Abstract: A resource allocation of multiple compressed AV streams delivered over the Internet is disclosed that achieves end-to-end optimal quality through a multimedia streaming TCP-friendly transport (MSTFP) protocol that adaptively estimates the network bandwidth while smoothing the sending rate. Resources allocated dynamically according to a media encoding distortion and network degradation algorithm. A scheme is also disclosed for dynamically estimating the available network bandwidth for streaming of objects, such as MPEG4 multiple video objects, in conjunction with the MSTFP protocol. The scheme can account for packet-loss rates to minimize end-to-end distortion for media delivery.

Abstract: A run-level coding module can be used in a video encoder that generates a processed video signal from a video input signal. The run-level coding module includes a run-level coder that generates a first plurality of run-level pairs from a first stream of quantized data. A first ring buffer buffers a first number of the first plurality of run-level pairs. The processed video signal is generated based on the buffered first number of run-level pairs.

Abstract: Methods and apparatuses for blending two images using vector table look up operations. In one aspect of the invention, a method to blend two images includes: loading a vector of keys into a vector register; converting the vector of keys into a first vector of blending factors for the first image and a second vector of blending factors for the second image using a plurality of look up tables; and computing an image attribute for the blended image using the blending factors.

Abstract: Disclosed are a method and system for video compression, wherein the video encoder has low computational complexity and high compression efficiency. The disclosed system comprises a video encoder and a video decoder, wherein the method for encoding includes the steps of converting a source frame into a space-frequency representation; estimating conditional statistics of at least one vector of space-frequency coefficients; estimating encoding rates based on the said conditional statistics; and applying Slepian-Wolf codes with the said computed encoding rates. The preferred method for decoding includes the steps of; generating a side-information vector of frequency coefficients based on previously decoded source data, encoder statistics, and previous reconstructions of the source frequency vector; and performing Slepian-Wolf decoding of at least one source frequency vector based on the generated side-information, the Slepian-Wolf code bits and the encoder statistics.

Abstract: A vector coding apparatus includes a decomposing unit to obtain a sign vector and an initial absolute vector, a sign coding unit to code the sign to obtain a sign code, a multi-level permutation-based coding unit to perform multi-level permutation-based coding to obtain an absolute vector code, a combining unit to combine the sign and absolute vector codes to obtain an initial vector code. A vector decoding apparatus includes a decomposing unit to decompose a code of an initial vector to obtain a sign code and an absolute vector code, a sign decoding unit to decode the sign code to obtain a sign vector, a multi-level permutation-based decoding unit to perform multi-level permutation-based decoding on the absolute vector code to obtain an initial absolute vector, and a combining unit to combine the sign and initial absolute vectors to obtain an initial vector. A media player includes the vector decoding apparatus.

Abstract: A method for quantizing vector. The method includes: performing a quantization process on a vector to be quantized by use of N basic codebook vectors and the adjustment vectors of each of the basic codebook vectors, generating a basic codebook vector and an adjustment vector used for quantizing the vectors to be quantized, N being a positive integer larger than or equal to 1. According to the present invention, based on the method a device for quantizing vector is disclosed. According to embodiments of the present invention, the quantization of an input vector is done by introducing the modification vectors for the base codebook vectors, therefore the memory amount of the base codebook vectors is reduced effectively, and the calculation amount is merely the calculation amount required for going through N codebooks. Therefore, the complexity of the vector quantization could be decreased effectively.

Abstract: In advanced video coding standards such as H.264, macro-blocks belong to more advanced MB types, such as skipped and non-skipped macro-blocks. In non-skipped macro-blocks, the encoder determines whether each of 8×8 luminance sub-blocks and 4×4 chrominance sub-block of a macro-block is to be encoded, giving the different number of sub-blocks at each macro-block encoding times. It has been found that the correlation of bits between consecutive frames is high. This correlation is even higher after macro-block normalization by considering advanced macro-block types. Based on this bit characteristic, a fast real-time H.264 rate control scheme is herein described. The empirical example results suggest that this scheme can achieve PSNR gain over JM10.2.

Abstract: To always maintain good image quality no matter whether the target bit rate for encoding is high or low. An encoder 210 includes: a converting section 103 for converting picture data representing a moving picture into a coefficient including a plurality of frequency components on a predetermined unit basis; a determining section 212 for determining a matrix to define a frequency bit allocation characteristic based on the resolution of the moving picture and a target bit rate for encoding that has been specified in advance; a quantization section 204 for quantizing each coefficient based on the matrix determined, thereby generating a quantized value; and an encoding section 105 for encoding the quantized value to generate encoded data of the moving picture.

