Showing papers on "Quantization (image processing) published in 1993"

Visual optimization of dct quantization matrices for individual images

Abstract: Many image compression standards (JPEG, MPEG, H.261) are based on the Discrete Cosine Transform (DCT). However, these standards do not specify the actual DCT quantization matrix. We have previously provided mathematical formulae to compute a perceptually lossless quantization matrix. Here I show how to compute a matrix that is optimized for a particular image. The method treats each DCT coefficient as an approximation to the local response of a visual 'channel'. For a given quantization matrix, the DCT quantization errors are adjusted by contrast sensitivity, light adaptation, and contrast masking, and are pooled non-linearly over the blocks of the image. This yields an 8x8 'perceptual error matrix'. A second non-linear pooling over the perceptual error matrix yields total perceptual error. With this model we may estimate the quantization matrix for a particular image that yields minimum bit rate for a given total perceptual error, or minimum perceptual error for a given bit rate. Custom matrices for a number of images show clear improvement over image-independent matrices. Custom matrices are compatible with the JPEG standard, which requires transmission of the quantization matrix.

Visually optimal DCT quantization matrices for individual images

Abstract: A custom quantization matrix tailored to a particular image is designed by an image-dependent perceptual method incorporating solutions to the problems of luminance and contrast masking, error pooling and quality selectability. >

FBI wavelet/scalar quantization standard for gray-scale fingerprint image compression

Abstract: The FBI has recently adopted a standard for the compression of digitized 8-bit gray-scale fingerprint images. The standard is based on scalar quantization of a 64-subband discrete wavelet transform decomposition of the images, followed by Huffman coding. Novel features of the algorithm include the use of symmetric boundary conditions for transforming finite- length signals and a subband decomposition tailored for fingerprint images scanned at 500 dpi. The standard is intended for use in conjunction with ANSI/NBS-CLS 1-1993, American National Standard Data Format for the Interchange of Fingerprint Information, and the FBI's Integrated Automated Fingerprint Identification System.© (1993) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Rate-constrained picture-adaptive quantization for JPEG baseline coders

Abstract: A recursive algorithm is presented for generating quantization tables in JPEG (Joint Photographic Experts Group) baseline coders from the actual statistics of the input image. Starting from a quantization table with large step sizes, corresponding to low bit rate and high distortion, one entry of the quantization table is updated at a time so that, at each step, the ratio of decrease in distortion to increase in bit rate is approximately maximized. This procedure is repeated until a target bit rate is reached. Simulation results demonstrate that, with picture-adaptive quantization tables designed by the proposed algorithm, the JPEG DCT (discrete cosine transform) coder is able to compress images with better rate-distortion performance than that achievable with conventional empirically designed quantization tables. >

A process-pipeline architecture for image/video processing

Abstract: A sequential process-pipeline (12) has a first processing stage (30) coupled to a CODEC (24) through a plurality of buffers, including an image data input buffer (28), an image data output buffer (26), and an address buffer (34). The address buffer stores addresses, each of which identifies an initial address of a block of addresses within an image memory (22). Each block of addresses in the image memory stores a block of decompressed image data. A local controller (18) is responsive to the writing of an address into the address buffer to initiate the operation of the CODEC to execute a Discrete Cosine Transformation Process and a Discrete Cosine Transformation Quantization Process.

Center-cut for color-image quantization

Abstract: Three simple and effective modifications made to the median-cut algorithm lead to a new and useful tool for color-image quantization. The first is to select the color box with the longest longest-dimension instead of the highest pixel count. The second is to cut through the center of the color box instead of bisecting the box's pixel count. The third is to use 3-2-4 instead of 3-3-3 bit-cutting for color reduction.

Local cosine transform: a method for the reduction of the blocking effect in JPEG

Abstract: This paper presents the local cosine transform (LCT) as a new method for the reduction and smoothing of the blocking effect that appears at low bit rates in image coding algorithms based on the discrete cosine transform (DCT). In particular, the blocking effect appears in the JPEG baseline sequential algorithm. Two types of LCT were developed: LCT-IV is based on the DCT type IV, and LCT-II is based on DCT type II, which is known as the standard DCT. At the encoder side the image is first divided into small blocks of pixels. Both types of LCT have basis functions that overlap adjacent blocks. Prior to the DCT coding algorithm a preprocessing phase in which the image is multiplied by smooth cutoff functions (or bells) that overlap adjacent blocks is applied. This is implemented by folding the overlapping parts of the bells back into the original blocks, and thus it permits the DCT algorithm to operate on the resulting blocks. At the decoder side the inverse LCT is performed by unfolding the samples back to produce the overlapped bells. The purpose of the multiplication by the bell is to reduce the gaps and inaccuracies that may be introduced by the encoder during the quantization step. LCT-IV and LCT-II were applied on images as a preprocessing phase followed by the JPEG baseline sequential compression algorithm. For LCT-IV, the DCT type IV replaced the standard DCT as the kernel of the transform coding. In both cases, for the same low bit rates the blocking effect was smoothed and reduced while the image quality in terms of mean-square error became better. Subjective tests performed on a group of observers also confirm these results. Thus the LCT can be added as an optional step for improving the quality of existing DCT (JPEG) encoders. Advantages over other methods that attempt to reduce the blocking effect due to quantization are also described.

