Abstract: An image retrieval device and an image retrieval method are provided which are capable of improving performance of image retrieval, of retrieving images at higher speed with simplified configurations and of retrieving images by simplified calculating processes.

Abstract: A system for coding images, and more particularly, to a system for compressing images to a reduced number of bits by employing a Discrete Cosine Transform (DCT) in combination with a visual model.

Abstract: A method of encoding a set of data representing physical quantities includes the steps of dividing the set of data into subsets, calculating a first encoding cost for each subset using a first encoding mode, calculating a second encoding cost for each subset using a second encoding mode, and selecting an encoding mode per subset as a function of the first and second encoding costs, in which the two encoding costs are calculated according to the same rate-distortion compromise (?), for the image overall.

Abstract: Methods and apparatus, including computer program products, for compression include decomposing an image into wavelet transform coefficients, applying a first quantizer to the wavelet transform coefficients, and applying a second quantizer to a set of wavelet transform coefficients representing a background of the image. The method can include rounding the wavelet transform coefficients and rounding the wavelet transform coefficients can include rounding to a nearest integer value. The method can include multiplying the set rounded quantized background wavelet transform coefficients by the second quantizer and rounding all wavelet transform coefficients. Rounding all wavelet transform coefficients can include rounding to a nearest integer value.

Abstract: The described embodiments provide a system that encodes a sequence of integers using a variable-length compression technique. During operation, the system scans the sequence of integers and observes the sizes of the integers to determine a threshold value, K, from the observed sizes. For a given integer of length N bits, if N?K is greater than zero, the system generates a tag for the encoded integer comprising a sequence of N?K zeros followed by a one, and generates a set of remaining bits for the encoded integer as a sequence of the N?1 least-significant bits which make up the integer. Otherwise, the system generates a tag for the encoded integer as a single one, and generates a set of remaining bits for the encoded integer by padding the N bits which make up the integer with zeros so that the set of remaining bits is K bits in length.

Abstract: The invention relates to an image coding apparatus which can improve the picture quality of an image. A quantization scale arithmetic operation section 43 calculates a coefficient of a quantization index based on a variance of values of a motion vector residual outputted from a motion vector residual variable buffer section 72 and an activity outputted from an activity buffer section 73. The quantization scale arithmetic operation section 43 calculates a quantization scale based on the coefficient of the quantization index calculated thereby and the quantization index outputted from the quantization index arithmetic operation section 42. The present invention can be applied to an image coding apparatus of the MPEG system.

Abstract: A method and an apparatus for compressing image data. The method includes dividing a line of an image into equal length fragments to form a coding unit, transforming and performing entropy coding to the coding unit, and compressing the image data based on the transformed entropy coded coding unit.

Abstract: This invention allows Huffman encoding using a common Huffman table according to basic quantization values Qi,j even when image data expressed by n bits falling within the range from L to K is JPEG-coded, and can suppress the Huffman table size from increasing. To this end, a basic quantization table storage unit stores quantization step values Q0,0 to Q7,7 used in baseline JPEG coding. A minimum quantization step generator outputs a minimum quantization step value Qn—min to a comparator/selector according to the number n of bits of each color component of image data to be coded. The comparator/selector compares the quantization step values Q0,0 to Q7,7 with Qn—min and selects larger ones, and outputs the comparison results to a quantizer as Q?i,j. The quantizer stores Q?i,j in a quantization table storage unit and quantizes orthogonal transformation coefficients output from a DCT transformer.

Abstract: A method for embedding digital watermark data in digital data contents includes the steps of obtaining a frequency coefficient of block data of digital data contents, obtaining a complexity of the block data, obtaining an amount of transformation of the frequency coefficient from the complexity and the digital watermark data, and embedding the digital watermark data by transforming the frequency coefficient. In addition, a method for reading digital watermark data includes the steps of calculating a probability of reading ‘1’ or ‘0’ in a read bit sequence by using a test method on the basis of binary distribution, determining the presence or absence of digital watermark data according to the probability, and reconstituting digital watermark data. Another method includes the steps of performing soft decision in code theory by assigning weights to the digital watermark sequence with a weighting function, and reconstituting digital watermark data.