Abstract: Improved systems and methods for error concealment of multiple description coding (MDC) encoded streams are provided based on a three loop interpolation of lost frames. Error concealment of the present invention can be combined with the error resilience provided by MDC to reconstruct lost frames, such that the propagated error to the following frames is reduced.

Abstract: Improved error resiliency of an encoding device, such as a video codec or encoder, operating in a compressed data transmission system, is achieved by enabling the encoding device to “shadow” or mimic the error conditions of a decoding device that receives and decodes compressed data sent by the encoding device. The encoding device is made aware of the specific error concealment scheme that the decoding device employs to reconstruct video frame data from a bit stream that contains bit or packet errors, and is also made aware of the location(s) in the frame to which the erroneous data corresponds and the particular manner in which the erroneous data was handled. The error-concealed data then can be used not only for the present encoding, but for subsequent encodings as well, to achieve synchronization between the encoder and decoder. Advantageously, the present invention improves on, but can work with, conventional error concealment schemes.

Abstract: Embodiments of the invention describe a compound conditional source coding method and system for communicating source data over a network. Length-n random uncompressed source data are drawn according to a distribution px(x), and serves as input data to an encoder. A set P of candidate side-information vectors is also input to the encoder. The encoder encodes the source data, utilizing the set of the candidate side-information vectors, to produce an encoded message. The message is transmitted to a decoder. The decoder decodes the received message to produce a source estimate, using selected side-information vector and an index of the selected side-information vector in the set of the candidate side-information vectors.

Abstract: A motion compensation module can be used in a video encoder for encoding a video input signal that includes a sequence of images that are segmented into a plurality of macroblocks. The motion compensation module includes a motion search module, that generates a motion search motion vector for a plurality of subblocks for a plurality of partitionings of a macroblock of a plurality of macroblocks. A motion refinement module generates a refined motion vector for the plurality of subblocks for the plurality of partitionings of the macroblock of the plurality of macroblocks, based on the motion search motion vector for each of the plurality of subblocks of the macroblock of the plurality of macroblocks. The motion refinement module can operate in a plurality of selected modes including a first mode corresponding to a first sub-pixel resolution and a second mode corresponding to a second sub-pixel resolution.

Abstract: A signal processing system adapted for sparse representation of signals is provided, comprising: (i) one or more training signals; (ii) a dictionary containing signal-atoms; (iii) a representation of each training signal using a linear combination of said dictionary's signal-atoms; (iv) means for updating the representation of the training signal; (v) means for updating the dictionary one group of atoms at a time, wherein each atom update may include all representations referring to said updated atom; and (vi) means for iterating (iv) and (v) until a stopping rule is fulfilled. The system uses the K-SVD algorithm for designing dictionaries for sparse representation of signals.

Abstract: There are provided an encoding device, a decoding device, an encoding method, and a decoding method that enhance optimality in a case of encoding moving image signals having no sample ratio distinction between color components such as a 4:4:4 format. In a case of conducting compression encoding by inputting digital moving image signals of the 4:4:4 format, there are prepared a first encoding process of encoding three color component signals of the input moving image signals in a common encoding mode, and a second encoding process of encoding the three color component signals of the input moving image signals in respective independent encoding modes. The encoding process is executed by selecting any one of the first encoding process and the second encoding process, and the compression data contains an identification signal for specifying which process is selected.

Abstract: The invention relates to a method of transmitting an audio and/or video program via a transmission channel at adjustable bit rates, the method implementing an adjustment of at least one encoding and/or transmission parameter as a function of at least one setpoint vector having at least one dimension and representing a quality of reception desired by said end user.

Abstract: A system and method for effectively encoding and decoding electronic information includes an encoding system with a tiling module that initially divides source image data into data tiles. A frame differencing module then outputs only altered data tiles to various processing modules that convert the altered data tiles into corresponding tile components. A quantizer performs a compression procedure upon the tile components to generate compressed data according to an adjustable quantization parameter. An adaptive entropy selector then selects one of a plurality of available entropy encoders to most effectively perform an entropy encoding procedure to thereby produce encoded data. The entropy encoder may also utilize a feedback loop to adjust the quantization parameter in light of current transmission bandwidth characteristics.

