Showing papers on "Image quality published in 1995"

Image quality measures and their performance

Abstract: A number of quality measures are evaluated for gray scale image compression. They are all bivariate, exploiting the differences between corresponding pixels in the original and degraded images. It is shown that although some numerical measures correlate well with the observers' response for a given compression technique, they are not reliable for an evaluation across different techniques. A graphical measure called Hosaka plots, however, can be used to appropriately specify not only the amount, but also the type of degradation in reconstructed images.

Spotlight synthetic aperture radar : signal processing algorithms

Abstract: Part 1 Introduction: spotlight SAR SAR modes importance of spotlight SAR early SAR chronology. Part 2 Synthetic aperture radar fundamentals: SAR system overview imaging considerations pulse compression and range resolution synthetic aperture technique for Azimuth resolution SAR coherence requirements signal phase equation inverse SAR (ISAR) SAR sensor parametric design. Part 3 Spotlight SAR and polar format algorithm: scope of processing task polar format overview polar data storage as a two-dimensional signal correction for non-planar motion polar format algorithm limitations Taylor series expansion procedures phase of image pixels image geometric distortion image focus error equations displacements and absolute positioning. Part 4 Digital polar format processing: sampling rate conversion polyphase filters polar interpolation image scale factors image distortion correction signal history projections stabilized scene polar interpolation subpatch processing and mosaicking. Part 5 Phase errors: classification of phase error management of phase error magnitude of phase error requirements on a practical SAR motion sensor moving target effects. Part 6 Autofocus techniques: mapdrift multiple aperture mapdrift phase difference phase gradient prominent point processing considerations for space-variant refocus. Part 7 Processor design examples: the common UNIX SAR processor the ground to air imaging radar processor. Part 8 SAR system performance: image quality metrics system performance budgeting requirements on system impulse response requirements on system noise geometric distortion secondary image quality metrics test arrays. Part 9 Spotlight processing applications: spotlight processing of scan and stripmap SAR data interferometric SAR forward look SAR vibrating target detection. Part 10 Range migration algorithm: model algorithm overview analytical development discussion efficient algorithms for range migration processing. Part 11 Chirp scaling algorithm: non-dechirped signal model algorithm overview analytical development discussion. Part 12 Comparison of image formation algorithms: image formation algorithm models computational complexity memory requirements other considerations.

Synthetic aperture imaging for small scale systems

Abstract: Multi-element synthetic aperture imaging methods suitable for applications with severe cost and size limitations are explored. Array apertures are synthesized using an active multi-element receive subaperture and a multi-element transmit subaperture defocused to emulate a single-element spatial response with high acoustic power. Echo signals are recorded independently by individual elements of the receive subaperture. Each method uses different spatial frequencies and acquisition strategies for imaging, and therefore different sets of active transmit/receive element combinations. Following acquisition, image points are reconstructed using the complete data set with full dynamic focus on both transmit and receive. Various factors affecting image quality have been evaluated and compared to conventional imagers through measurements with a 3.5 MHz, 128-element transducer array on different gel phantoms. Multielement synthetic aperture methods achieve higher electronic signal to noise ratio and better contrast resolution than conventional synthetic aperture techniques, approaching conventional phased array performance. >

Convolution-based interpolation for fast, high-quality rotation of images

Abstract: This paper focuses on the design of fast algorithms for rotating images and preserving high quality. The basis for the approach is a decomposition of a rotation into a sequence of one-dimensional translations. As the accuracy of these operations is critical, we introduce a general theoretical framework that addresses their design and performance. We also investigate the issue of optimality and present an improved least-square formulation of the problem. This approach leads to a separable three-pass implementation of a rotation using one-dimensional convolutions only. We provide explicit filter formulas for several continuous signal models including spline and bandlimited representations. Finally, we present rotation experiments and compare the currently standard techniques with the various versions of our algorithm. Our results indicate that the present algorithm in its higher-order versions outperforms all standard high-accuracy methods of which we are aware, both in terms of speed and quality. Its computational complexity increases linearly with the order of accuracy. The best-quality results are obtained with the sine-based algorithm, which can be implemented using simple one-dimensional FFTs. >

Applied Image Processing

Abstract: System design scene constraints image acquisition image preprocessing image understanding image analysis pattern classification applications and case studies visual inspection robotic vision and control.

