An error-less image compression of complex images has been achieved using a massively parallel computer. Algorithm involves utilization of a multi-level hierarchical structure of Kohonen type self-organizing learning ...
详细信息
The Conference materials contain 69 papers. Robotic vision systems, machine vision inspection techniques, segmentation of fused range/ intensity imagery, parallel and VLSI architectures for machine vision, comparison ...
详细信息
ISBN:
(纸本)0819408735
The Conference materials contain 69 papers. Robotic vision systems, machine vision inspection techniques, segmentation of fused range/ intensity imagery, parallel and VLSI architectures for machine vision, comparison of range segmentation algorithms, state of the arts in post-canny edge detection, simulation and visualization environments for autonomous robots, robot reach/grasp operations, fuzzymorphological neuralnetworks, reasoning techniques for vision systems, recognition techniques, object representation and matching, reactive robotic control strategies, and imageprocessing techniques are the main topics covered.
Work is being undertaken at Warwick University to analyse single photon emission tomography (SPET) images of the brain using artificialneuralnetworks. Considerable success has been achieved using the multi-layer per...
详细信息
Work is being undertaken at Warwick University to analyse single photon emission tomography (SPET) images of the brain using artificialneuralnetworks. Considerable success has been achieved using the multi-layer perceptron (MLP) topology together with a back propagation learning algorithm. The paper examines the application of a different type of network constructed of logical neurons to the same problem.< >
Aims to show that it is possible to solve the tracking and classification problems, successfully with the help of neuralnetworks. The authors propose two neuralnetworks, one for the visual tracking of moving objects...
详细信息
Aims to show that it is possible to solve the tracking and classification problems, successfully with the help of neuralnetworks. The authors propose two neuralnetworks, one for the visual tracking of moving objects and another one for their classification. The input images are taken sequentially from digitized TV pictures showing the 3-D objects movement along its trajectories.< >
Discusses the application of artificialneuralnetworks to texture analysis. Such networks we used as classifiers in order to distinguish among various classes of texture. A set of experiments was carried out in order...
详细信息
Discusses the application of artificialneuralnetworks to texture analysis. Such networks we used as classifiers in order to distinguish among various classes of texture. A set of experiments was carried out in order to study the network's classification performance as a function of both its size and the training volume. Another set of experiments was carried out in order to explore the effect of training strategy on the overall performance. The latter shows the effect of the input data arrangement on the learning rate and overall classification performance of the network. Recognition rates of up to 94.4 percent were achieved for tests made on unseen data.< >
Light intensity patterns produce at the output of a multi-mode optical fibre are complex in nature due to the interference of the many modes which propagate down its length. Small strains exerted on such a fibre chang...
详细信息
Light intensity patterns produce at the output of a multi-mode optical fibre are complex in nature due to the interference of the many modes which propagate down its length. Small strains exerted on such a fibre change its physical properties, such as length and refractive index, and hence alter the propagation characteristics of the light as it travels along the fibre. These effects result in an alteration of the complex modal interference pattern at the fibre output. It is possible to produce unique patterns at the fibre output over a wide range of strains, and use an artificialneural network as a means of associating a specific strain with a given pattern. This has allowed the construction of a simple, yet effective strain sensor. The system described consists of three basic blocks: a multi-mode fibre acting as a sensor; digital signal processing of the output pattern after acquisition; and finally, a neural network as the interface between the pattern and its corresponding strain value.< >
Introduces a method capable of recognising an aircraft from any viewpoint in uncluttered noisy scenes. The authors also describe a proposed exploitation of a recently developed hypermedia style user-interface for pres...
详细信息
Introduces a method capable of recognising an aircraft from any viewpoint in uncluttered noisy scenes. The authors also describe a proposed exploitation of a recently developed hypermedia style user-interface for presenting and manipulating information, and for enabling users to specify imageprocessing and pattern recognition operations. The paper is structured in two main parts. The first part describes the pattern recognition process and includes a description of the imageprocessing procedures, a detailed description of the Quadtree and normalised Quadtree data structures, the theory and properties of the Octree data structure, and finally the actual method used in order to train and test the artificialneural network. The second part describes an initial presentation of the user-interface graphical environment and its corresponding sub-parts for each individual stage of the pattern recognition process.< >
Adaptive Solutions' CNAPS architecture is a parallel array of digital processors. This design features a Single-Instruction Multiple-Data (SIMD) stream architecture. The architecture is designed to execute on- chi...
详细信息
ISBN:
(纸本)0819405515
Adaptive Solutions' CNAPS architecture is a parallel array of digital processors. This design features a Single-Instruction Multiple-Data (SIMD) stream architecture. The architecture is designed to execute on- chip learning for artificialneural Network (ANN) algorithms with unprecedented performance. ANNs have shown impressive results for solving difficult imageprocessing tasks. However, current hardware prevents many ANN solutions from being effective products. The CNAPS architecture will provide the computational power to allow real time ANN applications. Because of the high parallelism of the architecture,it is also ideal for digital imageprocessing tasks. This architecture will allow high performance applications that combine conventional imageprocessing methods and ANNs on the same system. This paper gives a brief introduction to the CNAPS architecture, and gives the system performance on implementation of neural network algorithms, and conventional imageprocessing algorithms such as convolution, and 2D Fourier transforms.
This paper describes a range of neural signal processing methods employed for B-Scan ultrsonic image enhancement and material identification. All approaches assume no a-priori knowledge of the environment. The potenti...
详细信息
ISBN:
(纸本)0852965311
This paper describes a range of neural signal processing methods employed for B-Scan ultrsonic image enhancement and material identification. All approaches assume no a-priori knowledge of the environment. The potential of a neural, sonar based material identification system has also been established.
The competitive learning technique is a well-known algorithm used in neuralnetworks which classifies the input vectors, so that the vectors (samples) belonging to the same class have similar characteristics. Each cla...
详细信息
ISBN:
(纸本)0819405787
The competitive learning technique is a well-known algorithm used in neuralnetworks which classifies the input vectors, so that the vectors (samples) belonging to the same class have similar characteristics. Each class is represented by one unit. Dynamic competitive learning is an unsupervised learning technique consisting of two additional parts related to conventional competitive learning: a method of generation of new units within a cluster and a method of generating new clusters. As seen in a description of the multilayered neuralnetworks, the number of clusters, their connections, and the generation of new units is determined dynamically during learning. The model is capable of high-level storage of complex data structures and their classification, including exception handling.
暂无评论