Tree-structured vector quantization of CT chest scans: image quality and diagnostic accuracy

Abstract: The authors apply a lossy compression algorithm to medical images, and quantify the quality of the images by the diagnostic performance of radiologists, as well as by traditional signal-to-noise ratios and subjective ratings. The authors' study is unlike previous studies of the effects of lossy compression in that they consider nonbinary detection tasks, simulate actual diagnostic practice instead of using paired tests or confidence rankings, use statistical methods that are more appropriate for nonbinary clinical data than are the popular receiver operating characteristic curves, and use low-complexity predictive tree-structured vector quantization for compression rather than DCT-based transform codes combined with entropy coding. The authors' diagnostic tasks are the identification of nodules (tumors) in the lungs and lymphadenopathy in the mediastinum from computerized tomography (CT) chest scans. Radiologists read both uncompressed and lossy compressed versions of images. For the image modality, compression algorithm, and diagnostic tasks the authors consider, the original 12 bit per pixel (bpp) CT image can be compressed to between 1 bpp and 2 bpp with no significant changes in diagnostic accuracy. The techniques presented here for evaluating image quality do not depend on the specific compression algorithm and are useful new methods for evaluating the benefits of any lossy image processing technique. >

High-definition video encoding system having color-sensitive quantization

Abstract: A video signal encoder uses an encoding system such as that developed by the Moving Picture Experts Group (MPEG). A key component of this encoding system increases the quantization step size of image data, thus decreasing its quantization resolution, to reduce the number of bits used to encode the data. Apparatus according to the present invention, monitors the color information being encoded and inhibits or reduces any increase in quantization step size for the luminance component and the color component of the image when signals representing saturated or nearly saturated red image components are being encoded.

Attractor image compression with a fast non-iterative decoding algorithm

Abstract: The problem of decoder convergence speed in an attractor (fractal-based) image compression system is considered. The general principles of such a system are described in the context of discrete linear spaces. A problem with previous implementations is that the decoding rests on an iterative procedure whose complexity is image-dependent. It is shown how this problem can be sidestepped by introducing a modification providing an iteration-free decoder. The coding results are not affected by this modification. The new decoding algorithm requires only one multiplication and three additions per pixel. The algorithm is suitable for applications where fast display of images is essential, such as retrieval of images from compressed databases. >

Apparatus for compressing and decompressing image data

Abstract: The present invention relates to an apparatus for compressing and decompressing the digital image data designed for minimizing degradation of the image quality and for improving the compression efficiency. Discriminative over the conventional package quantization method based on the statistical characteristics of an image, the invention is configured such that the image is firstly segmented into predetermined number of sub-blocks by the block segmentator, and each divided blocks are adopted by the different compression technique in accordance with the shape or complicity of the image.

Image processing device for processing grey level images

Abstract: An image processing device includes: a DOG filtering means for applying a DOG filter upon the grey level image of an object; and a tri-level thresholding means for thresholding the output of the DOG filtering means at two distinct threshold levels, thereby obtaining a tri-level image of the object. The 0-level regions of the tri-level image correspond to regions of the original image, such as the surface areas of the object, within which the variation of the grey level is small. The boundary between two 0-regions is marked by adjacent parallel strips of a (+)-region and a (-)-region, wherein the zero-crossing line between the (+)-region and the (-)-region clearly defines the contour line of the 0-regions. This method of tri-level quantization provides a basis for efficient template matching of the images of objects, recognition or identification of objects, the detection of movement vectors of objects, and the detection of the parallax of a continuously moving objects.

Optimum DCT quantization

Abstract: The paper offers a solution to the problem of determining good quantization tables for use with the discrete cosine transform. Using the methods proposed, the designer of a system can choose a selection of test images and a coefficient weighting scenario, from which a quantization table can be produced, optimized for the choices made. The method is based on simulated annealing searches which the space of quantization tables to minimize some chosen measure. >

A video-rate JPEG chip set

Abstract: A two-chip set has been designed, fabricated and is fully functional which performs the baseline JPEG image compression and decompression algorithm. The major functions of the devices include: DCT and IDCT, forward and inverse quantization, Huffman coding and decoding. The devices operate with pixel rates beyond 30 MHz at 70 degrees C and 4.75 V. Each die is less than 10 mm on a side and was implemented in a 1.0 µ CMOS cell-based technology to achieve a 9 man-month design time.