Odd aberrations and double-pass measurements of retinal image quality

Abstract: We investigated the formation of the aerial image in the double-pass method to measure the optical quality of the human eye. We show theoretically and empirically that the double pass through the eye’s optics forces the light distribution in the aerial image to be an even-symmetric function even if the single-pass point-spread function is asymmetric as a result of odd aberrations in the eye. The reason for this is that the doublepass imaging process is described by the autocorrelation rather than the autoconvolution of the single-pass point-spread functions, as has been previously assumed. This implies that although the modulation transfer function can be computed from the double-pass aerial image, the phase transfer function cannot. We also show that the lateral chromatic aberration of the eye cannot be measured with the double-pass procedure because it is canceled by the second pass through the eye’s optics.

Temporal and spatial error concealment techniques for hierarchical MPEG-2 video codec

Abstract: The MPEG-2 source coding algorithm is very sensitive to the channel disturbances. For instance a single bit error in the bitstream will cause a high degradation of picture quality due to error propagation. Hence, for picture replenishment error concealment techniques (ECTs) may be required at the receiver. The aim of the article is to study different ECTs for MPEG-2 hierarchical coded pictures applied for terrestrial broadcasting. For the base layer different temporal and spatial ECTs are investigated. Two temporal ECTs are considered: a simple temporal error concealment (EC) and a temporal EC with motion compensation. The latter method provides the best results in inter coded pictures (where motion vectors exist). For the intra coded pictures, where no motion information exists, two spatial interpolation techniques are considered. The main problem for spatial EC in MPEG-2 coded pictures is that only the top and the bottom macroblock can be used for interpolation, since one error in the bitstream causes one damaged horizontal stripe of macroblocks in the picture. Different ECTs for the upper layer of hierarchical coded pictures are also investigated. The possibility of upsampling the base layer for concealing the upper layer made by spatial scalability gives best results.

Vision based hand modeling and tracking for virtual teleconferencing and telecollaboration

Abstract: The authors present a hand model that simultaneously satisfies both the synthesis and analysis requirements of model based compression. The model can be fitted to any person's hand and can be done using a single camera. Once the model is fitted to a real human hand, it is then used in several tracking scenarios in order to verify its effectiveness. With successful tracking achieved, the model is ready to be incorporated into a virtual environment or model based compression scheme such as sign language communication over telephone lines or virtual teleconferences over computer networks at very low bit rates and at very high image quality. >

Continuous radial data acquisition for dynamic MRI

Abstract: Since image acquisition times in MRI have been reduced considerably over recent years, several new important application areas of MRI have appeared. In addition to pure static anatomic information, the evolution of a dynamic process may be visualized by a sequence of temporal snapshots of the process acquired within a short time period. This makes applications like interactive or interventional MRI as well as the acquisition of additional functional information feasible. For high temporal resolution, all these applications require a quasi real-time image acquisition during the time the interaction or dynamic process evolves. We present an approach to real-time imaging using a continuous radial acquisition scheme. The intrinsic advantages of radial or projection reconstruction (PR) techniques are used to minimize motion-related image distortions. Modifications of the acquisition scheme as well as dedicated reconstruction techniques are used to further reduce the temporal blurring due to the finite acquisition time of one entire data set in our approach. So far we have used this technique for the visualization of active joint motion.

Physically-valid view synthesis by image interpolation

Abstract: Image warping is a popular tool for smoothly transforming one image to another. "Morphing" techniques based on geometric image interpolation create compelling visual effects, but the validity of such transformations has not been established. In particular, does 2D interpolation of two views of the same scene produce a sequence of physically valid in-between views of that scene? We describe a simple image rectification procedure which guarantees that interpolation does in fact produce valid views, under generic assumptions about visibility and the projection process. Towards this end, it is first shown that two basis views are sufficient to predict the appearance of the scene within a specific range of new viewpoints. Second, it is demonstrated that interpolation of the rectified basis images produces exactly this range of views. Finally, it is shown that generating this range of views is a theoretically well-posed problem, requiring neither knowledge of camera positions nor 3D scene reconstruction. A scanline algorithm for view interpolation is presented that requires only four user-provided feature correspondences to produce valid orthographic views. The quality of the resulting images is demonstrated with interpolations of real imagery.

