International Journal of Pattern Recognition and Artificial Intelligence

The associative recall problem is treated by means of a novel neural net model. The classifier-based model (CBM) uses internal representations of objects automatically developed from observation. A feedback loop consisting of a Coulomb network is introduced into a classifier network. The classifier learns descriptions of various object types from observation of typical instances. The internal representations of the classifier are used by the Coulomb network to transform an incomplete or noisy input dynamically. Since the recall is based on abstract object descriptions, the output is not limited to an a priori specified collection of memories (objects). Results obtained for a simulation of the model show that it will retrieve a memory which satisfies the classification requirement and is compatible with the cue. The model can store any number of memories without degrading the quality of the memory retrieved. Either a schema or a distinct memory can be retrieved. The classifier prunes irrelevant objects from the object space, promoting faster convergence to an appropriate memory

The authors present a resource-level conflict detection and conflict resolution scheme which is combined with a state-level backward planning algorithm and provides efficient conflict detection and global conflict resolution for nonlinear planning. The scheme is to keep track of the usage of individual resources during planning and construct a resource-usage flow (RUF) structure (based on which conflict detection and resolution are accomplished). The RUF structure allows the system to perform minimal and nonredundant operations for conflict detection and resolution. Furthermore, resource-level conflict detection and resolution facilitates problem decomposition in terms of resources, thereby providing easy implementation in a parallel and distributed processing environment. Performance analysis indicates that the proposed architecture has a speed-up factor of the average depth of a plan network, D ( N _a), compared to the distributed NOAH where N _a (the total number of action nodes at the completion of planning) and D ( N _a) are considerably larger than the number of resources involved in planning as well as the number of initial goal states

Recently, local stereo matching has experienced large progress by the introduction of adaptive support-weights. In this paper, we aim at eliminating negative effects of occlusions by proposing an occlusion-based method to improve traditional support weights. Weights of occluded points are greatly reduced while computing matching costs, initial disparities and final disparities. Experimental results on the Middlebury images demonstrate that our method is very effective in improving disparities of points around occluded areas and depth discontinuities. According to the Middlebury benchmark, the proposed algorithm is now the top performer among local stereo methods. Moreover, this approach can be easily integrated into nearly all existing support weights strategies.

The authors present two parallel algorithms to perform template matching of an N × N image with an M × M template on an SIMD mesh connected computer with N × N processors. In the first algorithm, both the image and the template are represented by quadtrees, whereas in the second, the template is represented by a quadtree, and the image is represented by a matrix. The time complexities of the two algorithms are respectively upperbounded by α₁N + β₁M², and β₂M², where α₁, β₁, and β₂ are constants

A system of many-valued logical equations and its solving algorithm are presented. Based on this work, the authors generalize SLD resolution into many-valued logic and establish the corresponding truth-value calculus. As a result, M, an approximate reasoning system is constructed. Language and inference rules in M are presented. Inconsistencies of assignments and solving strategies are also analyzed in detail

An effort to incorporate machine learning capabilities within a general-purpose frame-based architecture is discussed. The authors describe Chunker, an explanation-based chunking mechanism built on top of Theo, a software framework to support development of self-modifying problem-solving systems. Chunker forms rules that improve problem-solving efficiency by generalizing and compressing the chains of inference which Theo produces during problem solving. After presenting the learning algorithm used by Chunker, the authors illustrate its application to learning search control knowledge, discuss its relationship to Theo's other three learning mechanisms, and consider the relationship between architectural features of Theo and the effectiveness of Chunker

The relationship between pattern classification using statistical decision theory and some feedforward neural networks are examined. The performance of eight networks is compared with that of the k -nearest neighbor decision rule with respect to memory requirements, computation time for classification, training time and adaptation or generalization capability

The paper proposes a general formalism for representation, inference and learning with general hybrid Bayesian networks. The formalism fuzzifies a hybrid Bayesian network into two alternative forms, which are called fuzzy Bayesian network (FBN) form-I and form-II. The first form replaces each continuous variable in the given directed acyclic graph (DAG) with a partner discrete variable and adding a directed link from the partner discrete variable to the continuous one. The second form only replaces each continuous variable whose descendants include discrete variables with a partner discrete variable and adding a directed link from that partner discrete variable to the continuous one. For the two forms of FBN, general inference algorithms exist which are extensions of the junction tree inference algorithm for discrete Bayesian networks

Efficient ways to prioritize and gather evidence within belief networks are discussed. The authors also suggest ways in which one can structure a large problem (a ship classification problem in the present case) into a series of small ones. This both re-defines much of the control strategy into the system structure and also localizes run-time control issues into much smaller networks. The overall control strategy thus includes the combination of both of these methods. By combining them correctly one can reduce the amount of dynamic computation required during run-time, and thus improve the responsiveness of the system. When dealing with the ship classification problem, the techniques described appear to work well

