Showing papers on "Quantization (image processing) published in 1990"
••
TL;DR: A new method for filling a color table is presented that produces pictures of similar quality as existing methods, but requires less memory and execution time.
Abstract: A new method for filling a color table is presented that produces pictures of similar quality as existing methods, but requires less memory and execution time. All colors of an image are inserted in an octree, and this octree is reduced from the leaves to the root in such a way that every pixel has a well defined maximum error. The algorithm is described in PASCAL notation.
347 citations
••
03 Jan 1990-Graphical Models \/graphical Models and Image Processing \/computer Vision, Graphics, and Image Processing
TL;DR: A new automatic peak detection algorithm is developed and applied to histogram-based image data reduction (quantization) and the results of using the proposed algorithm for data reduction purposes are presented in the case of various images.
Abstract: A new automatic peak detection algorithm is developed and applied to histogram-based image data reduction (quantization). The algorithm uses a peak detection signal derived either from the image histogram or the cumulative distribution function to locate the peaks in the image histogram. Specifically, the gray levels at which the peaks start, end, and attain their maxima are estimated. To implement data reduction, gray-level thresholds are set between the peaks, and the gray levels at which the peaks attain their maxima are chosen as the quantization levels. The results of using the proposed algorithm for data reduction purposes are presented in the case of various images.
222 citations
••
IBM1
TL;DR: In this paper, a fast algorithm for computing scaled output of the 2D discrete cosine transform (DCT) on 8 x 8 points is presented, where the DCT itself need not he computed, but rather a scalar multiple DCT with appropriate compensation incorporated into the scaling.
Abstract: The Discrete Cosine Transform (DCT) followed by scaling and quantiza.tion is an ml- portant operation in image processing. Because of the scaling, the DCT itself need not he computed, but rather a scalar mu]tiple of the DCT might do, with appropriate compensationincorporated into the scaling. We present a fast method for computing such scaled output ofthe 2-dimensional DCT on 8 x 8 points. We also present a similar algorithm for the inversescaled DCT.1. INTRODUCTIONThe discrete cosine transform (DCT) plays an important role in digital image processing.Of particular interest is the two-dimensional DCT followed by scaling and quantiza.tion. This has applications in data compression of continuous tone images [1, 11]. Because the DCT is so often used, research into fast algorithms for its implementation has been rather active[3, 4, 5, 7, 8, 9, 10, 12, 13, 14, 15].Given an array y(k) , 0 k K — 1, of input data, its one-dimensional DCT output is K—i (irn(2k+1)\ y(n) = c(n) cost i y(),
63 citations
•
11 Jul 1990TL;DR: In this paper, a 3D discrete cosine transform (DCT) was proposed to remove both spatial and temporal redundancy of a sequence of image frames to achieve high bandwidth compression.
Abstract: A three dimensional (3D) discrete cosine transform (DCT) uses one dimensional DCT networks for transforming and inverse-transforming blocks of data, such as image data. The 3D DCT configuration uses DCT transform coding to remove both the spatial and temporal redundancy of a sequence of image frames to achieve high bandwidth compression.
51 citations
•
09 Oct 1990TL;DR: In this article, a method of unscreening a digitally created halftone image to reconstruct a continuous tone image was proposed, including the determination of the parameters of the haloftone screen used to generate the halftones image, logically filtering the halftsone image, and smoothing the continuous tone levels of the reconstructed image to minimize the quantization errors introduced during the original screening or dithering process.
Abstract: The present invention is a method of unscreening a digitally created halftone image to reconstruct a continuous tone image, including the determination of the parameters of the halftone screen (14) used to generate the halftone image, logically filtering the halftone image (18) to determine approximate continuous tone levels, and optionally, smoothing (22) the continuous tone levels of the reconstructed image to minimize the quantization errors introduced during the original screening or dithering process.
37 citations
•
01 Mar 1990TL;DR: In this paper, an extractor extracts, from image data to be binarized, a feature amount of a target pixel including at least a maximum density difference among pixels in a local region including the target pixel.