Abstract: The present invention provides an image processing device, an image processing method, and an image reading method, by which encoded data can be easily generated from sub-sampled image data, without a complicated circuit or an increase in the data amount. A one-dimensional wavelet transform in a predetermined direction is omitted in stage 1, where components 1 and 2 of input image data have been sub-sampled. When coefficient data generated through such a two-dimensional wavelet transform are to be encoded, sub-bands that are not contained in the coefficient data in comparison with the coefficient data of a component 0 should be considered to have been truncated.

Abstract: A set of customizing operations for digital content is determined in accordance with network condition of a current network communication channel between a content server and one or more receiving devices, wherein the digital content is provided by the content server for transport to the receiving device and includes multiple frames of digital video data. The set of customizing operations specify multiple sequences or paths of customized video data in accordance with available video frame rates, and a customized video data sequence is selected from among the specified multiple sequences of customized video data in accordance with estimated received video quality and network condition for each receiving device.

Abstract: This invention calculates motion vectors during video encoding for sub-sampled data using a hierarchical search. The motion vector calculation employs a prior original frame as the reference frame rather than a reconstructed frame according to the prior art.

Abstract: An image coding apparatus is provided with a unit operable to scan a two-dimensional array of quantized coefficients into scanned quantized coefficients, the scanning being performed from a low frequency component toward a high frequency component; a unit operable to convert the scanned quantized coefficients into a run value and a level value, the run value indicating the number of continuous quantized coefficients, each having a zero value, and the level value indicating a value of a quantized coefficient having a non-zero value; a unit operable to code the run value; and a unit operable to code the level value, wherein the coding of the run value is performed, from a high frequency component toward a low frequency component, in accordance with information that represents a total number of uncoded quantized coefficients which have not been coded.

Abstract: An image coding apparatus provides a run-length encoding unit RLE1 that subjects quantized coefficients which are obtained by quantizing frequency components of an image signal to a variable length coding process by using a run value Run that indicates the number of successive zero coefficients and a level value Lev that indicates a value of a non-zero coefficient following the zero coefficients.

Abstract: An image decoding apparatus is provided with a unit operable to receive coded data including a coded run value and a coded level value, wherein the coded run value is obtained by coding a run value which indicates the number of continuous quantized coefficients, each having a zero value, and the coded level value is obtained by coding a level value which indicates a value of a quantized coefficient having a non-zero value; a unit operable to decode the coded run value to obtain a decoded level value; a unit operable to decode the coded level value to obtain a decoded level value; and a unit operable to obtain quantized coefficients of the block image from the decoded level value and the decoded run value, wherein the decoding of the coded run value is performed from a high frequency component toward a low frequency component, in accordance with information that represents a total number of undecoded quantized coefficients which have not been decoded.

Abstract: A method for memory management in video decoding systems that avoids some of the costs and disadvantages with video decoding systems in the prior art. Some embodiments of the present invention are especially well-suited for use with the H.264 video decoding standard. The illustrative embodiment is a memory management technique that controls which data is in the fastest memory available to a processor performing video decoding. In particular, the technique seeks to ensure that the data the processor will need is in the primary memory and expunges data that the processor will not need. The technique is based upon an analysis of predictive video decoding standards, such as H.264. By employing this technique, the illustrative embodiment ensures the expedient decoding of video frames.

Abstract: The invention provides a system and method for JPEG encoding an image that includes variable quantization for each block, depending upon the classification of each block, which is dependent on the classification of pixels in that block.

Abstract: A resource allocation of multiple compressed AV streams delivered over the Internet is disclosed that achieves end-to-end optimal quality through a multimedia streaming TCP-friendly transport (MSTFP) protocol that adaptively estimates the network bandwidth while smoothing the sending rate. Resources allocated dynamically according to a media encoding distortion and network degradation algorithm. A scheme is also disclosed for dynamically estimating the available network bandwidth for streaming of objects, such as MPEG-4 multiple video objects, in conjunction with the MSTFP protocol. The scheme can account for packet-loss rates to minimize end-to-end distortion for media delivery.

Abstract: Methods and apparatuses for blending two images using vector table look up operations. In one aspect of the invention, a method to blend two images includes: loading a vector of keys into a vector register; converting the vector of keys into a first vector of blending factors for the first image and a second vector of blending factors for the second image using a plurality of look up tables; and computing an image attribute for the blended image using the blending factors.

