs distortion measure by accumulation of the differential absolute value and average value for each block by summalizing the input signals within the block; a quantizer which quantizes the average value into one of corresponding level of the matching distortion; a comparator which compares the matching distortion calculated with the accumulator between input signals and the signals on the same position or near the position on previous frame or with the quantized average value, and outputs the block of signals on the previous frame to give the minimum distortion or index corresponding to the quantization average value; and a selector which supplies the differential absolute value between the block of input signals and the blocks of signals on the previous frame or the block of input signal to the accumulator.

Claim 4

An average value predicting motion compensation operator as set forth in claim 1 or claim 2, wherein in place of the differential absolute value operator, a differential square value operator is used, and difference square value of each picture element is calculated between the input signals and the reference blocks.

Claim 5

An average value predicting motion compensation operator as set forth in claim 1 or claim 2, wherein in the accumulator, the matching distortion of each reference block is supplied with different weighting function thereby the specific reference block as that giving the minimum distortion tends to be selected or not.

Claim 6

An image signal progressive build-up encoder/decoder wherein in case of one input image signal system, the input signal system is performed high efficiently coding and decoding, and obtained reconstructed image signal system is delayed by one frame, and the residual signal system between the input image signal system and the delayed signal system is again performed high efficiently coding and decoding frame by frame and operation is repeatedly executed thereby high quality image is performed progressive build-up in time progresses, said encoder/decoder comprising: an average value encoder/decoder which calculates average value within block of residual signal vectors obtained by forming the block of the residual signal system per lattice-shaped m x n (m, n: positive integer) sample, and performs high efficiently coding of the average value; a normalized vector quantization encoder/decoder which utilizes the residual signal vector as input vector and subtracts the average value from each element of the input vector and obtains the normalized output vector having the maximum inner product value to the input vector amount a set of normalized output vectors with average value 0, amplitude 1 generated so that total of distortion is minimized based on the distribution of normalized input vectors of which amplitude becomes 1, and obtains index to indicate the normalized output vector, and also performs high efficiently coding and decoding of the amplitude gain of the input vector given by the maximum inner product value, and the decoded amplitude gain is multiplied by the detected normalized output vector thereby the amplitude reconstructed output vector can be obtained; a sample value encoder/decoder which performs high efficiently coding and decoding of m x n block of samples in the residual signal vector sample by sample; a switch which adaptively controls ON/OFF operation of the average value encoder-decoder, the normalized vector quantization encoder/decoder and the sample value encoder/decoder based on the control signal frame by frame; a vector reconstructor which selects adaptively the decoded average value, the decoded residual signal vector by adding the decoded average value to the amplitude reconstructed output vector, and the decoded sample value based on the hereinafter described frame by frame control signal, ant then outputs them as the decoded vector; and a coding controller which generates the control signal and enables the adaptive control operation of the switch, each encoder/decoder and the vector reconstructor corresponding to the number of coded samples each frame.

Claim 7

An image signal progressive build-up encoder/decoder wherein in case of one input image signal system, the input signal system is performed high efficiently coding and decoding, and obtained reconstructed image signal system is delayed by one frame, and the residual signal system between the input image signal system and the delayed signal system is again performed high efficiently coding and decoding frame by frame and operation is repeatedly executed thereby high quality image is performed progressive build-up in time progresses, said encoder/decoder comprising: an average value encoder/decoder which calculates average value within block of residual signal vectors obtained by forming the block of the residual signal system per lattice-shaped m x n (m, n: positive integer) sample, and performs high efficiently coding of the average value; a normalized vector quantization encoder/decoder which utilizes the residual signal vector as input vector and subtracts the average value from each element of the input vector and obtains the normalized output vector having the maximum inner product value to the input vector amount a set of normalized output vectors with average value 0, amplitude 1 generated so that total of distortion is minimized based on the distribution of normalized input vectors of which amplitude becomes 1, and obtains index to indicate the normalized output vector, and also performs high efficiently coding and decoding of the amplitude gain of the input vector given by the maximum inner product value, and the decoded amplitude gain is multiplied by the detected normalized output vector thereby the amplitude reconstructed output vector can be obtained; a sample value encoder/decoder which performs high efficiently coding and decoding of m x n block of samples in the residual signal vector sample by sample; a switch which adaptively controls ON/OFF operation of the average value encoder-decoder, the normalized vector quantization encoder/decoder and the sample value encoder/decoder based on the control signal frame by frame; a vector reconstructor which selects adaptively the decoded average value, the decoded residual signal vector by adding the decoded average value to the amplitude reconstructed output vector, and the decoded sample value based on the hereinafter described frame by frame control signal, ant then outputs them as the decoded vector; and a coding controller which generates the control signal and enables the adaptive control operation of the switch, each encoder/decoder and the vector reconstructor corresponding to the number of coded samples each frame.