Abstract: An extractor extracts, from image data to be binarized, a feature amount of a target pixel including at least a maximum density difference among pixels in a local region including the target pixel. A predictor predicts an image type of the target pixel in accordance with a minimum density value among the pixels in the local region. A selector selects one of the feature amount of the target pixel from the extractor and a feature amount obtained by adding the feature amount of the target pixel to feature data of neighboring pixels in accordance with the image type predicted by the predictor. A discriminator discriminates the image type of the target pixel in accordance with the feature amount selected by the selector. A weighting section weights the discrimination result from the discriminator with respect to the neighboring pixels of a predetermined range preceding the target pixel in accordance with distances from the target pixel to the neighboring pixels so as to generate feature data of the neighboring pixels selectively used by the selector. A determinator adaptively determines a threshold value for binarizing the image data in accordance with the discrimination result. A binarizing section binarizes the image data in units of target pixels by using the threshold value determined by the determinator.
22 citations
••
01 Nov 1990TL;DR: A DCT algorithm with adaptive thresholding and quantization is combined with variable block size segmentation to further improve the coding performance and a new segmentation criterion is proposed.
Abstract: Bell Communications Research, Inc.445 South Street, Morristown, NJ 07960AbstractDiscrete cosine transform (DCT) coding has been emerging as a key element for image data compres-sion. Conventional DCT coding algorithms, which treat all the image areas indiscriminately, unfortunatelygive nonuniform image quality for various image contents. This motivates work on DCT schemes adaptive tothe image contents so that a better tradeoff between bit rate and image quality can be achieved. In this paper,a DCT algorithm with adaptive thresholding and quantization is combined with variable block size segmenta-tion to further improve the coding performance. A new segmentation criterion is proposed. Some simulationresults are given to illustrate the superiority of this adaptive DCT algorithm with segmented blocks. It is alsoshown that this algorithm poses itself as a promising compression method to deal with the composite imagesconsisting of text/graphics and natural scenes.I. IntroductionTransform coding of digital images has been studied extensively for the last decade. Thanks to its highperformance on the bit rate and image quality, it has become very popular in real applications such as stilland video image transmission and storage [5,13]. The fundamental concept of transform coding is to decorre-late the image data into a more compact form via an orthogonal transformation so that the image can berepresented by fewer bits, and therefore, compression is achieved. Subject to the assumption of spatiallywide-sense stationarity, the Karhunen-Loeve (K-L) transformation is the optimum orthogonal transformationin the mean-square error sense for decorrelating the image data. Nevertheless, the K-L transform is hardlyever used in practice due to the fact that its basis functions (eigenfunctions) are image-dependent. An alter-native is the more implementable discrete cosine transform (DCT) whose performance is close to K-Ltransform for the first-order Markov image model [12]. Although real-life image data may not fit exactly into
19 citations
••
01 Jul 1990TL;DR: A protocol for subjective and objective evaluation of the fidelity of compressed/decompressed images to the originals and the results of its application to four representative and promising compression methods are presented.
Abstract: Image compression at rates of 10:1 or greater could make PACS much more responsive and economically attractive. This
paper describes a protocol for subjective and objective evaluation of the fidelity of compressed/decompressed images to the
originals and presents the results ofits application to four representative and promising compression methods. The methods
examined are predictive pruned tree-structured vector quantization, fractal compression, the discrete cosine transform with equal
weighting of block bit allocation, and the discrete cosine transform with human visual system weighting of block bit
allocation.
Vector quantization is theoretically capable of producing the best compressed images, but has proven to be difficult to
effectively implement. It has the advantage that it can reconstruct images quickly through a simple lookup table.
Disadvantages are that codebook training is required, the method is computationally intensive, and achieving the optimum
performance would require prohibitively long vector dimensions. Fractal compression is a relatively new compression
technique, but has produced satisfactory results while being computationally simple. It is fast at both image compression and
image reconstruction. Discrete cosine iransform techniques reproduce images well, but have traditionally been hampered by
the need for intensive computing to compress and decompress images.