This paper describes a new system for extracting and classifying bibliography regions from the color image of a book cover. The system consists of three major components: preprocessing, color space segmentation, and test region extraction and classification. Preprocessing extracts the edge lines of the book, gets the basic information, and geometrically corrects and segments the input image, into the parts of front cover, spine, and back cover. The same as in all color image processing research, the segmentation of the color space is an essential and important step here. Instead of the RGB color space, the HSI color space is used in this system. The color space is segmented into achromatic and chromatic regions first; and both the achromatic and chromatic regions are segmented further to complete color space segmentation. Then test region extraction and classification follows. After detecting fundamental features (stroke width and local label width) text regions are determined. By comparing the text regions on the front cover with those on the spine, all extracted test regions are classified into suitable bibliography categories: author, title, publisher, and other information, without applying OCR

The application of a connectionist model to an image binarization method called connectionist model binarization (CMB) is discussed. CMB employs a multilayer network of a connectionist model whose input and output are a histogram and a desirable threshold for binarization, respectively. This network is trained with a back-propagation algorithm to output a threshold which gives a visually suitable binarised image against any histogram. The details of CMB are described, and its learning strategy and binarization performance are discussed

In order to overcome the shortcomings of Principal Component Analysis (PCA) and bionic methods in feature extracting and dimension reduction, a method for extracting Gabor features of face images based on Gabor wavelet is presented. First, Gabor features are extracted from face images. After reduced by 2DPCA algorithm, the features are reduced further by rough set. Then the nearest classifier is trained for classification. The experiments being performed on Yale Face Database B human face image database show the method presented in this paper is superior to that of bionic recognition algorithm and PCA algorithm.

In this paper, we propose a matching method for images captured at different times and under different capturing conditions. Our method is designed for change detection in street scapes using normal automobiles that has an off-the-shelf car mounted camera and a GPS. Therefore, we should analyze low-resolution and low frame-rate images captured asynchronously. To cope with this difficulty, previous and current panoramic images are created from sequential images which are rectified based on the view direction of a camera, and then compared. In addition, in order to allow the matching method to be applicable to images captured under varying conditions, (1) for different lanes, enlarged/reduced panoramic images are compared with each other, and (2) robustness to noises and changes in illumination is improved by the edge features. To confirm the effectiveness of the proposed method, we conducted experiments matching real images captured under various capturing conditions.

A neural network of cascaded restricted Coulomb energy (RCE) networks is constructed for object recognition. A number of RCE networks are cascaded together to form a classifier where the overlapping decision regions in a previously learned network are solved by the next network. The similarities among objects which have complex decision boundaries in the feature space are resolved by this multinetworks approach. The generalization ability of a RCE network recognition system, referring to the ability of the system to correctly recognize a new pattern even when the number of learning exemplars is small, is increased by the proposed coarse-to-fine learning strategy. A new feature extraction technique is proposed for mapping the geometrical shape information of an object into an ordered feature vector of fixed length which is the required form for input to this neural network

A new approach to the partitioning of large dictionaries by virtual views is presented. The basic idea is that additional knowledge sources of text recognition and text analysis are employed for fast dictionary look-up in order to prune the search space through static or dynamic views. The heart of the system is a redundant hashing technique which involves a set of hash functions dealing with noisy input efficiently. Currently, the system is composed of two main system components: the dictionary generator and the dictionary controller. While the dictionary generator initially builds the system by using profiles and source dictionaries, the controller allows the flexible integration of different search heuristics. Results prove that the system achieves a respectable speed-up of dictionary access time

This paper presents a probabilistic approach to recognize handwritten Chinese characters. According to the stroke writing sequence, strokes and interleaved stroke relations are built manually as a 1D string, called online models, to describe a Chinese character. The recognition problem is formulated as an optimization process in a multistage directed graph, where the number of stages is the length of the modelled stroke sequence. Nodes in a stage represent extracted strokes. The Viterbi algorithm, which can handle stroke insertion, deletion, splitting, and merging, is applied to compute the similarity between each modelled character and the unknown character. The unknown character is recognized as the one with the highest similarity. Experiments with 500 characters uniformly selected from the database CCL/HCCR1 are conducted, and the recognition rate is about 94.3%

Proposes an algorithm, which satisfies real time process requirements and has a high recognition rate, for recognizing a limited set of multifont Chinese characters used in a special application. Stroke is one of the major features of Chinese characters. Since the aim of the proposed method is to recognize a special set of printed Chinese characters, only the horizontal/vertical strokes and the crossings among these two types of strokes are used as the features of characters. In the learning stage, a random model for each type of characters is established. In the recognition stage, the character model which has the minimal error with the input pattern is considered as the recognition result. The proposed recognizer is insensitive to noise. Some experimental results show that the method exactly provides high recognition rate and has high computing speed

This paper presents a weighted support vector machine (WSVM) to improve the outlier sensitivity problem of standard support vector machine (SVM) for two-class data classification. The basic idea is to assign different weights to different data points such that the WSVM training algorithm learns the decision surface according to the relative importance of data points in the training data set. The weights used in WSVM are generated by kernel-based possibilistic c-means (KPCM) algorithm, whose partition generates relative high values for important data points but low values for outliers. Experimental results indicate that the proposed method reduces the affect of outliers and yields higher classification rate than standard SVM does when outliers exist in the training data set.