Claim 8

An image signal progressive build-up encoder/decoder as set forth in claim 5, wherein in the coding controller, at the first coding stage, only the average value encoder/decoder are turned on, and rough image is reconstructed by the decoded vector which all m x n samples are substituted by the decoded average value, and at the second coding stage, the average value encoder/decoder and the normalized vector quantization encoder / decoder are turned on, and the decoded vector substituted by the decoded residual signal vector is obtained, and the image of which quality is better than that of the first stage can be built up, and at the third coding stage or later, only the sample value encode/decoder is turned on, the decoding vector comprising the decoding value of m x n pixels of sample value is obtained, and the image of which quality is better than that of the second stage can be built up.

Claim 9

An image signal progressive build-up encoder/decoder as set forth in claim 5 or claim 6, wherein in the average value encoder/decoder, the normalized vector quantization encoder/decoder and the sample value encoder/decoder, a set of different coding and decoding control modes corresponding to various encoded information generation rate of each coding stage are prepared, and based on the total encoded information generation rate of one frame by frame coding stage and the number of the stage, the coding and decoding control mode, operation of the switch and the vector reconstructor are adaptively selected and controlled frame by frame at the subsequent coding stage.

Claim 10

An image signal progressive build-up encoder/decoder as set forth in any of claim 5, 6, 7, wherein square-root of the mean square error of each sample between the residual signal vector and the decoded vector outputted from the vector reconstructor is calculated as the error amplitude value, and based on the average of the error amplitude value during one frame period and the number of coding stage, the coding and decoding control mode and operation of the switch and the vector reconstructor are adaptively selected and controlled frame by frame at the subsequent coding stage.

Claim 11

An interframe motion image encoder wherein a first frame memory which can store the motion image signals at least by one frame is provided, interframe predicting signal is obtained utilizing the first frame memory, and the interframe prediction error signals are performed high efficiently coding and decoding and the encoded data are transmitted, and at the same time the decoded image signals are written in the first frame memory and updated for the interframe prediction of the subsequent frame or later, said encoder comprising: a discriminator which evaluates approximation of the interframe predicting signal to the input motion image signals to be coded, and then compares the approximation with the prescribed threshold value, and in the case of the large approximation, it generates the signal indicating that the interframe predicting signals are treated as the decoded signal; an encoder and decoder which performs coding and decoding operation in the case of the small approximation; a second frame memory which can store the input motion image signals at least by one frame, and memorises the input signals before coding operation; an estimator which performs subtraction of the image signals of the previous frame read from the second frame memory and the input image signal, and then performs integration frame by frame and obtains the estimation value of the information generation amount of the current frame; and a coding controller which judges the threshold value of the current coding frame corresponding to the corrected motion amount in order to smooth the information generation rate based on the method: the motion amount is estimated from the threshold value and the information generation amount of the previous coding frame, and then the motion amount is corrected by the estimated information generation amount of the current coding frame outputted from the estimator according to estimate the motion amount of the current coding frame, and the corrected motion amount is determined.

Claim 12

An interframe motion image encoder wherein a first frame memory which can store the motion image signals at least by one frame is provided, interframe predicting signal is obtained utilizing the first frame memory, and the interframe prediction error signals are performed high efficiently coding and decoding and the encoded data are transmitted, and at the same time the decoded image signals are written in the first frame memory and updated for the interframe prediction of the subsequent frame or later, said encoder comprising: a discriminator which evaluates approximation of the interframe predicting signal to the input motion image signals to be coded, and then compares the approximation with the prescribed threshold value, and in the case of the large approximation, it generates the signal indicating that the interframe predicting signals are treated as the decoded signal; an encoder and decoder which performs coding and decoding operation in the case of the small approximation; a second frame memory which can store the input motion image signals at least by one frame, and memorises the input signals before coding operation; an estimator which performs subtraction of the image signals of the previous frame read from the second frame memory and the input image signal, and then performs integration frame by frame and obtains the estimation value of the information generation amount of the current frame; and a coding controller which judges the threshold value of the current coding frame corresponding to the corrected motion amount in order to smooth the information generation rate based on the method: the motion amount is estimated from the threshold value and the information generation amount of the previous coding frame, and then the motion amount is corrected by the estimated information generation amount of the current coding frame outputted from the estimator according to estimate the motion amount of the current coding frame, and the corrected motion amount is determined.