A protocol was developed for side-by-side observer comparison of reconstructed images with originals. Three 1024 X 1024
CR (Computed Radiography) images and two 512 X 512 X-ray CT images were viewed at six bit rates (0.2, 0.4, 0.6, 0.9,
1.2, and 1.5 bpp for CR, and 1.0, 1.3, 1.6, 1.9, 2.2, 2.5 bpp for X-ray CT) by nine radiologists at the University of
Washington Medical Center. The CR images were viewed on a Pixar II Megascan (2560 X 2048) monitor and the CT images
on a Sony (1280 X 1024) monitor.
The radiologists' subjective evaluations of image fidelity were compared to calculations of mean square error (MSE),
normalized mean square error (NMSE), percentage mean square error (PMSE), and fractal normalized mean square error
(FMSE) for each compression method and bit rate.
18 citations
••
01 Jul 1990TL;DR: In this paper, a 3D cosine transform based image compression method was proposed to take advantage of the correlations between adjacent pLtels in an image for time-sequenced studies.
Abstract: Transform based compression methods achieve their effect by taking advantage of the correlations between adjacent pLtels in an image. The increasing use of three-dimensional imaging studies in radiology requires new techniques for image compression. For time-sequenced studies such as digital subtraction angiography, pixels are correlated between images, as well as within an image. By using three-dimensional cosine transforms, correlations in time as well as space can be exploited for image compression. Sequences of up to eight 512 x 512 x 8-bit images were compressed using a single full volume three-dimensional cosine transform, followed by quantization and bit-allocation. The quantization process is a uniform thresholding type and an adaptive three-dimensional bit-allocation table is used. The resultant image fidelity vs. compression ratio was shown to be superior to that achieved by compressing each image individually.© (1990) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
17 citations
••
TL;DR: A technique has been developed for medical image sequences storage and transmission in order to obtain very high compression ratio: in dynamic nuclear medicine studies it can achieve a compression ratio as high as 100:1 without significant degradation.
Abstract: As data compression plays now an important role in the development of medical PACS, a technique has been developed for medical image sequences storage and transmission in order to obtain very high compression ratio: in dynamic nuclear medicine studies it can achieve a compression ratio as high as 100:1 without significant degradation. The implemented technique combines two methods which multiply their effects. In a first step, a principal component analysis (PCA) of the image series is performed. It extracts a limited number of principal components and their associated images. For data compression it is not necessary to perform an oblique factor analysis to estimate the so-called ‘physiological functions’ and their spatial distributions as in factor analysis of dynamic structures (FADS). In a second step, the principal images are compressed by means of a transform coding procedure: an adaptive block-quantization technique using the 2D discrete cosine transform (DCT) is implemented, followed by a statistical quantization method to encode the DCT coefficients. To reconstruct the principal images, an inverse DCT is applied. Then the original series is computed from the reconstructed images combined with the principal components which have been stored without any modification. The reconstructed series is compared to the original series, as well as the time activity curves generated on different regions of interest (ROI) and the factor estimates obtained using FADS performed on the two series. Method and evaluation are illustrated on an example of first pass radionuclide angiocardiography.
16 citations
••
TL;DR: An efficient distortion measure based on a novel subspace is proposed for 4*4 image mean residual vector quantisers (MRVQ) and the dimensionality of the coder distortion measure is reduced from 16 to 4, maintaining the visual image quality.
Abstract: An efficient distortion measure based on a novel subspace is proposed for 4*4 image mean residual vector quantisers (MRVQ). The dimensionality of the coder distortion measure is reduced from 16 to 4, maintaining the visual image quality. The computation complexity and memory requirements of the quantiser are significantly reduced. >
••
01 May 1990TL;DR: Simulation results show that CVQ-TA can give a good visual perceptual quality and can be applied for large block size (8*8 or 16*16), which is more efficient for block coding.
Abstract: A new VQ-based image coding method, classified vector quantization using texture analysis (CVQ-TA) is proposed. The most notable differences between CVQ-TA and CVQ are: (1) the classification in CVQ-TA operates in the transform domain by means of texture analysis, and (2) product-code vector quantization is used to reduce the complexity increasing with the subblock size. Because of these characteristics, CVQ-TA can be applied for large block size (8*8 or 16*16), which is more efficient for block coding. Simulation results show that CVQ-TA can give a good visual perceptual quality. >
••
TL;DR: A quantization scheme where the minimum-distortion reconstruction is always provided in the original image space is developed and presented and its design algorithm is presented.