Automatic white balance for digital still camera

Abstract: A new algorithm of automatic white balance for electronic images has been developed. This algorithm uses fuzzy logic rules for determination of the color parameters in order to minimize the color temperature difference of various light sources. This algorithm has been implemented in a newly designed electronic still camera, and resulted in highly improved image quality. >

A comparison of several algorithms for SAR raw data compression

Abstract: Proposes new algorithms for synthetic aperture radar (SAR) raw data compression and compares the resulting image quality with the quality achieved by commonly used methods. The compression is carried out in time and frequency domain, with statistic, crisp, and fuzzy methods. The algorithms in the time domain lead to high resolution and a good signal-to-noise ratio, but they do not optimize the performance of the compression according to the frequency envelope of the signal power in both range and azimuth directions. The hardware requirements for the compression methods in the frequency domain are significant, but a higher performance is obtained. Even with a data rate of 3 bits/sample, a satisfactory phase accuracy is achieved which is an essential parameter for polarimetric and interferometric applications. Preliminary analysis concerning the suitability of the proposed algorithms for different SAR applications shows that the compression ratio should be adaptively selected according to the specific application. >

Electronic image pickup device

Abstract: PROBLEM TO BE SOLVED: To obtain high image quality applicable even to a print photograph by using a flair elimination circuit to eliminate a harmful signal component and including a specific negative lens and a diffraction optical element for an image forming lens system. SOLUTION: An image of a pickup object is formed on a light receiving face of a solid-state image pickup element 3 by an image forming optical system 1. An electric signal generated therein is image-processed by a signal processing circuit 6 and converted into a video signal such as an RGB component signal, and a flair elimination circuit 7b eliminates a harmful signal component. Furthermore, a sensor drive circuit 5 drives the solid-state image pickup element 3 to control image storage and read control. The image optical system 1 is provided with an image forming lens system 10, a filter group 11 and a prism 12. The image forming lens system 10 includes negative lenses L3, 4 satisfying a condition of 0.1<-fN/fL<2, where fL is a focal distance of the entire image forming lens and a diffraction optical element L10 satisfying a relation of fL/ fDOE<0.07, where fDOE is a focal distance of a diffraction face. Thus, high image quality applicable even to a print photograph and a poster is obtained. COPYRIGHT: (C)1997,JPO

Quality asessment of coded images using numerical category scaling

Abstract: The large variety of algorithms for data compression has created a growing need for methods to judge (new) compression algorithms. The results of several subjective experiments illustrate that numerical category scaling techniques provide an efficient and valid way not only to obtain compression ratio versus quality curves that characterize coder performance over a broad range of compression ratios, but also to assess perceived image quality in a much smaller range (e.g. close to threshold level). Our first object is to discuss a number of simple techniques that can be used to assess perceived image quality. We show how to analyze data obtained from numerical category scaling experiments and how to set up such experiments. Second, we demonstrate that the results from a numerical scaling experiment depend on the specific nature of the subject's task in combination with the nature of the images to be judged. As results from subjective scaling experiments depend on many factors, we conclude that one should be very careful in selecting an appropriate assessment technique.

Automated shimming at 1.5 T using echo-planar image frequency maps.

Abstract: Using echo-planar imaging, we developed an automated image-based procedure to shim the static (B0) field. Our method uses the rapid acquisition capability of echo-planar imaging to collect the required frequency data rapidly, rendering the shim data acquisition time negligible in comparison with the total study time. We address image distortion issues involved in echo-planar imaging acquisition of the data and formulate analytic methods for arriving at an optimal shim for the NMR imaging experiment in a single iteration. We investigated the use of cost functions other than least-squares (Chebychev, high-order numeric) and found that choice between the cost functions we tested was irrelevant to resultant image quality, at least when used in conjunction with low-order shims. With appropriate integration, the method has become routine practice for investigators at our laboratory.

Accelerating fractal image compression by multi-dimensional nearest neighbor search

Abstract: In fractal image compression the encoding step is computationally expensive. A large number of sequential searches through a list of domains (portions of the image) are carried out while trying to find the best match for another image portion. Our theory developed here shows that this basic procedure of fractal image compression is equivalent to multi-dimensional nearest neighbor search. This result is useful for accelerating the encoding procedure in fractal image compression. The traditional sequential search takes linear time whereas the nearest neighbor search can be organized to require only logarithmic time. The fast search has been integrated into an existing state-of-the-art classification method thereby accelerating the searches carried out in the individual domain classes. In this case we record acceleration factors from 1.3 up to 11.5 depending on image and domain pool size with negligible or minor degradation in both image quality and compression ratio. Furthermore, as compared to plain classification our method is demonstrated to be able to search through larger portions of the domain pool without increased the computation time.