Claim 13

An image signal progressive build-up encoder wherein to same one input image frame the input signals are performed high efficiently coding and decoding and reconstructed image signals are delayed by one frame period, and the residual signals between the input image signals and the delayed signals are performed high efficiently coding and decoding frame by frame respectively and operation is repeatedly performed thereby high quality and resolution image is built up progressively as time progresses, said encoder comprising: a vector/subvector transformer which forms lattice-shaped am sample x bn line (a, b, m, n: natural number, and am = a x m, and bn = b x n) residual signals into an L-dimensional (L = am x bn) vector, and forms lattice-shaped m x n average values calculated within the blocks of a sample x b line residual signals respectively into a K-dimensional (K = m x n) subvector; an average value encoder and decoder which performs high efficiently coding and decoding of the arithmetic means of the sample within the residual signal subvector; K-dimension normalized vector quantization encoder and decoder which treats the residual signal subvector as K-dimension input vector, and performs vector quantization of the input vector through the inner product operation of K dimensions into K-dimension normalized output vector of average value 0 and amplitude 1, and performs high efficiently coding and decoding of the amplitude gain of the residual signal subvector given by the inner product value between normalized output vector and the input vector; a subvector/vector de-transformer which arranges samples within the decoded K-dimension residual signal subvector obtained by multiplying the normalized output vector and the decoded amplitude gain, and each element of the K-dimension amplitude reconstructed output vector corresponding to lattice-shaped m sample x n line subvector is added to the decoded average value, thereby performs interpolation of the decoded residual signal vector having L dimensions; and performs decrease size of the lattice-shaped block of the residual signals step by step and to execute the coding and decoding operation repeatedly.

Claim 14

An image signal progressive build-up encoder wherein to same one input image frame the input signals are performed high efficiently coding and decoding and reconstructed image signals are delayed by one frame period, and the residual signals between the input image signals and the delayed signals are performed high efficiently coding and decoding frame by frame respectively and operation is repeatedly performed thereby high quality and resolution image is built up progressively as time progresses, said encoder comprising: a vector/subvector transformer which forms lattice-shaped am sample x bn line (a, b, m, n: natural number, and am = a x m, and bn = b x n) residual signals into an L-dimensional (L = am x bn) vector, and forms lattice-shaped m x n average values calculated within the blocks of a sample x b line residual signals respectively into a K-dimensional (K = m x n) subvector; an average value encoder and decoder which performs high efficiently coding and decoding of the arithmetic means of the sample within the residual signal subvector; K-dimension normalized vector quantization encoder and decoder which treats the residual signal subvector as K-dimension input vector, and performs vector quantization of the input vector through the inner product operation of K dimensions into K-dimension normalized output vector of average value 0 and amplitude 1, and performs high efficiently coding and decoding of the amplitude gain of the residual signal subvector given by the inner product value between normalized output vector and the input vector; a subvector/vector de-transformer which arranges samples within the decoded K-dimension residual signal subvector obtained by multiplying the normalized output vector and the decoded amplitude gain, and each element of the K-dimension amplitude reconstructed output vector corresponding to lattice-shaped m sample x n line subvector is added to the decoded average value, thereby performs interpolation of the decoded residual signal vector having L dimensions; and performs decrease size of the lattice-shaped block of the residual signals step by step and to execute the coding and decoding operation repeatedly.

Claim 15

An image signals progressive build-up encoder as set forth in claim 10, wherein at first coding stage, the input image signals are combined by lattice-shaped m sample x n line and average value within the block is treated as one sample, and the samples are further combined in lattice-shaped m x n samples into K-dimension input vector and K-dimension mean separated and normalized vector quantization is performed for the K-dimension input vector, and at the subsequent coding stage, the K-dimension mean separated and normalized vector quantization is performed again respectively by combination of the lattice-shaped m sample x n line the residual signals thereby hierarchical coding and decoding for each block can be performed.