Abstract: A quantization error compensation includes a dividing unit to divide an input current image signal into a first high-frequency signal and a low-frequency signal; a changing unit to replace bits of the first high-frequency signal with a predetermined signal to output a second high-frequency signal; an adding unit to add the low-frequency signal and the second high-frequency signal to create a composite signal; a quantization unit to cut n number of bits of the composite signal and to output the cut composite signal and the n number of bit signal; an equalizing unit to output a brightness equalizing value of the cut composite signal; a calculating unit to calculate a compensation value using the n number of the bit signal and a difference between brightness equalizing values with respect to a current and next brightness levels; and a compensation unit to add the compensation and the current brightness equalizing values.

Abstract: A distortion removal system for reducing the quantization distortion that results from quantization of a signal by sample-by-sample quantization techniques is disclosed. The distortion removal system uses the signal previously quantized with the sample-by-sample quantization techniques to determine an expected quantization distortion. The expected quantization distortion is determined for each of a plurality of frames of the signal. The expected quantization distortion is removed from each of the frames of the signal.

Abstract: A frame group target bits number calculating unit (31) reads the total number of bits on frame group basis allocatable to residual frames from a storing unit (36), updates subtracting the number of generated bits used to code a preceding frame, and transmits to a next frame target bits number calculating unit (32). The next frame target bits number calculating unit (32) calculates the target number of bits to be allocated to a next frame based on the received total number of bits and the number of residual frames with considering a frame rate value, Rf. An average frame bits number calculating unit (33) calculates the average number of bits allocated to preceding frames. A calculating unit (34) multiplies the calculated average number of bits by a predetermined coefficient, compares the resulting value with above target number of bits, selects the greater number and transmits it to a quantization step calculating unit (35).

Abstract: A DTV signal includes frames, and the frames include data segments. Data segments are replaced by codewords which represent digital data. The DTV signals are broadcast and received by receivers. The receivers recover the digital data from the codewords in the DTV signals, for example by using matched filtering. By using long codewords, the digital data can be broadcast over a longer range and recovered by simpler circuitry in comparison to the television programming contained in the DTV signal. In one implementation, the DTV signal is an American Television Standards Committee (ATSC) DTV signal.

Abstract: Methods and apparatuses for blending two images using vector table look up operations. In one aspect of the invention, a method to blend two images includes: loading a vector of keys into a vector register; converting the vector of keys into a first vector of blending factors for the first image and a second vector of blending factors for the second image using a plurality of look up tables; and computing an image attribute for the blended image using the blending factors.

Abstract: The present invention relates to an apparatus and a method for vector descriptor representation and multimedia data retrieval, which can quantize a plurality of feature values described by a vector descriptor respectively, represent the quantized feature values in the form of bit or orthogonally transform the quantized bector feature values, and rearrange the feature values represented in the form of bit from the highest bit to the lowest bit or rearrange the transformed coefficient from low frequency to high frequency to represent the vector descriptor hierarchically.

Abstract: This invention relates to the creation of dictionary functions for the encoding of video signals using matching pursuit compression techniques. After an initial set of reference dictionary images is chosen, training video sequences are selected, and motion residuals are calculated. High energy portions of the residual images are extracted and stored when they match selection criteria with the reference dictionary. An energy threshold is used to limit the number of video signal “atoms” encoded for each frame, thus avoiding the encoding of noise. A new dictionary is then synthesized from the stored portions of the image residuals and the original reference dictionary. The process can then be repeated using the synthesized dictionary as the new reference dictionary. This achieves low bit rate signals with a higher signal-to-noise ratio than have been previously achieved.

Abstract: The present invention extends the generalized Lloyd algorithm (GLA) for vector quantizer (VQ) codebook improvement and codebook design to a new linearly-constrained generalized Lloyd algorithm (LCGLA). The LCGLA improves the quality of VQ codebooks, by forming the codebooks from linear combinations of a reduced set of base codevectors. The present invention enables a principled approach for compressing texture images in formats compatible with various industry standards. New, more flexible compressed texture image formats are also made possible with the present invention. The present invention enhances signal compression by improving traditional VQ approaches through the integrated application of linear constraints on the multiple pattern and signal prototypes that represent a single pattern or block of signal samples.