Abstract: For high-efficiency image compression, previously, an SVD (singular value decomposition)-based coder was developed using vector quantization, called SVD-VQ. This paper proposes an improved quantization SVD-VQ scheme. For every input subblock, the SVD-VQ coder scalar-quantizes a singular value and vector-quantizes two singular vectors, separately. The SVD-VQ decoder reproduces a subblock as the product of these quantization outputs, but does not necessarily produce a reconstruction with the minimum distortion in an image space. This paper develops a quantization scheme where the minimum-distortion reconstruction is always provided in the original image space and presents its design algorithm. The improved SVD-VQ shows A/N performance improvement of 0.5 - 1.0 dB over the conventional SVD-VQ, and is similar in performance to the adaptive DCT (discrete cosine transform) coder.
•
19 Apr 1990
TL;DR: In this article, the authors proposed a method to compress a picture into an object code quantity early and to obtain best picture quality by setting variably a quantization width of quantization to an optimum value in response to the object code quantities.
Abstract: PURPOSE: To compress a picture into an object code quantity early and to obtain best picture quality by setting variably a quantization width of quantization to an optimum value in response to the object code quantity. CONSTITUTION: Upon the receipt of a picture data outputted from an orthogonal conversion circuit 4, a quantization circuit 6 multiplies a quantization width coefficient with the quantization width for each frequency component at a 1st quantization in response to the pickup mode to apply quantization of the conversion coefficient in the corrected quantization width. The quantization is implemented by using an optimum quantization coefficient decided by a preceding processing at a 2nd time. A quantization width prediction circuit 12 receives information of object code quantity from a control circuit 18 at the start of a 1st path and sets the initial value of the quantization width coefficient to output the result to the quantization circuit 6. Then optimum quantization width coefficient is predicted to obtain a close object code quantity from the code quantity of the entire picture inputted from a code quantity calculation circuit 14 and the object code quantity being a maximum data quantity allowed for one picture prior to the start of the 2nd path. Thus, the code quantity is made close to the object code quantity and best picture quality is obtained. COPYRIGHT: (C)1992,JPO&Japio
••
01 Jan 1990
TL;DR: The present work shows that, without substantially increasing the coder complexity, it is indeed possible to achieve acceptable picture quality while attaining a high compression ratio.
Abstract: This thesis deals with the development and analysis of a computationally simple vector quantization image compression system for coding monochrome images at low bit rate. Vector quantization has been known to be an effective compression scheme when a low bit rate is desirable, but the intensive computation required in a vector quantization encoder has been a handicap in using it for low rate image coding. The present work shows that, without substantially increasing the coder complexity, it is indeed possible to achieve acceptable picture quality while attaining a high compression ratio.
Several modifications to the conventional vector quantization coder are proposed in the thesis. These modifications are shown to offer better subjective quality when compared to the basic coder. Distributed blocks are used instead of spatial blocks to construct the input vectors. A class of input-dependent weighted distortion functions is used to incorporate psychovisual characteristics in the distortion measure. Computationally simple filtering techniques are applied to further improve the decoded image quality. Finally, unique designs of the vector quantization coder using electronic neural networks are described, so that the coding delay is reduced considerably.
Except for the basics of the vector quantization described in the first chapter, each chapter is independent from the others because each chapter deals with a separate aspect of the coder. Therefore, each chapter beyond the first can be read separately.
••
TL;DR: This work applies flow-based computational vision methods (previously implemented using only simple flow) to the long-sequence flow, and finds the derivation of robust visual information overcomes many of the effects of noise and quantization errors.
••
01 Aug 1990TL;DR: A new method for the noiseless compression of medical images is described, which can compress typical medical images 20% to 30% better than the conventional method.