Claim 16

An image signal progressive build-up encoder as set forth in claim 10 or claim 11, wherein at the subsequent coding stage, the square means of the sample value within the block by combination of the lattice-shaped m sample x n line residual signals is compared with the threshold value, and block by block conditional replenishment is executed for coding and decoding of only the block having the square means value larger than the threshold value, and binary informations indicating the result of the threshold comparison is performed high efficiently coding and decoding.

Claim 17

An interframe vector quantizer in hierarchical multi-stage structure, comprising: a frame memory which can store image signals of at least one frame a subtractor which subtracts interframe predictive signals from input motion image signals to form interframe differential signals; average value and amplitude operation circuit which forms the interframe differential signals obtained by the subtractor into the blocks with horizontal m picture elements by vertical n lines (m, n being positive integers) and calculates the first average value and amplitude; a block discrimination circuit of the first stage which compares the first average value and amplitude with the threshold value of the first stage an average value separation and normalization circuit of the first stage which forms the first average values into blocks of horizontal m by vertical n corresponding spatially to m x n blocks of the interframe differential signals and calculates second average value and amplitude and subtracts the second average value from the block of the first average values; a code table of the first stage which stores a set of output vectors suited for the distribution of the input vector assuming that the block of the first average values is the input vector of the first stage a distortion operation circuit of the first stage which calculates distortion between the input vector of the first stage and the output vector; a minimum distortion detecting circuit which detects the optimum output vector of the first stage which gives the minimum distortion and determines the index of the optimum output vector of the first stage; a decoding circuit of the first stage which multiplies the optimum output vector of the first stage by the second amplitude and adds the second average value thereby obtains the decoded vector of the first stage; a subtractor which subtracts the decoded vector of the first stage from the interframe differential signal and obtains the error signal; an average value separation and normalization circuit of the next stage which forms the error signals into the blocks with the horizontal m elements by vertical n lines so that the blocks of the error signals corresponding spatially to the blocks of the interframe differential signals formed by the average value and amplitude operation circuit of the first-stage, and calculates the third average values and amplitude regarding blocks except for the block where the first average value and the amplitude are less than the threshold value of the first stage, and subtracts the third average value from the block of error signals and normalizes at the third amplitude; a block discrimination circuit of the next stage which compares the third average value and amplitude with the threshold value of the next stage a code table of the next stage which stores a set of output vectors suited for the distribution of the input vectors of the next stage assuming that the blocks of the error signals except for ones where the third average value and amplitude are less than the threshold value of the next stage in the block discrimination circuit of the second stage; a distortion operation circuit of the next stage which calculates distortion distance between the input vector of the next stage and the output vector; a minimum distortion detecting circuit of the next stage which detects the optimum output vector of the next stage which gives the minimum distortion among the distortion calculated by the distortion operation circuit of the next stage and determines the index of the optimum output vector; a decoding circuit of the next stage which multiplies the optimum output vector of the next stage by the third amplitude and adds the third average value thereby obtains the decoded vector of the next stage; an adder which adds the decoded vector of the first stage and the decoded vector of the next stage and forms the decoded interframe differential signals; an adder which adds the decoded interframe differential signals to .the interframe predictive signals and obtains the decoded image signal; a variable length coding section which operates variable length coding of the second and third average value and amplitude and the index of the optimum output vector of the first stage and the next stage and the result of the block discrimination circuit of the first stage and the next stage and a buffer which temporarily stores data subjected to variable length coding and performs smoothing of the amount of information to be transmitted, wherein when the first average value and amplitude are detected to be less than the threshold value of the first stage by the block discrimination circuit of the first stage, and the decoded vector at the first stage and the corresponding blocks of the next stage are assumed as zero vectors, when the third average value and amplitude are less than the threshold value of the next stage the decoded vector of the next is assumed as zero vector, and threshold value at the first stage and the next stage is hierarchically controlled when the storage amount in the buffer is supposed as at least one of the control condition.