Abstract: Methods and apparatus implementing systems and techniques for determining quantizer values for a signal. In general, in one implementation, a technique includes: obtaining a target distortion and a transformed signal having sub-bands resulting from a source signal being transformed by a wavelet-based decomposition; distributing the target distortion among the sub-bands based on energy weights to produce target sub-band distortions; ascertaining actual sub-band distortion values that result at different grades of quantization; determining quantizer values for the sub-bands, including interpolating between at least two of the quantization grades; and quantizing the sub-bands with the determined quantizer values. The signal may be a digital image. Additionally, in another implementation, determining the quantizer values can involve determining approximations of sub-band coefficient quantization error rather than performing the interpolation, in which case, the ascertaining operation need not be performed.

Abstract: An implementation of a technology is described herein for deriving robust non-local characteristics and quantizing such characteristics for blind watermarking of a digital good.

Abstract: An implementation of a technology is described herein for deriving robust non-local characteristics and quantizing such characteristics for blind watermarking of a digital good. This technology finds the proper balance between minimizing the probability of false alarms (i.e., detecting a non-existent watermark) and the probability of misses (i.e., failing to detect an existing watermark). The technology, described herein, performs quantization index modulation (QIM) based upon non-local characteristics of the digital good. Non-local characteristics may include statistics (e.g., averages, median) of a group of individual parts (e.g., pixels) of a digital good. This abstract itself is not intended to limit the scope of this patent. The scope of the present invention is pointed out in the appending claims.

Abstract: An aspect of an image signal transforming method is a method of generating one or more transformed samples from a plurality of input samples, which includes a first transformed sample generating step of performing a first filtering process by a filter, on at least one first input sample (an input sample from a terminal) out of a plurality of first input samples used for generation of a first transformed sample, to generate first filtered data, and performing a first arithmetic process (subtraction by a subtractor) on another first input sample not used for the generation of the first filtered data (an input sample from another terminal), and the first filtered data generated, to generate the first transformed sample.

Abstract: A method for compressing a high dynamic range (HDR) texture. A first block of texels of the HDR texture in a red-green-blue (RGB) space may be transformed to a second block of texels in a luminance-chrominance space. The first block may have red values, green values and blue values. The second block may have luminance values and chrominance values. The chrominance values may be based on a sum of the red values, a sum of the green values and a sum of the blue values. The luminance values and the chrominance values may be converted to an 8-bit integer format. The luminance values may be modified to restore a local linearity property to the second block. The second block may be compressed.

Abstract: A decoding apparatus includes a dequantization-value generating section, distribution-information acquiring section, and a correcting section. The dequantization-value generating section generates a plurality of dequantization values for each quantization index value. The distribution-information acquiring section acquires distribution information of original data corresponding to each quantization index. The correcting section corrects at least a portion of the dequantization values generated by the dequantization-value generating section, based on the distribution information acquired by the distribution-information acquiring section.

Abstract: A video predictive decoding method and apparatus for predicting a current block of a picture. The method includes storing at least one previous product in a memory. The previous product corresponds to a block of a plurality of blocks of the picture. The previous product is the product of a quantized AC coefficient and a quantization scale of the block that the previous product corresponds to. The method further includes determining which block to use as a prediction block from the plurality of blocks, reading from the memory at least one previous product corresponding to the prediction block, and calculating at least one quantized AC coefficient of the current block using the at least one previous product read from the memory.

Abstract: To improve the data compression ratio in the data compression/decompression method utilizing the down-sampling process. Image data in a DS area division unit (130) is divided into a plenty of DS areas and DS areas which are not important are converted into contracted data of the block size by down-sampling in a down-sampling unit (131). A relocation unit (132) allocates two or more contracted data in one DS area, inserts file data into the other portion in the DS area, and removes the DS area where no contracted data is allocated, thereby reducing the entire image size. The image data thus contracted is inputted into a JPEG encoder (14) and subjected to a DCT process and a quantization process for each block. When performing restoration, decompression is performed by the JPEG decoder and the original location is restored. As for the contracted data, interpolation is performed to restore the image data.