Image coding/decoding apparatus for efficient processing by sharing members in coding/local decoding and decoding processing

Abstract: An image coding/decoding apparatus intended for efficient processing by sharing members in coding, local decoding, and decoding processing. Processes such as DCT and IDCT, zigzag scan conversion and inversion, or quantization and inverse quantization performed in coding, local decoding, and decoding processing are similar to each other. DCT/IDCT, zigzag scan conversion/inversion, and quantization/inverse quantization are provided where either of the two functions can be selected for execution in synchronization with the processing timing for each block. Since the time required for one process for data for each block is very short, overall processing is not affected even if the members are used in a time division manner. By sharing processing, the hardware scale can be made small and by using a data bus in a time division manner, an external data bus can also be eliminated.

Marginal analysis prioritization for image compression based on a hierarchical wavelet decomposition

Abstract: A novel approach for jointly optimizing scalar quantization and tree-based quantization of hierarchical wavelet decompositions is presented. An image compression algorithm is developed around this approach, utilizing a pruned-tree image representation. Marginal analysis is applied to optimize jointly the pruned-tree representation and scalar quantization. Simulation results demonstrate that the proposed algorithm offer substantially improved signal-to-noise ratio at matching bit rates, compared with similarly structured compression algorithms. >

On-line adaptive vector quantization with variable size codebook entries

Abstract: A new image compression algorithm employs some of the most successful approaches to adaptive lossless compression to perform adaptive on-line (single pass) vector quantization. The authors have tested this algorithm with a host of standard test images (e.g. gray scale magazine images, medical images, space and scientific images, fingerprint images, and handwriting images) and with no prior knowledge of the data or training, for a given fidelity the compression achieved typically equals or exceeds that of the JPEG standard. The only information that must be specified in advance is the fidelity criterion. >

Input modification system for multilevel dithering

Abstract: This disclosure relates to an image processing system which relies upon quantization and dithering techniques to enable an output device, which has a given number of output levels, to accurately reproduce a image which is generated by an input device, which has a greater or equal number of input levels. Generally, neither the number of input nor output levels need to be a power of two. The present invention is implemented in a number of different embodiments. These embodiments generally rely upon an image processor which, depending on the particular implementation, includes memory devices and an adder, a comparator, or a bit shifter. Additional embodiments use an image adjustment system or an image modification system to refine the raw input levels of the input device, in order to create an improved output image. Also, the particular embodiments of the image processors can be used in connection with imaging systems having bi-tonal, monochromatic, or color input and output devices.

Image compression coder having improved scale factor control

Abstract: Disclosed is an image compression coding device in which time domain image data is transformed to frequency domain data and then compressed by multiplying the frequency domain data by quantization coefficients scaled by a scale factor based on a non-linear relationship of log(ACVdata) versus log(SF). One or more statistical passes are made through the frequency domain data using initial scale factors in which ACV data is obtained based on the newly defined relationship. The New Scale Factor (NSF) is then calculated based on the initial scale factors (ISF1, ISF2), the accumulated code volumes for the two scale factors (ACVdata1, ACVdata2), and the target code volume (TCV) of the compressed file.

Mathematical model to quantify JPEG block artifacts

Abstract: A method for quantifying the magnitude of block artifacts caused by JPEG type of image compression has been developed. The behavior of block artifacts as a function of compression ratio and edge enhancement operations can now be mathematically tracked. This mathematical technique is found to be useful for development of image compression techniques with regard to eliminating block artifacts.

Digital image processor for color image transmission

Abstract: The present invention improves either the processing speed or circuit complexity (or both) of a digital image processor by eliminating the conventional multiplication operation from the RGB to YUV and YUV to RGB transforms having NTSC compatible equation coefficients. In particular, the change of color-basis matrix for converting RGB data to obtain YUV data is factored into a scale factor and a coefficient matrix. The coefficient matrix contains coefficients such that its product with the RGB input vector can be performed using only seven binary additions and five shift operations. This process effectively removes the conventional multiplication operations from the RGB to YUV conversion. The scale factor is conveniently absorbed into the quantization of the YUV data before transmission. The quantization process includes a multiplication already and the two multiplication steps are folded into one step. Therefore, the absorption of the scale factor into the quantization step does not increase the processing time because no multiplication steps are added. This process of factoring the transform matrix is also applied to the inverse transform process. In the inverse transform process the change of color-basis matrix for converting YUV data to RGB data is such that its implementation reduced to five addition steps and two shift steps, again effectively eliminating the conventional multiplication operations. Also, the diagonal matrix is absorbed into the descaling process so that no multiplication steps are added to the inverse transform process.