Claim 18

An interframe vector quantizer in hierarchical multi-stage structure, comprising: a frame memory which can store image signals of at least one frame a subtractor which subtracts interframe predictive signals from input motion image signals to form interframe differential signals; average value and amplitude operation circuit which forms the interframe differential signals obtained by the subtractor into the blocks with horizontal m picture elements by vertical n lines (m, n being positive integers) and calculates the first average value and amplitude; a block discrimination circuit of the first stage which compares the first average value and amplitude with the threshold value of the first stage an average value separation and normalization circuit of the first stage which forms the first average values into blocks of horizontal m by vertical n corresponding spatially to m x n blocks of the interframe differential signals and calculates second average value and amplitude and subtracts the second average value from the block of the first average values; a code table of the first stage which stores a set of output vectors suited for the distribution of the input vector assuming that the block of the first average values is the input vector of the first stage a distortion operation circuit of the first stage which calculates distortion between the input vector of the first stage and the output vector; a minimum distortion detecting circuit which detects the optimum output vector of the first stage which gives the minimum distortion and determines the index of the optimum output vector of the first stage; a decoding circuit of the first stage which multiplies the optimum output vector of the first stage by the second amplitude and adds the second average value thereby obtains the decoded vector of the first stage; a subtractor which subtracts the decoded vector of the first stage from the interframe differential signal and obtains the error signal; an average value separation and normalization circuit of the next stage which forms the error signals into the blocks with the horizontal m elements by vertical n lines so that the blocks of the error signals corresponding spatially to the blocks of the interframe differential signals formed by the average value and amplitude operation circuit of the first-stage, and calculates the third average values and amplitude regarding blocks except for the block where the first average value and the amplitude are less than the threshold value of the first stage, and subtracts the third average value from the block of error signals and normalizes at the third amplitude; a block discrimination circuit of the next stage which compares the third average value and amplitude with the threshold value of the next stage a code table of the next stage which stores a set of output vectors suited for the distribution of the input vectors of the next stage assuming that the blocks of the error signals except for ones where the third average value and amplitude are less than the threshold value of the next stage in the block discrimination circuit of the second stage; a distortion operation circuit of the next stage which calculates distortion distance between the input vector of the next stage and the output vector; a minimum distortion detecting circuit of the next stage which detects the optimum output vector of the next stage which gives the minimum distortion among the distortion calculated by the distortion operation circuit of the next stage and determines the index of the optimum output vector; a decoding circuit of the next stage which multiplies the optimum output vector of the next stage by the third amplitude and adds the third average value thereby obtains the decoded vector of the next stage; an adder which adds the decoded vector of the first stage and the decoded vector of the next stage and forms the decoded interframe differential signals; an adder which adds the decoded interframe differential signals to .the interframe predictive signals and obtains the decoded image signal; a variable length coding section which operates variable length coding of the second and third average value and amplitude and the index of the optimum output vector of the first stage and the next stage and the result of the block discrimination circuit of the first stage and the next stage and a buffer which temporarily stores data subjected to variable length coding and performs smoothing of the amount of information to be transmitted, wherein when the first average value and amplitude are detected to be less than the threshold value of the first stage by the block discrimination circuit of the first stage, and the decoded vector at the first stage and the corresponding blocks of the next stage are assumed as zero vectors, when the third average value and amplitude are less than the threshold value of the next stage the decoded vector of the next is assumed as zero vector, and threshold value at the first stage and the next stage is hierarchically controlled when the storage amount in the buffer is supposed as at least one of the control condition.

Claim 19

An interframe vector quantizer as set forth in claim 13, wherein when distortion distance between input vector of the first stage and output vector read from the code book of the first stage is determined in the distortion operation circuit of the first stage, element of the input vector of the first stage corresponding to the block having the first average value and amplitude less than the threshold value in comparison in the block discrimination circuit of the first stage is ignored in process of the distortion calculation.

Claim 20

An interframe vector quantizer as set forth in claim 13, wherein before the third average value and amplitude are subject to variable length coding in the variable length coding section, a quantizer of average value and amplitude of subsequent stage having a plurality of quantization characteristics is previously provided, and the plurality of quantization characteristics are adaptively changed by the second amplitude obtained in the average value separation and normalization circuit of the first stage.

Claim 21

An interframe vector quantizer as set forth in claim 13, wherein plural sets of output vectors having different characteristics are stored in the code table of the next stage, and the plurality of quantizing characteristics are adaptively changed by the second amplitude obtained in the average value separation and normalization circuit of the first stage.

