This paper describes the motivation behind the development of the Cellular neural Network Associative Processor (C-NNAP) and how the architecture is used for object recognition. It describes why neuralnetworks have n...
详细信息
This paper describes the motivation behind the development of the Cellular neural Network Associative Processor (C-NNAP) and how the architecture is used for object recognition. It describes why neuralnetworks have not been successful in object recognition problems and goes on to show how the C-NNAP architecture is able to implement a hierarchical object recognition system that overcomes these problems.
The proceedings contains 170 papers. Topics discussed include imageprocessing, image coding, labelling and classification, medical applications, motion, stereo and three dimensional, image analysis, image interpretat...
详细信息
The proceedings contains 170 papers. Topics discussed include imageprocessing, image coding, labelling and classification, medical applications, motion, stereo and three dimensional, image analysis, image interpretation, image coding and communications, shape description and recognition, imageprocessingapplications, computer architectures, image segmentation, neuralnetworks, industrial inspection, filtering and morphology, texture and color, transport, security and remote sensing.
In this paper an extension to the standard error back-propagation learning rule for multi-layer feed forward neuralnetworks is proposed, that enables them to trained for context dependent information. The context dep...
详细信息
In this paper an extension to the standard error back-propagation learning rule for multi-layer feed forward neuralnetworks is proposed, that enables them to trained for context dependent information. The context dependent learning is realised by using a different error function (called Average Risk: AVR) in stead of the sum of squared errors (SQE) normally used in error backpropagation and by adapting the update rules. It is shown that for applications where this context dependent information is important, a major improvement in performance is obtained.
Window filters are used in imageprocessing for reducing noise, enhancing contrast, edge detection, and other pre-processing operations. Selecting an appropriate filter for the task depend on the skill and experience ...
详细信息
Window filters are used in imageprocessing for reducing noise, enhancing contrast, edge detection, and other pre-processing operations. Selecting an appropriate filter for the task depend on the skill and experience of the imageprocessing specialist. In this article, the development of a UWF general purpose filter is discussed. In addition, the performance of the filter on simple emulation task, and the subsequent development of improved training methods based on analogy to the task of training people to perform complex tasks are also accounted.
This study addresses the rapidly advancing subject of visual science including its inherent interdisciplinary nature and the benefits to be gained by those involved in digital imageprocessing being aware of the subje...
详细信息
This study addresses the rapidly advancing subject of visual science including its inherent interdisciplinary nature and the benefits to be gained by those involved in digital imageprocessing being aware of the subject and its key results. It is demonstrated how imageprocessing can be inspired by the form, system or strategy of the visual systems of man and animals. The biological example used for this purpose is concerned with the application of instructional design theories developed for humans being applied to neuralnetworks.
neuralnetworks have been used to classify high resolution remote-sensed data. Experiments have demonstrated the potential of neuralnetworks for clustering large number of ground cover instances using supervised meth...
详细信息
neuralnetworks have been used to classify high resolution remote-sensed data. Experiments have demonstrated the potential of neuralnetworks for clustering large number of ground cover instances using supervised methods. This paper will describe a new algorithm of unsupervised learning, based on artificial neural network. Its performance has been compared with the competitive learning algorithm. The efficiency of this approach has been demonstrated through experimental results obtained on real-world of multispectral remote sensing data.
Automation of the image shearing measurement technique applied to optical fibre geometry is discussed. The use of both edge detection filters and a multi-layer perceptron neural network to identify the critical 'j...
详细信息
Automation of the image shearing measurement technique applied to optical fibre geometry is discussed. The use of both edge detection filters and a multi-layer perceptron neural network to identify the critical 'just touch' condition are compared. The repeatability of cladding diameter measurements using both methods are presented which demonstrate the superiority of the neural network technique.
Synthetic-Aperture Radar (SAR) is a high-resolution remote sensing platform with an all-weather capability. By mounting the SAR on an aircraft or spacecraft, long range imagery can be produced. However, the image prod...
详细信息
Synthetic-Aperture Radar (SAR) is a high-resolution remote sensing platform with an all-weather capability. By mounting the SAR on an aircraft or spacecraft, long range imagery can be produced. However, the image produced sometimes consists of speckle due to the scattering of the coherent radiation used in the imaging process. Given a high-quality, distortion free SAR image generated by a SAR processor, the present work aims to remove the speckle to reveal the underlying cross-section. To perform this, an approach which involve the training of a neural network to emulate the simulated annealing of SAR images is used.
Hopfield networks can be used as a tool for classification. However, the previous method of using the Hopfield network as an associative memory has certain shortcomings. In view of this, a new method for using the Hop...
详细信息
Hopfield networks can be used as a tool for classification. However, the previous method of using the Hopfield network as an associative memory has certain shortcomings. In view of this, a new method for using the Hopfield network for classification is proposed. In this method, the training elements are incorporated in the network as elements attracted to other elements rather than as attractors themselves. Finally, the efficiency and effectiveness of the method is verified by applying it to the diabetes database and to the classification of cervical cells.
This paper presents a robust neural network edge labelling strategy in which a network is trained with data from an imaging model of an ideal step edge. In addition to the Sobel operator, we employ pre-processing step...
详细信息
This paper presents a robust neural network edge labelling strategy in which a network is trained with data from an imaging model of an ideal step edge. In addition to the Sobel operator, we employ pre-processing steps on image data to exploit the known invariances due lighting variation and rotation and so reduce the complexity of the mapping which the network has to learn. The composition of the training set to achieve labelling of the image lattice with Bayesian posterior probabilities is described. Results are shown for real images and comparison made with the Canny edge detector, the effects of adding zero-mean Gaussian noise are also shown. To elucidate the roles of the Sobel operator and the network, a probabilistic Sobel labelling strategy has been derived - its results are inferior to those of the neural network.
暂无评论