Abstract: An image information coding apparatus includes cut-out unit for cutting out image information into blocks each having a predetermined size, transforming unit for preprocessing to generate frequency information from each block, coding unit for separating the frequency information into a first frequency component relating to an average value of the block and a second frequency component different form the first frequency component and for coding the first and second frequency components independently. The coding unit includes scalar quantization means for performing scalar quantization on the frequency information, extracting unit for extracting zero information from the scalar quantized frequency information which includes the second frequency components, and zero coding unit for coding the zero information.

Abstract: Improved method and apparatus for vector quantization (VQ) to build a codebook for the compression of data. The codebook or "tree" is initialized by establishing N initial nodes and creating the remainder of the codebook as a binary codebook. Children entries are split upon determination of various attributes, such as maximum distortion, population, etc. Vectors obtained from the data are associated with the children nodes, and then representative children entries are recalculated. This splitting/reassociation continues iteratively until a difference in error associated with the previous children and current children becomes less than a threshold. This splitting and reassociating process continues until the maximum number of terminal nodes is created in the tree, a total error or distortion threshold has been reached or some other criterion. The data may then be transmitted as a compressed bitstream comprising a codebook and indices referencing the codebook.

Abstract: Improved proprecessing and postprocessing for vector quantization, for example, for encoding an image. In one embodiment, the luminosity of the vectors in an image are used to determine the codes for vector quantization. In another embodiment, a median filter is performed to eliminate motion artifacts. In another embodiment, temporal filtering is applied once the difference between an earlier frame and a current frame exceeds a threshold. Embodiments are also provided for adaptive temporal filtering based on temporal "no change" blocks and their errors. Embodiments are provided for different regions of images which reference different codebooks, and regions of variable size. Embodiments are also provided for shared and variable size codebooks for different images or other data.

Abstract: A technique for encoding and decoding an image includes the following steps: subsampling the image to obtain a number of subsampled frames of spatially offset image-representative signals; transforming each of the subsampled frames to obtain a corresponding number of frames of transform coefficient-representative signals; forming vector-representative signals from corresponding coefficient-representative signals of the respective frames of coefficient-representative signals; performing a thresholding operation on the vector-representative signals to obtain thresholded vector-representative signals; vector quantizing the thresholded vector-representative signals to obtain encoded signals; storing the encoded signals; and decoding the encoded signals to obtain a recovered version of the image.

Abstract: A handwriting signal processing front-end method and apparatus for a handwriting training and recognition system which includes non-uniform segmentation and feature extraction in combination with multiple vector quantization. In a training phase, digitized handwriting samples are partitioned into segments of unequal length. Features are extracted from the segments and are grouped to form feature vectors for each segment. Groups of adjacent from feature vectors are then combined to form input frames. Feature-specific vectors are formed by grouping features of the same type from each of the feature vectors within a frame. Multiple vector quantization is then performed on each feature-specific vector to statistically model the distributions of the vectors for each feature by identifying clusters of the vectors and determining the mean locations of the vectors in the clusters. Each mean location is represented by a codebook symbol and this information is stored in a codebook for each feature.

Abstract: In a vector transform coding (VTC) scheme for image and video compression, a multi-layered codebook structure and a dynamic bit-allocation scheme in the vector transform domain are employed. The main advantage of this technique is that distortion is controlled by dynamically allocating more bits to vectors causing larger distortions and less bits to vectors causing smaller distortions.

Abstract: An image signal analyzing system capable of accomplishing shape analysis for processing or coding an image with easy, by space division using two orthogonal representative vectors, or by differential representative vectors given by the difference between representative vectors. With such a system, inner products between mean separated blocks produced by a mean separated unit and a differential representative vector set stored in a memory in advance are computed by inner product computing unit disposed into a binary tree shape having n stages. Positive/negative signs expressing resulting inner products in the respective stages of the binary tree computed by the inner product computing unit are held, as indexes, by n index holding means sequentially, and an index representing the shape of input image information is obtained from the n indexes held by the index holding unit.

Abstract: Before transmitting signals to a receiver, the signals are subjected to adaptive prediction to generate a residual signal for transmission, and the residual signal is then transformed into frequency domain coefficients, the coefficients are grouped together to form vectors, and the vectors are then quantized.

Abstract: Video data compression techniques reduce necessary storage size and communication channel bandwidth while maintaining acceptable fidelity. Vector quantization provides better overall data compression performance by coding vectors instead of scalars. The search algorithm and VLSI architecture for implementing it is herein disclosed, and such a search algorithm is useful for real-time image processing. The architecture employs a single processing element and external memory for storing the N constant value hyperplanes used in the search, where N is the number of codevectors. The design does not perform any multiplication operation using the constant value hyperplane tree search, since the tree search method is independent of any L.sub.q metric for q between one and infinity. Memory used by the design is significantly less than memory employed in existing architecture.

Abstract: A vector quantization method and apparatus which processes signal vectors constructed from digital image data (such as a digital video stream). The digital video stream is treated as a sequence of images, each represented by a rectangular array of pixels. The pixels are processed in blocks (typically 4.times.4 blocks) of two types, known as smooth blocks and detail blocks. The detail blocks contain the original image pixels, while the smooth blocks contain pixels from a subsampled version of the image. The pixels in the blocks are formatted into image vectors, again in two types. Each type of image vector is separately compressed using a vector quantization coder and a separate code book for each type. Smooth image vectors are compressed using a smooth code book and detail image vectors compressed using a detail code book.

Abstract: There is disclosed a method and apparatus for data coding for image transmission or storage in which vector quantization enables good image quality at reduced transmission or storage bit rates by directly employing spatial contiguity with a prior block of data, or a one-row-prior block of data in a way that directly controls intensity continuity between spatially close pixels of the adjacent blocks in two directions. The technique includes both "side-match" (no overlap) and "overlap-match" vector quantization. The technique is relative demanding of memory space and logic capacity in the encoding of the image information; but the decoding can be relatively simple and thus is well adapted for archiving and subsequent retrieval of two-dimensional information.

Abstract: A technique for encoding and decoding an image includes the following steps: subsampling the image to obtain a number of subsampled frames of spatially offset image-representative signals; transforming each of the subsampled frames to obtain a corresponding number of frames of transform coefficient-representative signals; forming vector-representative signals from corresponding coefficient-representative signals of the respective frames of coefficient-representative signals; performing a thresholding operation on the vector-representative signals to obtain thresholded vector-representative signals; vector quantizing the thresholded vector-representative signals to obtain encoded signals; storing the encoded signals; and decoding the encoded signals to obtain a recovered version of the image.