Claim 22

A dynamic vector quantizer provided with a coding section comprising: an average value separation circuit which separates the average value within a block from input vectors formed with every k samples (k being integer of two or more) of input signal; an average value coding circuit which encodes the average value; a normalization circuit which normalizes the average value separated input vector obtained through average value separation by the average value separation circuit according to the amplitude of the input vector; a fixed code book which stores a plurality of normalized output vectors suited for the distribution of the normalized input vectors; a dynamic code book which stores the plurality of normalized input vectors supposing them as normalized output vectors, the stored content being updated; an inner product operation section which calculates inner product value between the average value separated input vector and a plurality of normalized output vectors stored in the fixed code book and the dynamic code book; a maximum inner product detecting section for finding maximum inner product value among a plurality of inner product values calculated by the inner product operation section; an amplitude coding circuit which encodes either the maximum inner product value or the normalization coefficient; a dynamic code book control section which calculates a degree of approximation between the input vector and the decoded vector after vector quantization of the input vector using the maximum inner product value and the normalization coefficient, and controls the updating operation of the dynamic code book based on comparison result of the degree of approximation and threshold value previously set; a coded data multiplexing section which multiplexes either discrimination code relating to the update procedure, the output data of the average value coding circuit, the output data of the amplitude coding circuit and the maximum inner product value or normalized output vector and the discrimination code is attached according to the prescribed format, and then transmits it; means which calculates the degree of approximation with the normalization coefficient and the maximum inner product value and selects the maximum inner product value as input signal to the amplitude coding circuit when the degree of approximation is larger than the threshold value, and defines the input signal to the coding data multiplexing section with the discrimination code, of the output data of the average value coding circuit, the output data of the amplitude coding circuit and the normalized output vector; and means which replaces the oldest normalized input vector stored in the dynamic code book corresponding to the input vector with the normalized input vector thereby executes the update operation when the degree of approximation is less than the threshold value, and selects the normalization coefficient as input signal to the amplitude coding circuit, and further defines the input signal to the coding data multiplexing section with the normalized input vector and the output data of the average value coding circuit, the output data of the amplitude coding circuit and the discrimination symbol indicating that the updating operation has been done thereto, decoding section comprising: a coded data demultiplexing section which separates the coded data transmitted from the coding section into the average value coded data, the amplitude coded data and the discrimination symbol according to the prescribed format; an average value decoding circuit which decodes the average value coded data an amplitude decoding circuit which decodes the amplitude coded data; a fixed code book and a dynamic code book corresponding to the fixed code book and the dynamic code book at the side of the coding section respectively; means which reads the normalized output vector on the address indicated by the discrimination symbol from the fixed code book or the dynamic code book when the discrimination symbol indicates the normalized output vector, and performs prescribed operation based on the normalized output vector, and decoded amplitude value outputted from the amplitude decoding circuit, and the decoded average value outputted from the average value decoding circuit thereby obtains the decoded vector to the input vector; and means which replaces the oldest vector stored in the dynamic code book with the normalized input vector received following the discrimination symbol in the case that the discrimination symbol indicates the update operation thereby executes similar to the update in the coding section, and performs prescribed operation of the normalized input vector based on the decoded amplitude value and decoded average value thereby obtains the decoded vector to the input vector.

Claim 23

A dynamic vector quantizer provided with a coding section comprising: an average value separation circuit which separates the average value within a block from input vectors formed with every k samples (k being integer of two or more) of input signal; an average value coding circuit which encodes the average value; a normalization circuit which normalizes the average value separated input vector obtained through average value separation by the average value separation circuit according to the amplitude of the input vector; a fixed code book which stores a plurality of normalized output vectors suited for the distribution of the normalized input vectors; a dynamic code book which stores the plurality of normalized input vectors supposing them as normalized output vectors, the stored content being updated; an inner product operation section which calculates inner product value between the average value separated input vector and a plurality of normalized output vectors stored in the fixed code book and the dynamic code book; a maximum inner product detecting section for finding maximum inner product value among a plurality of inner product values calculated by the inner product operation section; an amplitude coding circuit which encodes either the maximum inner product value or the normalization coefficient; a dynamic code book control section which calculates a degree of approximation between the input vector and the decoded vector after vector quantization of the input vector using the maximum inner product value and the normalization coefficient, and controls the updating operation of the dynamic code book based on comparison result of the degree of approximation and threshold value previously set; a coded data multiplexing section which multiplexes either discrimination code relating to the update procedure, the output data of the average value coding circuit, the output data of the amplitude coding circuit and the maximum inner product value or normalized output vector and the discrimination code is attached according to the prescribed format, and then transmits it; means which calculates the degree of approximation with the normalization coefficient and the maximum inner product value and selects the maximum inner product value as input signal to the amplitude coding circuit when the degree of approximation is larger than the threshold value, and defines the input signal to the coding data multiplexing section with the discrimination code, of the output data of the average value coding circuit, the output data of the amplitude coding circuit and the normalized output vector; and means which replaces the oldest normalized input vector stored in the dynamic code book corresponding to the input vector with the normalized input vector thereby executes the update operation when the degree of approximation is less than the threshold value, and selects the normalization coefficient as input signal to the amplitude coding circuit, and further defines the input signal to the coding data multiplexing section with the normalized input vector and the output data of the average value coding circuit, the output data of the amplitude coding circuit and the discrimination symbol indicating that the updating operation has been done thereto, decoding section comprising: a coded data demultiplexing section which separates the coded data transmitted from the coding section into the average value coded data, the amplitude coded data and the discrimination symbol according to the prescribed format; an average value decoding circuit which decodes the average value coded data an amplitude decoding circuit which decodes the amplitude coded data; a fixed code book and a dynamic code book corresponding to the fixed code book and the dynamic code book at the side of the coding section respectively; means which reads the normalized output vector on the address indicated by the discrimination symbol from the fixed code book or the dynamic code book when the discrimination symbol indicates the normalized output vector, and performs prescribed operation based on the normalized output vector, and decoded amplitude value outputted from the amplitude decoding circuit, and the decoded average value outputted from the average value decoding circuit thereby obtains the decoded vector to the input vector; and means which replaces the oldest vector stored in the dynamic code book with the normalized input vector received following the discrimination symbol in the case that the discrimination symbol indicates the update operation thereby executes similar to the update in the coding section, and performs prescribed operation of the normalized input vector based on the decoded amplitude value and decoded average value thereby obtains the decoded vector to the input vector.