Abstract: A method of compressing an original dynamic range of an original image, includes a first step of obtaining a reduced image corresponding to the original image by performing quantization and downsampling on the original image that has been input; a second step of calculating a look-up table based on the reduced image, wherein the look-up table indicates a mapping relationship between the original dynamic range of the original image and a dynamic range of a medium; and a third step of mapping each of the pixels in the original image from the original dynamic range of the original image onto the dynamic range of the medium based on the look-up table.

Abstract: A filter for implementing Floyd Steinberg two-dimensional error diffusion algorithms allows high-speed processing of video and images. The filter is shown in direct form with proper bit precision with implementations that permit the filter to operate at high speed. Furthermore, a reduction in the gate count is achieved over the direct form. The results of static timing analysis obtained post synthesis are also summarized.

Abstract: Systems and methods are described that apply a watermark to data, such as data representing an image. In one implementation, the complexity of the image is measured. A quantization step size is calculated, based in part on the measured complexity of the image. A watermark or message is embedded into the image using the quantization step sizes derived for each coefficient of interest. In a further implementation, a mark decoding system is configured to extract the embedded message from the image data.

Abstract: A method for compression includes applying a discrete cosine transform to said image data to obtain transform coefficients, quantizing the transform coefficients by applying a quantization level scaled through a gain value, adjusting the gain value as a function of desired image parameters by executing a first time said quantization operation applying a first gain value and obtaining first quantized data, estimating statistically a second gain value suitable to obtain the desired image parameters, and executing a second time said quantization operation applying said second gain value. The operation of statistical estimation of the second gain value includes evaluating a threshold value as a function of the desired image parameters and setting to zero a percentage of coefficients of the first quantized data as a function of the threshold value.

Abstract: An implementation of a technology is described herein for deriving robust non-local characteristics and quantizing such characteristics for blind watermarking of a digital good.

Abstract: An electronic watermark generating apparatus color-converts an RGB image, and divides the color-converted image into a plurality of blocks. Then, the apparatus embeds to-be-embedded information as an electronic watermark into blocks in which an irreversible compression image has been quantized, the image being produced by carrying out quantization on a block-by-block basis, and restores the RGB image by carrying out de-quantization with respect to the blocks. The above apparatus color-converts the restored RGB image, and divides the color-converted image into a plurality of blocks. Thereafter, the apparatus carries out quantization on a block-by-block basis, and extracts to-be-embedded information from the thus quantized blocks. The apparatus embeds the to-be-embedded information as an electronic watermark again in the quantized blocks when it is not checked that the to-be-embedded information has been correctly extracted.

Abstract: A DWT unit applies wavelet transform to an input signal to output transform coefficients, and a quantization unit quantizes those transform coefficients with a quantization step size determined according to target image quality. Then, a rate control unit controls the rate of coded data according to information on the quantization step size. Also, an image-quality control unit controls the rate of only part of coded data which is determined from a priority table. This achieves a compression encoder which operates at high speed with minimal operations.

Abstract: An image processing device includes a memory that is adapted to store an input image; an inverted image generator that is adapted to generate an inverted image, the inverted image corresponding to a predetermined part of the input image; a target image generator that is adapted to generate a target image, the target image including the inverted image and the input image; a quantization unit that is adapted to quantize the target image, in accordance with the error diffusion method; an image extractor that is adapted to extract an output image from the quantized target image, the output image corresponding to the input image; an output unit that is adapted to output the output image extracted by the image extractor.

Abstract: A transform coding system and method are disclosed which utilize a modified quantization technique which advantageously foregoes the need for inverse quantization at the decoder. New techniques for optimizing an entropy code for the modified quantizer and for constructing the entropy codes are also disclosed.