Abstract: A new method for the noiseless compression of medical images is described The method uses the wellknown DPCM technique (ie , linear prediction) for decorrelating a given image However, instead of encoding the pixels of the decorrelated image using a memoryless model as in the conventional method, a source model with several conditioning events (or contexts) is employed The contexts are based on the horizontal and vertical components of the gradient in the given image as well as the predicted value of a pixel The statistics under each context of the model are obtained adaptively In order to encode the decorrelated image as an outcome of such a complex source model, the powerful arithmetic coding technique is employed Experimental results show that the new method can compress typical medical images 20% to 30% better than the conventional method© (1990) COPYRIGHT SPIE--The International Society for Optical Engineering Downloading of the abstract is permitted for personal use only
•
13 Mar 1990
TL;DR: In this article, a three-dimensional orthogonal transformation is applied on the L images (P1-P(L-1) at every block group at a corresponding position, respectively in the 3D OTR circuit, and a required OTR coefficient is supplied to a motion decision part 4 and a definition decision part 5.
Abstract: PURPOSE:To perform the encoding of image data with high efficiency by a moving image by performing three-dimensional orthogonal transformation by a three-dimensional block in which a time base direction is considered. CONSTITUTION:An image memory 1 stores digital image data corresponding to L continuous images on a time base supplied as input. The segmentation circuit 2 of the image segments the data as the block having size of (MXN), and supplies it in a three-dimensional orthogonal transformation circuit 3. As for segmented digital image data, the orthogonal transformation is applied on the L images (P1-P(L-1)) at every block group at a corresponding position, respectively in the three-dimensional orthogonal transformation circuit 3, and a required orthogonal transform coefficient is supplied to a motion decision part 4 and a definition decision part 5. By performing bit allocation for adaptive quantization by utilizing the information of the motion of the image and that of the definition of the image, the encoding with high efficiency can be attained.
•
21 May 1990
TL;DR: In this paper, the authors proposed to prevent local deterioration in the quality of reproduced picture by obtaining an AC component coefficient through the application of 2-dimensional orthogonal conversion such a discrete cosine conversion DCT to a picture data, applying quantization, splitting the result into a bit plane and applying run length encoding compression while placing priority to the coefficient with larger absolute valve.
Abstract: PURPOSE:To prevent local deterioration in the quality of a reproduced picture by obtaining an AC component coefficient through the application of 2-dimension orthogonal conversion such a discrete cosine conversion DCT to a picture data, applying quantization, splitting the result into a bit plane and applying run length encoding compression while placing priority to the coefficient with larger absolute valve. CONSTITUTION:An orthogonal conversion section 1 splits a multi-gradation picture data into a block of M picture element X M picture element and used a DCT to apply NXN 2-dimension orthogonal conversion. The AC component from the orthogonal conversion section 1 is given to a compression section 2 to convert the AC component into a variable length code and to execute the run length encoding compression. The data compressed by run length encoding is given to an expansion section 3, the compression data is decoded to generate the original AC component. Then the decoded AC component is given to the orthogonal conversion section 4 together with the DC component from the said orthogonal conversion section 1 to apply inverse conversion to the DC and AC components to reproduce the original multi-gradation picture data.
••
01 Sep 1990TL;DR: In this paper, the authors describe source models of bit rate for digital TV codecs based on the Discrete Cosine Transformation as decorrelation technique, which are based on cyclostationary processes.
Abstract: This paper describes source models of bit rate for digital TV codecs based on the Discrete Cosine Trans-formation as decorrelation technique. These models are based on cyclostationary processes. Properties ofcyclostationary processes are such that the process is preserved over the whole TV transmission. Therefore,the influence of image Statistics is presented on the coders, the transmission networks, and, the decoders.An emphasis is given onvariable bit rate codecs transmitting on ATM networks. 1 INTRODUCTION Digital applications of this paper deal with CCIR 601 4:2:2 standard TV video format: in interlaced TVformat, images are divided in two fields, each field consists in 288 lines of either 720 pixels for luminanceor 360 pixels for each chrominance; the frequency is of 25 frames per second.The standard TV codecs, under investigation, are based on Discrete Cosine Transformation (DCT) andaim to reduce the incoming bit rate at 216 Mbits/s with a factor of 5 to 100 to eliminate the redundant
••
01 Sep 1990TL;DR: This paper presents a method for the classification and coding of textures based upon the use of trans-form vector quantization, and suggests avenuws for future research which will yield significant improvements in future work.
