摘要:

Manipulation of computed tomography image data at the scanner console to produce slices of the body at arbitrary positions is currently either fairly slow or else very expensive to achieve in real-time. The paper presents a modelling technique based on octrees which builds a three-dimensional model from the scanned slices and then allows the production of arbitrarily sliced images and projections, volume calculations, etc. The Millar polyhedron is introduced as a new technique used to build the octree. It exploits the volume coherence in the image data in such a way that the octree nodes can be directly specified from its output. It is clearly demonstrated that the transputer approach can allow the production of a real-time system at lower cost. Experimental results indicate that the algorithms are inherently concurrent and are suited to implementation in a parallel processing environment such as that provided by the transputer. They also indicate the degree of data compression obtained by using the technique, compared with direct storage of the raw data.

DOI:10.1049/ip-i-2.1990.0017

出版商:IEE    

年代:1990

数据来源: IET

摘要:

A system is presented which is capable of performing perspective spatial transformations at video rate, by means of a mapping function which can vary at frame rate. The mapping is implemented by two successive applications of a one-dimensional algorithm operating in scan-line order, using a framestore to store an intermediate image. A novel one-dimensional filtering algorithm provides a continuous mapping from source to destination co-ordinates, significantly reducing aliasing. Emphasis is placed on the design of a novel stream processing architecture which, coupled with an efficient pipelined co-ordinate generation system, provides single-cycle per output pixel performance.

DOI:10.1049/ip-i-2.1990.0018

出版商:IEE    

年代:1990

数据来源: IET

摘要:

A method of simultaneous image segmentation and edge detection based on grey-level co-occurrence matrices is described. An analysis of the distributions within a co-occurrence matrix defines an initial pixel classification into both region and interior or boundary classes. Local consistency of pixel classification is enforced by minimising the entropy of local region and boundary information, where region information is expressed by conditional probabilities, estimated from the co-occurrence matrices, and boundary information by conditional probabilities which are determined apriori. The method robustly segments an image into homogeneous areas and generates an edge map. The technique extends easily to general edge operators; examples are given of the techniques applied to both synthetic and infrared imagery for the [1 –1] and Canny edge operators. The results are compared with other techniques.

DOI:10.1049/ip-i-2.1990.0019

出版商:IEE    

年代:1990

数据来源: IET

摘要:

Used to model pattern development in biological and chemical systems, reaction-diffusion (RD) equations possess self-organising properties that are useful in processing many classes of images. Here we describe their use in processing two classes: images with quasi-zero-dimensional patterns (textures), and images with quasi-one-dimensional patterns (fingerprints). The paper presents details of linear and nonlinear analysis of RD equations, outlining their pattern-forming principles. Details of how to engineer RD systems are presented, together with examples and parameter sets.

DOI:10.1049/ip-i-2.1990.0020

出版商:IEE    

年代:1990

数据来源: IET

摘要:

A method of automatically ‘stretching’ a series of hand thermograms to a standard form is described. It is based on an elastic model which has the iterative and co-operative properties of relaxation labelling, while not requiring large storage space for label updating. The model was chosen to be iterative because it is difficult to produce a low-error association in a single pass. At each iteration some of the mutual differences between image pairs are removed, making the association between the new pairs more reliable. In this way the system gradually makes its way towards a match. The experiments described in this paper show that mismatches decrease with each iteration, and thus the reliability of both local and global matches is increased. The method is co-operative because a feature at one location influences decisions made in its surrounding area. Nearest neighbour relationships are used to derive local difference vectors between features. These vectors are combined in a weighted averaging scheme to produce corrected difference vectors which are then applied to an image to create a new image for further matching. An FFT based technique has been developed to enhance the efficiency of the corrected difference vector evaluation. The method has been found to be efficient and cost effective for 2-dimensional image matching.

DOI:10.1049/ip-i-2.1990.0021

出版商:IEE    

年代:1990

数据来源: IET

摘要:

A method is described for improving resolution in long-wavelength holography by expanding the two-dimensional hologram aperture. Simulation results are presented to illustrate resolution improvements obtained when doubling the linear size of square apertures for imaging single-point, multiple-point and quasi-continuous objects in the presence of noise. A technique for improving the quality of images reconstructed from predicted holograms using a number of focus measures is described. Alternative strategies for reconstructing the two-dimensional holograms are discussed and their performance compared regarding prediction errors and noise. Data is supplied on CPU time usage for hologram expansion and reconstruction on a VAX-11/785 computer.

DOI:10.1049/ip-i-2.1990.0022

出版商:IEE    

年代:1990

数据来源: IET

摘要:

The problem of image restoration from noisy measurements as encountered in nuclear medicine is considered. A model for the emission/detection process is introduced, which is based on the Poisson statistics of the emissions and a point-spread function for the imaging system. A new approach for treating the measurements is given, in which they are represented by a spatial non-causal interaction model before maximum entropy restoration, which describes the statistical dependence among the image values and their neighbourhood. The particular application of the algorithms presented here relates to gamma-ray imaging systems, and is aimed at improving the resolution–noise-suppression product. Results for actual gamma camera data are presented and compared with more conventional techniques.

DOI:10.1049/ip-i-2.1990.0023

出版商:IEE    

年代:1990

数据来源: IET

摘要:

A comparison of retinal fluorescein angiograms taken before cardiopulmonary bypass with those taken just before the end of operation, has shown that microvascular occlusions can develop during operation. Quantification of the retinal damage in terms of the area of non-perfusion depends on careful examination by humans. Assessment of these areas is necessarily subjective and digital image analysis would be preferable to make the measurements more objective. Visual or automatic comparison of images is facilitated by prior geometrical alignment. A method for registering retinal images which has been used successfully on 65 image pairs is described. Techniques for automatically detecting changes in the microvascular structure are illustrated. Using a subset of 40 good quality image pairs, good agreement between computer detected occlusions and human assessment of non-perfusing areas is reported.

DOI:10.1049/ip-i-2.1990.0024

出版商:IEE    

年代:1990

数据来源: IET

摘要:

The paper describes a high-level programming language for image processing, called IAL (Image Algebra Language). IAL has been designed for low-level image processing applications. It has been implemented on transputer networks. IAL provides mechanisms whereby a programmer can specify image processing operations algebraically on the complete image. The implementation automatically parallelises the program, distributes the image data and controls all interprocess communication. The language is based on image algebra (a mathematical notation developed specifically for low-level image processing at the University of Florida). The paper describes the features of the language and provides an outline of some of the strategies used in implementing it on a transputer network.

DOI:10.1049/ip-i-2.1990.0025

出版商:IEE    

年代:1990

数据来源: IET

摘要:

A multi-layer network is described which is able to recognise simple shapes in a shift, size, and rotation invariant manner. The use of layers of units to smooth and then to shift the image eliminates the need for the very large numbers of cells which are often proposed in shift invariant networks. The network was trained using back-propagation and is not intended to be plausible as a model of biological vision at the level of cell and connection detail. Some interesting parallels with human vision are noted in the emergent behaviour of the network.

DOI:10.1049/ip-i-2.1990.0026

出版商:IEE    

年代:1990

数据来源: IET