Abstract: An image compression and decompression method encodes and decodes pixel data based on a color conversion method. Base on the relationships of corresponding color components of two adjacent pixels, the corresponding color components are encoded either by a white and black modification, a down-sampling or an edge modification. Based on the relationships of encoded color components of the two adjacent pixels, the corresponding encoded color components are decoded either by an inverse white and black modification, an up-sampling or an inverse edge modification.

Abstract: An encoding device 100 includes a quantization parameter generating circuit 111 that generates a provisional quantization parameter, a quantizing circuit 121 that generates quantized data by quantizing a signal to be quantized on the basis of the provisional quantization parameter, a binarizing circuit 131 that binarizes the quantized data to output binary symbol data, an arithmetic coding circuit 141 that generates coded data by arithmetic-coding the binary symbol data, a quantization parameter calculating circuit 112 that generates a suitable quantization parameter on the basis of a symbol amount of the binary symbol data, a code amount of the coded data, an upper limit of the symbol amount, and an target code amount, a quantizing circuit 122 that quantizes the signal to be quantized on the basis of the suitable quantization parameter.

Abstract: An image encoding apparatus includes a quantization control section that supplies a rounded Q-scale to a quantization section for quantizing each macroblock according to the rounded Q-scale. A Q-scale is generated for each macroblock based on the amount of generated codes produced by encoding. The Q-scale is rounded within a set range by a rounding control section. The rounded Q-scale is smaller in amount than the Q-scale. Since the same Q-scale values will continue in succession, these can be omitted from a stream, allowing the reduced amount of quantization information and the increased amount of information allocated to images.

Abstract: In an encoding apparatus or an encoding method, motion-image data is input, an encoding parameter is output such that a predetermined number of codes are used for encoding the input motion-image data in units of predetermined sizes, the output encoding parameter is stored in a storage medium, and the input motion-image data is encoded by adaptively selecting the output encoding parameter or the encoding parameter stored in the storage medium.

Abstract: This disclosure describes rate control techniques that can improve video encoding. In particular, the described rate control techniques exploit relationships between the number of bits encoded per frame and the number of non-zero coefficients of the video blocks after quantization. The number of number of non-zero coefficients of the video blocks after quantization is referred to as rho (?). The value of ? is generally proportional to the number of bits used in the video encoding. This disclosure utilizes a relationship between ? and a quantization parameter (QP) in order to achieve rate controlled video encoding. More specifically, this disclosure provides techniques for generating a lookup table (LUT) that maps values of ? to different QPs.

Abstract: Techniques are generally described for detecting double JPEG compression in images. Example detection techniques of double JPEG compression may include receiving JPEG images for analysis and extracting 2-dimensional (2-D) arrays of JPEG coefficients from the images. 2-D difference arrays may be generated from the array of JPEG coefficients, with the entries in the difference array reflecting relative changes in values of pairs of entries in the array of JPEG coefficients. The detection techniques also model the difference arrays using random processes, and evaluate whether the random processes reveal statistical artifacts in the JPEG images. These statistical artifacts result from double JPEG impression performed on the JPEG images.

Abstract: This invention enables to generate encoded data without noticeable image quality degradation when reproducing an image at a lower resolution not to mention the original resolution. To accomplish this, when setting is done to transmit an image captured by a digital camera to a network, code stream forming information CF is set to “2” to arrange the encoded data of each tile in a resolution order. To suppress image quality degradation when reproducing at an intermediate resolution, stream conversion information SC is set to “2”. When encoding image data in compression processing, block overlap processing of suppressing discontinuity of data at the boundary between adjacent blocks is executed as many times as the count set in the stream conversion information. The obtained encoded data is arranged in accordance with the code stream forming information CF and output.

Abstract: A video encoding method comprises selecting one combination, for each block of an input video signal, from a plurality of combinations each including a predictive parameter and at least one reference picture number determined in advance for the reference picture, generating a prediction picture signal in accordance with the reference picture number and predictive parameter of the selected combination, generating a predictive error signal representing an error between the input video signal and the prediction picture signal, and encoding the predictive error signal, information of the motion vector, and index information indicating the selected combination.