Abstract: This paper presents a method for the classification and coding of textures based upon the use of trans-form vector quantization. Techniques for texture classification and vector quantization similarly process small, nonoverlapping blocks of image data which are extracted independently from the image. Localspatial frequency features have been identified as being appropriate for texture classification, indicatingthat a transform vector quantization scheme should be capable of characterizing and classifying texturedregions. A data set consisting of 7 natural textures is used to demonstrate the utility of this approach.The experimental results show acceptable classification rates and suggest avenuws for future researchwhich will yield significant improvements in future work. 1 Introduction Various approaches have been adopted for the solution of the texture classification problem, includingthe use of structural pattern recognition, second order statistical analysis and local spatial frequency
••
20 May 1990TL;DR: In this paper, an 8-b quantization scheme was proposed to reduce the data volume for single-look complex scattering matrix data measured by polarimetric imaging radar systems, where the scattering matrices are not symmetrized before compression, thereby retaining information about background noise and system effects.
Abstract: An 8-b quantization scheme to reduce the data volume for single-look complex scattering matrix data measured by polarimetric imaging radar systems is described. The scattering matrices are not symmetrized before compression, thereby retaining information about background noise and system effects. The data volume is reduced by a factor of 3.2. It is shown, with measured data, that the signal to quantization noise ratio for the compression scheme is more than 35 dB for the cross-polarized channels, and more than 45 dB for the copolarized channels.
••
22 Jan 1990TL;DR: A new technique for compression of binary images is proposed that uses the Radon transform to convert a binary image to a set of 1-dimensional (1-D) non-binary sequences, which are coded using 1-D techniques.
Abstract: A new technique for compression of binary images is proposed. The Radon transform is used to convert a binary image to a set of 1-dimensional (1-D) non-binary sequences, which are coded using 1-D techniques. A binary image can be reconstructed from a very small number of projections and this leads to significant compression. The compression ratio for a binary image of size NxN is inversely proportional to N.
•
28 Nov 1990TL;DR: In this paper, a picture quality decision is made to obtain a satisfactory reproducing image by performing encode noise elimination in accordance with picture quality by controlling the characteristics of a coring means and a high-frequency emphasis means by the picture quality of a decoding image.
Abstract: PURPOSE:To obtain a satisfactory reproducing image by performing encode noise elimination in accordance with picture quality by controlling the characteristics of a coring means and a high-frequency emphasis means by the picture quality of a decoding image. CONSTITUTION:A picture quality decision means 1 calculates a mean value extending over the whole image with quantization width used in the encoding of an image block, and compares it with a prescribed threshold value, and outputs a comparison result. The coring means 2 inputs a decoded digital image signal, and substitutes a value 0 for a low amplitude component whose absolute value is less than a certrain threshold value Tc out of high-frequency components of spatial frequency. At this time, the value of the threshold value Tc is controlled by the output of the decision means 1. The high-frequency emphasis means 3 performs correction in accordance with the Laplacean value of a remarked picture element and the output of the decision means 1 to compensate resolution lowered by the means 2.
••
TL;DR: A novel bit-rate reduction method developed for the MUSE (multiple subNyquist sampling encoding) signal is discussed, using an orthogonal transformation by which the Muse data-rate is reduced from 1/2 to 1/3 of the original, maintaining a picture quality acceptable for the home-use digital VTR of a HDTV (high-definition television) system.
Abstract: A novel bit-rate reduction method developed for the MUSE (multiple subNyquist sampling encoding) signal is discussed. This method uses an orthogonal transformation by which the MUSE data-rate is reduced from 1/2 to 1/3 of the original, maintaining a picture quality acceptable for the home-use digital VTR of a HDTV (high-definition television) system. By simultaneous application of two-dimensional DCT (discrete cosine transform) and nonlinear quantization methods for a quincuncial pixel arrangement formed within a field, an optimum data assignment between DCT coefficients and nonlinear quantization values is accomplished by using a variable length coding method. >
•
07 Nov 1990
TL;DR: In this article, the authors propose to improve a picture quality by executing the setting of a threshold so as to conserve the frequency components of a low band as possible in consideration for the visual characteristic of a human.