Claim 24

A dynamic vector quantizer as set forth in claim 17, wherein in the coding section, a plurality of normalized output vectors within the fixed code book are arranged in tree structure in step per 2" (n: natural number), and a normalized output vector giving the maximum inner product value to the input vector is selected among 2" pieces of the normalized output vectors in sequence from upper step of the tree structure towards lower step and tree searching method is used.

Claim 25

A dynamic vector quantizer as set forth in claim 17, wherein the normalization circuit performs normalization using the root of the sum of the squared average value-separated-input vector as the normalization coefficient.

Claim 26

A dynamic vector quantizer as set forth in claim 17, wherein the fixed code book stores a plurality of normalized output vectors with average value zero and magnitude 1 determined depending on the statistical characteristics of plural normalized input vectors.

Claim 27

A dynamic vector quantizer as set forth in claim 27, wherein the degree of approximation between the input vector and the decoded vector after vector quantization is estimated using the squared normalization coefficient and the squared maximum inner product value.

Claim 28

A dynamic vector quantizer as set forth in claim 17, wherein the discrimination symbol indicates the existence of the index corresponding to the output vector, the normalized output vector is multiplied by the decoded amplitude value and the product is added to the decoded average value.

Claim 29

A dynamic vector quantizer as set forth in claim 17, wherein the discrimination symbol indicates that the update operation has been done, the normalized input vector is multiplied by the decoded amplitude value and the product is added to the decoded average value.

Claim 30

An image coding transmission system wherein at the transmitting side digitalized image signal is inputted and coded and then transmitted, and at the receiving side transmitted signal is decoded and outputted as image signal, at the transmitting side of one station the information indicating time capability of the decoding process of the station is transmitted, and at the receiving side the transmitted information indicating time capability of the decoding process of the opposite station is detected and the coding processing speed of the transmitting side may be controlled based on the information.

Claim 31

An image coding transmission system wherein at the transmitting side digitalized image signal is inputted and coded and then transmitted, and at the receiving side transmitted signal is decoded and outputted as image signal, at the transmitting side of one station the information indicating time capability of the decoding process of the station is transmitted, and at the receiving side the transmitted information indicating time capability of the decoding process of the opposite station is detected and the coding processing speed of the transmitting side may be controlled based on the information.

Claim 32

An image coding transmission system as set forth in claim 24, wherein based on the information indicating time capability of the decoding process of the opposite station, the number of image frames to be coded per unit time in the self station may be controlled.

Claim 33

An image coding transmission system as set forth in claim 24, wherein based on information indicating time capability of the decoding process of the opposite station, the number of image frames to be transmitted per unit time in the self station may be controlled.

Claim 34

An image coding transmission system as set forth in claim 24, wherein based on information indicating time capability of the decoding process of the opposite station, minimum value of time interval between the transmitting of the code corresponding to the lead of each image frame at the self station may be controlled.