Abstract: The method, system, and apparatus of a shared error resiliency path through coefficient reordering is disclosed. In on embodiment, determining that an input data is data-partitioned, performing a discrete cosine transform and a quantization of the input data to form a quantized data, separating a first coefficient representing a DC coefficient of the quantized data for each block of the input data, rearranging other coefficients representing AC coefficients of the quantized data for each block of the input data in a fashion produces a zig-zag scan output similar to that of a non-data-partitioned data, bypassing a DC encoding module, determining whether any of the rearranged AC coefficients of the quantized data need to be encoded, performing a zig-zag scan on the rearranged AC coefficients of the quantized data when they need to be encoded, and encoding the rearranged AC coefficients of the quantized data based on the zig-zag scan.

Abstract: An image encoding method is for obtaining, from a pixel data string containing pixel values of a plurality of pixels in a pixel array where the pixels are arrayed, a pixel data string containing a quantized representative value into which at least one of the pixel values is encoded. Encoding is performed by allocating a specific quantized representative value when a pixel value of a pixel is a specific pixel value, and by allocating a quantized representative value other than the specific quantized representative value when the pixel value is other than the specific pixel value.

Abstract: Techniques and tools for content classification and adaptive quantization are described. In an example implementation, a video encoding tool classifies blocks as textured, dark smooth or other smooth. The tool classifies a block as textured or non-textured by comparing the energy of AC coefficients for the block to a texture threshold, which can be set using a non-linear mapping of possible texture classification levels to possible texture thresholds. If a block is not textured, the tool further classifies the block as dark smooth or smooth depending on average intensity value for the block. Using the classification information and one or more control parameters to control bit allocation for dark smooth content relative to other smooth content, the tool encodes the video and outputs encoded video information. Example multi-pass approaches to setting the control parameters are also described.

Abstract: An image processor includes a quantization unit receiving first data before quantization and outputting second data after quantization, a prediction unit obtaining a difference value between the second data and third data being prediction data and outputting the difference value as fourth data, and an encoding unit encoding the fourth data. The quantization unit includes a first processing unit dividing the first data by a quantization coefficient, so as to obtain fifth data including a fraction as a result of division and a second processing unit rounding up or rounding off the fraction such that a value of the fourth data becomes smaller based on comparison between the third data and the fifth data, so as to obtain the second data.

Abstract: A video encoding method comprises selecting one combination, for each block of an input video signal, from a plurality of combinations each including a predictive parameter and at least one reference picture number determined in advance for the reference picture, generating a prediction picture signal in accordance with the reference picture number and predictive parameter of the selected combination, generating a predictive error signal representing an error between the input video signal and the prediction picture signal, and encoding the predictive error signal, information of the motion vector, and index information indicating the selected combination.

Abstract: A video encoding method comprises selecting one combination, for each block of an input video signal, from a plurality of combinations each including a predictive parameter and at least one reference picture number determined in advance for the reference picture, generating a prediction picture signal in accordance with the reference picture number and predictive parameter of the selected combination, generating a predictive error signal representing an error between the input video signal and the prediction picture signal, and encoding the predictive error signal, information of the motion vector, and index information indicating the selected combination.

Abstract: A video encoding method comprises selecting one combination, for each block of an input video signal, from a plurality of combinations each including a predictive parameter and at least one reference picture number determined in advance for the reference picture, generating a prediction picture signal in accordance with the reference picture number and predictive parameter of the selected combination, generating a predictive error signal representing an error between the input video signal and the prediction picture signal, and encoding the predictive error signal, information of the motion vector, and index information indicating the selected combination.

Abstract: An inputted digital signal of a first format (DV video signal) is restored to a variable-length code by having its framing cancelled by a de-framing section 11, then decoded by a variable-length decoding (VLD) section 12, inversely quantized by an inverse quantizing (IQ) section 13, and inversely weighted by an inverse weighting (IW) section 14. Then, required resolution conversion in the orthogonal transform domain (frequency domain) is carried out on the inversely weighted video signal by a resolution converting section 16. After that, the video signal having the resolution converted is weighted by a weighting (W) section 18, then quantized by a quantizing (Q) section 19, coded by variable-length coding by a variable-length coding (VLC) section 20, and outputted as a digital signal of a second format (MPEG video signal).