Abstract: PURPOSE:To improve a picture quality by executing the setting of a threshold so as to conserve the frequency components of a low band as possible in consideration for the visual characteristic of a human. CONSTITUTION:The encoder is equipped with a cosine transformation part 1, a threshold processing part 2, a quantization processing part 3, a variable length encoding part 4, and a buffer part 5 for adjusting the speed. The threshold is set higher as the group of the coefficient of a high-order with a comparatively small amplitude (power), and as to the coefficient at the threshold or below, the value is managed as zero. Thus, the generation rate of a zero run is increased, the encoding efficiency is improved, and even in the case of a low bit rate encoding, the image to be visually suitable can be obtained.
••
•
26 Jul 1990
TL;DR: In this article, the authors proposed a method to identify a picture characteristic according to the variance of a density level excepting for the picture element of the lowest level and a picture element number based on a result counting the number of the picture elements for each density level of a multilevel picture.
Abstract: PURPOSE:To execute optimum binarization processing concerning respective original pictures by identifying a picture characteristic according to the variance of a density level excepting for the picture element of the lowest level and a picture element number based on a result counting the number of the picture elements for each density level of a multilevel picture and executing the binarization processing which is matched to the picture characteristic. CONSTITUTION:The picture, to which multilevel quantization is executed, is read from a scanner 3 by a multilevel picture read part 1 and for this read multilevel picture, the number of the picture elements for each density level is counted in a picture element histogram count part 5. Next, in a variance calculation part 6, the variance of the density level excepting for the picture element of the lowest level and the picture element number is obtained. In a picture characteristic identification part 7, the picture characteristic is identified by the obtained variance and in a binarization part 8, the multilevel picture is binarized by the binarization system of a method suitable for the identified picture characteristic. Thus, even in the case of an original to have the various types of printing conditions, the optimum binarization processing can be executed concerning the respective original pictures.
••
01 Sep 1990TL;DR: The above method gives a more accurate estimation of the displacement field and it is shown to be more robust in the presence of occlusion and noise, compared to the mean-squared error based block-matching algorithm.
Abstract: In this paper, a distributed detection approach for displacement estimation in image sequences is presented. This method is derived from a Bayesian framework and reduces to a M-ary Hypothesis test among a representative set of possible displacement vectors. It is shown that the mean-squared error based block-matching (BM) algorithm is a special case of this general approach. In our approach, at each point of the current frame a set of overlapping localized detectors outputs a number of estimates for the displacement vector. Then, a distributed detection network is adopted for the fusion of the these estimates. Since the computational load is high, suboptimal but computationally efficient solutions are proposed. The above method gives a more accurate estimation of the displacement field and it is shown to be more robust in the presence of occlusion and noise, compared to the BM algorithm. Experimental results on video-conference image sequences are presented.
••
30 Jan 1990TL;DR: The concept of a combined quality and bit rate controller is introduced that is capable of satisfying both the viewer by a constant perceived image quality and the network by suitable coder output bit rate statistics.
Abstract: In conventional constant transmission rate image coders the image quality is varied in dependence on the current activity in the image sequence in order to obtain a constant data rate. Flexible transmission rate will be available in the future along with packetized networks and a constant image quality could be achieved. Using a 2 Mbit/s constant transmission rate codec for video conference we have studied three modifications of this coder to adapt it to variable transmission rate image coding: open loop, closed loop and constant MSE distortion. It is shown that none of these modes yields satisfying results neither from the image quality nor from the channel rate requirements point of view. After a thorough discussion of the pros and cons of above modifications the concept of a combined quality and bit rate controller is introduced. It is capable of satisfying both the viewer by a constant perceived image quality and the network by suitable coder output bit rate statistics. For a constant image quality a quality measure was developed reflecting the perceived image quality very well. Using this quality measure we are able to adapt quantization to the visibility of errors and thus prevent waste of bits.