Abstract: A method for efficient low power motion estimation of a digital video image is provided in which processing requirements are reduced based upon the content being processed. The method performs motion estimation of a current video image using a search window of a previous video image. The method may include forming mean pyramids of a reference macroblock and the search area and a full search at a lowest resolution. A number of candidate motion vectors (CMVs) propagated to lower levels may be dependent on a quantized average deviation estimate (QADE) of a current macroblock and the maximum distortion band obtained during training for that QADE value at that particular level. Training over a sequence may be triggered at the beginning of every sequence. This training technique may be used to determine the value of the maximum distortion band for all QADEs of the macroblocks occurring over the training frames.

Abstract: A database managing apparatus, which can immediately obtain desired data from a database. A database managing apparatus includes a CPU, an input/output device, a main memory and an external memory. The CPU has a controller that initially obtains one article record from stored records via the input/output device. The controller reads out definition data in a database definition file. Then, the controller classifies the obtained records according to attributions based on the definition data. The controller does not compress the data regarding the record group belonging to the attribution A, which is the record group to be searched, but compresses data regarding the record groups belonging to the attributions B-E, which are the record groups other than the record group to be searched. As a result, the controller can reduce the unnecessary decompression of record data belonging to other attributions than the attribution to be searched. As a result, the controller can quickly retrieve the requested record.

Abstract: A method and apparatus for representing motion in a sequence of digitized images derives a dense motion vector field and vector quantizes the motion vector field.

Abstract: A process for encoding digital video signals organized in frames comprises the operations of dividing the frames into blocks starting from macroblocks subjected to motion-compensation and applying to the blocks a discrete cosine transform in such a way as to generate respective sets of coefficients. The sets of coefficients are then assembled by being organized into sets of vectors by a assembling module. Once the variance of the vectors has been detected, the vectors themselves are quantized on a number of available bits by a pyramid vector quantizer, associating to the vectors respective quantization pyramids having given sizes according to the variance detected and to the number of available bits. Finally, the vectors are encoded with respective codewords.

Abstract: A transformation unit 102 transforms an entered image to multi-resolution space. A quantizer 105 performs vector quantization on a local pattern of an image of a multi-resolution representation. A perspective-order calculating unit 107 extracts a plurality of code words, positions corresponding thereto and/or angle of rotation and/or scale, and the perspective-order relationship of a plurality of these representative vectors, from the quantized image. An algebraic encoder 108 encodes the input image based upon the extracted information. As a result, there are provided an image processing apparatus and method for asymmetric encoding without motion in three dimensions and extraction of a three-dimensional structure. Further, a transformation unit 1103 transforms an entered image to vector field, and a singularity detector 1104 detects a singularity in the transformed image.

Abstract: The present invention relates to a method for eliminating a blocking effect in a compressed video signal signal. The method of the invention includes: the encoding step of eliminating a blocking effect by compensating the motion of a signal compressed in block unit to be transmitted; the decoding step of restoring the motion compensated video signal to the original video signal by reducing the prediction residual between the motion compensated video signal and the original video signal and the blocking effect; and the post-filtering step of performing post-filtering in a blocking elimination filter in order to eliminate a blocking effect and ring effect remained in the compensated signal. The equation for obtaining the original pixel is made simple by eliminating the remaining blocking effect and ring effect using a loop/post filter.

Abstract: A new dequantization scheme for DCT-based transform coding, such as JPEG, MPEG and H.26x, is disclosed. The new approach drastically reduces blocking artifacts without smoothing the decoded image. Most discrete cosine transform (DCT) based video coding suffers from blocking artifacts where boundaries of 8×8 DCT blocks become visible on decoded images. The blocking artifacts become more prominent as the bit rate is lowered. In the present invention, a new dequantization technique is disclosed for discrete cosine transform (DCT) based encoding to sharply reduce the blocking artifacts. The dequantization scheme of the present invention sharply reduces blocking artifacts in decoded images through regularization. The performance comparison with the standard JPEG as well as MPEG and H.26x decoding shows visual improvements as well as numerical improvements in terms of the peak-signal-to-noise ratio (PSNR) and the blockiness measure (BM) to be defined.