An error concealment algorithm for images subject to channel errors

Abstract: We present an algorithm to conceal bit errors in still images and image sequences that are coded using the discrete cosine transform (DCT) and variable length codes (VLCs). No modification is necessary to an existing encoder, and no additional bit rate is required. The concealment algorithm is kept simple so that real-time decoding and concealment is possible. A single bit error in these images can cause a block to split into several blocks or several blocks to merge into one. This causes the DCT coefficients of all subsequent blocks to be correctly decoded but stored in the wrong location in the image. Furthermore, the DC coefficient of all subsequent blocks may be incorrect. The error concealment algorithm uses transform domain information to identify the location of the affected blocks and to correct errors. The image quality after error concealment is shown to be significantly improved. >

Apparatus and method for compressing still images

Abstract: Transforms such as the DCT are useful for image compression. One close relative of the DCT is preferred for its arithmetic simplicity. A method and apparatus is described whereby the image compression is done with no multiplications while still compatible with a JPEG (Joint Photographic Experts Group) Transform. Other enhancements are made to improve image quality.

Combined techniques of singular value decomposition and vector quantization for image coding

Abstract: The combination of singular value decomposition (SVD) and vector quantization (VQ) is proposed as a compression technique to achieve low bit rate and high quality image coding. Given a codebook consisting of singular vectors, two algorithms, which find the best-fit candidates without involving the complicated SVD computation, are described. Simulation results show that the proposed methods are better than the discrete cosine transform (DCT) in terms of energy compaction, data rate, image quality, and decoding complexity. >

Advanced ccd based x-ray image sensor system

Abstract: An improved CCD-based x-ray image sensor system enables the use of an uncooled or only slightly cooled CCD array (18a) within a standard size x-ray film cassette (1). The sensor system provides a number of advanced functions such as remote diagnostic capability, variable image resolution, real-time exposure control, automatic x-ray detection, a low-power "sleep" mode, and automatic, closed loop optimization of image quality.

A fractal approach to the classification of Mediterranean vegetation types in remotely sensed images

Abstract: A method is presented to assess fractal dimensions from remotely sensed images. The method is a three-dimensional version of the walking dividers method which has been applied to two digital images of southern France to distinguish various types of Mediterranean landscape units. The first image is a Landsat Thematic Mapper image, while the second image was acquired by the airborne Geophysical Environmental Research Imaging Spectrometer. The method has been tested on some artificial images to demonstrate procedures and results. The method can distinguish range-lands, maquis and closed garrigue and to a lesser extent agricultural regions on the TM image. Fractal dimensions for open garrigue and badlands are similar. However, the reflection properties of the land-cover units do not behave like real fractals at the scale considered, and different methods to compute the fractal dimension do not yield the same results. Results of the airborne image are disappointing, probably due to somewhat poor image quality. Finally, some advantages and disadvantages of the method are discussed.

Region-based fractal image compression using heuristic search

Abstract: Presents work carried out on fractal (or attractor) image compression. The approach relies on the assumption that image redundancy can be efficiently exploited through self-transformability. The algorithms described utilize a novel region-based partition of the image that greatly increases the compression ratios achieved over traditional block-based partitionings. Due to the large search spaces involved, heuristic algorithms are used to construct these region-based transformations. Results for three different heuristic algorithms are given. The results show that the region-based system achieves almost double the compression ratio of the simple block-based system at a similar decompressed image quality. For the Lena image, compression ratios of 41:1 can be achieved at a PSNR of 26.56 dB. >

Image conversion apparatus for transforming compressed image data of different resolutions wherein side information is scaled

Abstract: An image conversion apparatus for converting and outputting the spatial resolution, temporal resolution, or image quality of compressed image data input thereto is provided. A compression-coded frame-unit image signal An input from the input signal line is changed to real-time image data Rn by an image decoding unit. A image coding unit then converts the real-time image data Rn to compressed image data comprising fewer picture elements and having a different spatial or temporal resolution different from that of the real-time image data Rn.

Adaptive spatial-temporal postprocessing for low bit-rate coded image sequences

Abstract: A 3D nonlinear postprocessing system and method are utilized to reduce coding artifacts produced by block-based motion-compensated transform coding. In the system and method, a separable 3D filtering structure is used: space-variant FIR-Median Hybrid filtering (in 106) is used in the spatial domain, followed by a motion-compensated (on 112) nonlinear filtering in the temporal domain (in 110). By using this structure and method, the coding artifacts in a reconstructed image sequence can be effectively reduced without blurring edges or moving objects in the image sequence. Significant improvement in the picture quality of low bit-rate coded video sequences (111) is thereby achieved.