Abstract: A pattern recognition method comprises steps of inputting a pattern of a recognition object performing feature extraction from the input pattern to generate a feature vector, increasing the number of quantization in an order from quantization number 1 or quantization number 2 to calculate a quantization threshold of each of the quantization number, wherein the quantization threshold of quantization number (n+1) using a quantization threshold of quantization number n (n>=1) is calculated and a quantization function having a quantization threshold corresponding to quantization number S (S>n) is generated, quantizing each component of the feature vector of the input pattern using the quantization function to generate an input quantization feature vector having each of the quantized component, storing a dictionary feature vector of the recognition object, or a quantized dictionary feature vector in which each component of the dictionary feature vector of the pattern of a recognition object is quantized; cal

Abstract: In an image forming system, an information processing apparatus (host computer) outputs print data to a printer connected to the apparatus through an interface. The apparatus includes an interface identifying module for identifying the type of the interface, a query module for determining, based on the identified interface type, a quantization method for quantizing print data and/or a compression method, and an image processing module for performing quantization and/or compression on the print data by the determined quantization method and/or compression method. The apparatus outputs, to the printer, the print data processed by the image processing module.

Abstract: A bit stream is decoded to generate first quantized DCT-coefficients. The first quantized DCT-coefficients are inversely quantized using a first quantization table, thereby generating DCT-coefficients. The resulting DCT-coefficients are quantized using a second quantization table, thereby generating second quantized DCT-coefficients. Elements in the second quantization table are smaller in value than those in the first quantization table. Predetermined additional information is embedded into the second quantized DCT-coefficients. The second quantized DCT-coefficients having the additional information embedded therein are encoded to produce a bit stream having the additional information embedded therein. The second quantization table is set to be smaller in value than the first quantization table.

Abstract: A check processing method involves capturing at a first compression rate certain portions of a received check, and evaluating, based on that captured image data, whether certain information of the check can be reproduced by image data at a second compression rate that is greater than the first compression rate. If so, then image data representing the subject portions of the check is captured at the second compression rate image data, stored, and used to electronically process a corresponding check payment transaction. If not, then a specific process is executed without using image data at the second compression rate. This processing step outputs a specific signal or stores image data containing representing the subject portions of the received check at the first compression rate. Such process may be embodied in a check processing apparatus and may be carried out in response to a program of instructions.

Abstract: With adaptive multiple quantization, a video or other digital media codec can adaptively select among multiple quantizers to apply to transform coefficients based on content or bit rate constraints, so as to improve quality through rate-distortion optimization. The switch in quantizers can be signaled at the sequence level or frame level of the bitstream syntax, or can be implicitly specified in the syntax.

Abstract: In a method of digital image compression, the spatial frequency content of a digital image is characterized using a relatively sparse sampling of data blocks from the image, and a quantization table is constructed based on the characterization of the spatial frequency content.

Abstract: In uniform superposition of two predictions toward a weighted superposition, such as in coding B frames, weighting is based on reliability of a reference picture used for determination of the prediction. Since there is a relationship between the reliability and the respective quantization parameter, coupling to the quantization parameter that is also known in the receiver also can be effected advantageously.

Abstract: A method of transforming a n-bit data packet to a m-bit data packet with a lookup table. The lookup table includes at least one entry data packet and at least one respective delta value associated with each entry data packet. The method includes the acts of receiving an input data packet having n-bits, indexing the lookup table with at least a portion of the input data packet to obtain one of the at least one entry data packet, and decompressing the obtained entry data packet with the at least one respective delta value associated with the obtained entry data packet, thereby resulting in an output data packet having m-bits. The decompressing act includes using a portion of the input data packet to determine the number of delta values called for decompressing the obtained entry data packet. The method can be used in, for example, an image processor.