Transferring the current Websites to Semantic Websites, using ontology population, is a research area within which classification has the main role. The existing classification algorithms and single level execution of them are insufficient on web data. Moreover, because of the variety in the context and structure of even common domain Websites, there is a lack of training data. In this paper we had three experiences: 1- using information in domain ontology about the layers of classes to train classifiers (layered classification) with improvement up to 10% on accuracy of classification. 2- experience on problem of training dataset and using clustering as a preprocess. 3- using ensembles to benefit from both two methods. Beside the improvement of accuracy from these experiences, we found out that with ensemble we can dispense with the algorithm of classification and use a simple classification like Naïve Bayes and have the accuracy of complex algorithms like SVM.

The authors present the concept of a `segmental neural net' (SNN) for phonetic modeling in continuous speech recognition (CSR) and demonstrate how than can be used with a multiple hypothesis (or N-Best) paradigm to combine different CSR systems. In particular, they have developed a system that combines the SNN with a hidden Markov model (HMM) system. They believe that this is the first system incorporating a neural network for which the performance has exceeded the state of the art in large-vocabulary, continuous speech recognition. To take advantage of the training and decoding speed of HMMs, the authors have developed a novel hybrid SNN/HMM system that combines the advantages of both types of approaches. In this hybrid system, use is made of the N-best paradigm to generate likely phonetic segmentations, which are then scored by the SNN. The HMM and SNN scores are then combined to optimize performance

The authors investigate the path-planning problem for two planar robot manipulators which cooperate in carrying a rectangular object from an initial position and orientation to a destination position and orientation in a two-dimensional environment. To approach the problem, they model the robot arms, the carried object, and the straight line connecting two of the robot bases as a six-link closed chain. The path-planning problem for the six-link closed chain is solved by the collision-free feasible configuration finding algorithm and the collision-free path-finding algorithm. The former finds all collision-free feasible configurations (CFFCs) for the six-link closed chain at each quantized interval for two of the six joint angles. The latter builds a connection graph of the CFFCs and the transitions between any two groups of CFFCs at adjacent joint intervals. A graph search method is used to find a collision-free path for each robot joint. The algorithms can deal with cluttered environments and can be proved to converge

A feasible approach is proposed to determine the 3D location and orientation of a cylinder with a label stuck on it at a known height. The proposed approach uses a rectangle-shaped standard mark composed of two point-pairs for performing monocular image analysis. A monocular image of a labelled cylinder is first taken. Image processing and numerical analysis techniques are then used to select two point-pairs located on the boundary of the visible part of the cylindrical label. According to 3D imaging geometry, the location and orientation of the cylinder relative to the camera are uniquely determined by using 3D vector analysis and simple algebraic computation. Owing to the full linearity of the derivation, this approach can perform at high speed

This paper proposes an efficient method for detecting ghost and left objects in surveillance video, which, if not identified, may lead to errors or wasted computation in background modeling and object tracking in surveillance systems. This method contains two main steps: the first one is to detect stationary objects, which narrows down the evaluation targets to a very small number of foreground blobs; the second step is to discriminate the candidates between ghost and left objects. For the first step, we introduce a novel stationary object detection method based on continuous object tracking and shape matching. For the second step, we propose a fast and robust inpainting method to differentiate between ghost and left objects by constructing the real background using the candidate 's corresponding regions in the input and the background images. The effectiveness of our method has been validated by experiments over a variety of video sequences.

The authors describe their system for writer independent, off-line unconstrained handwritten word recognition. First, they present a new method to automatically determine the parameters of Gabor filters to extract features from slant and tilt corrected images. An algorithm is also developed to translate 2D images to 1D domain. Finally, they propose a modified dynamic programming method with fuzzy inference to recognize words. Their initial experiments have shown encouraging results

Since networks of main roads are basic information for classification of the use of the Earth's surface, automatic detection of roads from satellite images is a very important problem. In this paper, a new detection theory is proposed which can overcome drawbacks of current theories and detect plural roads in an image with high speed and high precision. Firstly, a binary image representing edges of a given image is used to evaluate the possibility for a road to pass on each edge pixel. We propose an 8-direction filter, a clearing filter, and a parallel-edge-detection filter, which can compile insufficient local information to obtain global information enough to detect a road. After these filters, the possibility of a road passing on a edge pixel can be effectively evaluated. Secondly, by using the Hough transform and the optimal search method it is possible to detect a complete road. This detection theory does not depend on the size of image and can detect almost all the main roads in an image including intersecting and T-type roads