Segmentation based image compression system

Abstract: An image processing system encodes a natural image into a segmented or mosaic image having well-defined edges and a residual image. The segmented image is encoded using a lossless encoding technique while the residual image is encoded using a lossy technique. This encoded image may be recorded on a video tape such that the segmented image may be recovered in picture-in-shuttle modes such as fast forward and fast rewind. In addition, the recorded image may be decoded and reencoded through several generational levels without experiencing significant degradation in perceived image quality. The segmented image is produced by an encoder which employs a multi-scale edge finder that is able to resolve transitions occurring over two, four and eight pixels into an edge located between two pixels. In addition, the lossless encoder includes circuitry which chain-encodes the segmented image into a low-resolution image and multiple complementary images such that the low-resolution image may be stored as a single data packet and, thus, recovered as a unit from the tape in picture-in-shuttle mode.

Scanned remote imaging method and system and method of determining optimum design characteristics of a filter for use therein

Abstract: An imaging method and system for optimum sampling, image restoration and enhancement is disclosed where imaging errors associated with scanning a fiber optic image guide are identified and filtered to restore image quality to the available resolution limit. Scanned image samples are synchronized with spot positions corresponding to the center of each fiber to avoid generating spurious data resulting from unwanted modulation of the scanning beam by the fiber optic array, manifested as large amplitude multiplicative noise. A detector, such as a position sensitive device, is utilized which allows for monitoring both spot intensity and position. Peak detection on the intensity waveform determines the instant at which the beam is to be sampled. The method applies to any imaging system where a beam of electromagnetic energy such as radiant energy is scanned through a plurality of optical fibers to irradiate the object under examination. Various embodiments are disclosed which, depending upon application requirements, affect the level and sophistication of the signal processing hardware. The imaging system may utilize grey scale, color, motion (as in a Doppler vibrometry system) or 3D information.

Aliasing reduction in staring infrared imagers utilizing subpixel techniques

Abstract: We introduce and analyze techniques for the reduction of aliased signal energy in a staring infrared imaging system. A standard staring system uses a fixed two-dimensional detector array that corresponds to a fixed spatial sampling frequency determined by the detector pitch or spacing. Aliasing will occur when sampling a scene containing spatial frequencies exceeding half the sampling frequency. This aliasing can significantly degrade the image quality. The aliasing reduction schemes presented here, referred to as microscanning, exploit subpixel shifts between time frames of an image sequence. These multiple images are used to reconstruct a single frame with reduced aliasing. If the shifts are controlled, using a mirror or beam steerer for example, one can obtain a uniformly sampled microscanned image. The reconstruction in this case can be accomplished by a straightforward interlacing of the time frames. If the shifts are uncontrolled, the effective sampling may be nonuniform and reconstruction becomes more complex. A sampling model is developed and the aliased signal energy is analyzed for the microscanning techniques. Finally, a number of experimental results are presented that illustrate the perlormance of the microscanning methods.

System and method for automatically processing image data to provide images of optimal perceptual quality

Abstract: An image processing system that automatically optimizes the perceptual quality of images undergoing a series of selected image-processing operations. The automatic optimization capability allows the system to bypass expensive and time-consuming trial-and-error methods associated with interactive methods. The system consists of a set of image-processing operations, an architecture, and an intelligent control. These elements take into consideration profiles of sources from which the images are generated, profiles of intended applications, and the impact that image processing operations (individually or in concert) will have on perceived image quality. The analysis uses a body of relationships linking human perception of image quality with objective metrics (such as sharpness, grain, tone, and color) of image content. The relationships used are based upon extensive psychovisual testing, using human observers and photographic images. The intelligent control embodies the test results and thereby functions as a synthetic human observer. Controlling a collection of adjustable parameters in the requested image-processing operations, it works automatically to maximize subjective quality for the resulting picture. When optimal parameter values have been determined, the system then subjects the image data, itself, to the corresponding processing operations and provides an image to the end application. This is all achieved without operator intervention other than initially selecting the set of processing operations. The functionality of the system may be implemented in a variety of physical architectures, including computer workstations, and including communicating sets of discrete components in which the computations and control are distributed among the components.

Image processor for correcting distortion of central portion of image and preventing marginal portion of the image from protruding

Abstract: A distorted optical image formed on an image pickup surface by an optical system is photoelectrically converted by an image pickup device and written in memory cells of an image memory designated by write addresses. When the optical image is read, the image quality at the central portion is improved and the image is prevented from missing at the marginal portion so that the image can preferably be observed by correcting the distortion of an image in the area at the central side of the image pickup surface due to the optical system by read addresses obtained by converting the write addresses and preventing an image in the area at the marginal side from protrusion.