Abstract: An encoding apparatus and method for an orientation interpolator node are provided. The encoding apparatus for an orientation interpolator node, which provides information on the rotation of an object in a 3-dimensional space, includes a field data input unit for extracting field data to be encoded at present from a key which indicates information on a position on a time axis where a change of a rotational movement on a time axis occurs and key values which indicate rotation information corresponding to the position information, by parsing the orientation interpolator node; an adaptive differential pulse code modulation (ADPCM) processing unit for converting the key value data into a quaternion, and then ADPCM processing the quaternion using a rotation differential converting matrix, and differential pulse code modulation (DPCM) processing the key data; and a quantizing unit for quantizing the key data and key value data and outputting the quantized data.

Abstract: A pixel calculating device that performs vertical filtering on pixel data in order to reduce frame data in a vertical direction. The pixel calculating device includes a decoding unit 401 for decoding compressed video data to produce frame data, frame memory 402 for storing the frame data, a filtering unit 403 for reducing the frame data in a vertical direction by the vertical filtering to produce a reduced image, buffer memory 404 for storing the reduced image outputted from filtering unit 403, and a control unit 406 for controlling filtering unit 403 based on a decoding state of the video data by decoding unit 401 and a filtering state of the frame data by filtering unit 403, so that overrun and underrun do not occur in filtering unit 403.

Abstract: A wavelet pyramid-based still-image and video compression apparatus uses line-vectors, context-dependent quantization thresholds, hierarchical tree-structured displacement coding, synchronous prediction, and conditional zerotree branch updates to improve intra- and inter-frame coding speed and efficiency, and to improve video stream noise resilience. The disclosed intra-frame coding methods can be used to improve the coding efficiency of wavelet-based still-image compression algorithms.

Abstract: A method for data compression. An encoder receives data vectors from an original data set. The encoder uses a vector quantization codebook to encode the data vectors into encoded vectors. The codebook is constructed from a compound data set, where the compound data set includes real data vectors and artificial data vectors. The encoded vectors are indexed in the codebook and the indexes are transmitted across communication channels or transmitted to storage.

Abstract: A dc-free coding scheme that also provides channel gain employs a code that is based on an alphabet of two and comprises a list of N-dimensional vectors, where N is even, with a maximum run of two symbols. From this list, all vectors which begin or end in two identical symbols are eliminated. This avoids runs of longer than two consecutive symbols when vectors are concatenated. With this scheme, if the baseband transmission of a “+” is realized with positive voltage V and transmission of a “−” is realized with a negative voltage −V, then any concatenation of symbols is dc-free. Improved coding gain is realized by mapping modulated signals onto convolutional code trellises.

Abstract: A motion compensated video coding method which can be applied especially in transfer of video streams using low transmission bit rate is presented. In the motion compensated coding method, the motion of picture elements between a piece of reference video information and a piece of current video information is estimated and then modeled using certain basis function and coefficients. The coefficients are quantized, and the quantizer is selected according to a certain selection criterion, for example, based on a target image quality or on a target transmission bit rate. Preferably the selection criterion is such that it automatically adjust the accuracy with which the motion of picture elements is represented to be related to the accuracy with which the prediction error information is represented. A decoding method, an encoder and a corresponding decoder are also described.

Abstract: The invention proposes a method of determining an index of a code vector (VCk) belonging to a predetermined lattice, characterised in that it includes the steps of: determining (E47) a leader vector (Y, VLk) for the code vector (VCk), seeking (E50, E51) a permutation number (NP) between the code vector and the leader vector, determining (E51) the index (Ik) for the code vector according to the permutation number. The invention makes it possible to code a code vector, for example in the context of the lattice vector quantization of a digital signal.

Abstract: A decoder in a communication system using channel optimized vector quantization and including an encoder stores active encoder centroids ci corresponding to active indices i, an active index being defined as an index that has a predetermined source probability pi>0 of being selected by the encoder for transmission to the decoder, source probabilities pi and the bit error rate epsilon to dynamically estimate a decoded vector using a sub-optimal algorithm in which only the most significant contributions are considered.

Abstract: The present method relates to a method for encoding image data using vector quantization. According to the invention, a small first codebook is determined. Each image vector of the image data is then encoded by determining a codevector within the first codebook that best approximates the image vector within the image data. A first index map is generated by replacing each image vector with an index indicative of the codevector's location within the first codebook. Then difference data are evaluated based on the original image data and the encoded image data. Each error vector of the difference data is then encoded using another small codebook. In another index map the error vectors are then replaced with an index indicative of the codevector's location within the other codebook. Evaluation of the error based on the difference data and the encoded difference data provides new difference data which is used to evaluate the fidelity of the approximation process performed for compression.