Picture data compression processing method

Abstract: PURPOSE:To improve data compressibility while maintaining the picture quality of the important part of pictures by raising a bit number at the time of performing quantization for the highly important part of an object in the pictures and lowering the bit number for the part not so important. CONSTITUTION:Wavelet transformation is repeated N times to an original picture data Sorg and wavelet transformation coefficient signals WW0-WWN, WV0-WVN, VW0-VWN and VVN are obtained. Picture element numbers in respective main and sub directions are (1/2)N for the coefficient signals compared to the original picture data Sorg. The pictures indicated by the respective transformation coefficient signals WVi, VWi and WWi are the ones for which the source pictures are reduced and labeling executed to the coefficient signal WWN corresponds to the respective transformation coefficient signals as well. Thus, at the time of the quantization of the transformation coefficient signals, the bit number of the quantization is changed corresponding to the result of the labeling. Then, for the respective transformation coefficient signals, the picture quality is maintained by raising the bit number for the important part, the bit number is lowered for the unimportant part and the overall compressibility is improved.

Adaptive DCT coding of video signals

Abstract: A coefficient-by-coefficient adaptive discrete cosine transform (DCT) is investigated. The interframe DCT has been known to possess an adverse effect in that it generates mosquito and blocking noise. By analyzing DCT for the intraframe and the interframe video signals, a new adaptive intra/interframe DCT is developed. Cases without and with the movement-compensation are discussed separately. Through simulations, it is shown that a substantial improvement gain (1.0-5.0 dB/sub p-p/ or 0.2-2.0 dB/sub p-p/, respectively) can be obtained by the new adaptive method. The improvement is brought about by the reduction of the blocking noise conventionally generated by the coarse quantization of the higher order interframe DCT coefficients. By calculating the improvement bound, a further possibility is discussed. >

An adaptive BTC-VQ image compression algorithm operating at a fixed bit-rate

Abstract: Block Truncation Coding-Vector Quantization (BTC-VQ) is a simple, fast, non-adaptive block-based image compression technique with a moderate bit-rate of 1.0 bit/pixel. By making the algorithm adaptive it is possible to further lower the bit-rate. In this paper we show how the algorithm may optimally adapt itself to the local nature of the picture while keeping the bit-rate fixed at a given value. Examples are given of image compressions down to 0.65 bit/pixel. We further show how the adaptation process may be carried out in a fast and efficient manner.

An evaluation of JPEG compression for on-line satellite images transmission

Abstract: Image compression is a key technology to realize on-line satellite image transmission economically and quickly Among various image compression algorithms, the JPEG algorithm is the international standard for still color image compression In this study, various kinds of satellite images were compressed with the JPEG algorithm The relation between compression ratio and image quality were evaluated As for the image quality evaluation, both subjective evaluation and objective evaluation were performed It was determined that all the test satellite images could be compressed to at least 1/10 of the original data volume preserving high visual image quality The degradation of spatial distribution quality of the compressed images were evaluated using power spectrum of original and compressed images >

On the optimal transform for vector quantization of images

Abstract: Transform coding is widely used for image compression. If vector quantization is used in the transform domain, the definition of the optimal transform should be modified because the performance of vector quantization is better when the components within each vector are more correlated. The attributes of the optimal transform for vector quantization are discussed. It is shown how to construct the optimal transform for vector quantization based on the first order Markov model of images. >

Image coding/decoding apparatus

Abstract: An image coding/decoding apparatus intended for efficient processing by sharing members in coding, local decoding, and decoding processing. Processes such as DCT and IDCT, zigzag scan conversion and inversion, or quantization and inverse quantization performed in coding, local decoding, and decoding processing are similar to each other. DCT/IDCT, zigzag scan conversion/inversion, and quantization/inverse quantization are provided where either of the two functions can be selected for execution in synchronization with the processing timing for each block. Since the time required for one process for data for each block is very short, overall processing is not affected even if the members are used in a time division manner. By sharing processing, the hardware scale can be made small and by using a data bus in a time division manner, an external data bus can also be eliminated.

A video coding scheme with a high compression ratio for consumer digital VCRs

Abstract: A video coding scheme based on an intra-2frame DCT (discrete cosine transform) coding is described. This scheme is intended to reduce the bit rate of the 4:2:2 component signal down to 12.5 Mb/s without relying on motion estimation and compensation. The algorithm has been successfully simulated and achieved a reasonable quality for consumer use of digital VCRs. >