Abstract: The invention provides a perceptually-based system for pattern retrieval and matching, suitable for use in a wide variety of information processing applications. An illustrative embodiment of the system uses a predetermined vocabulary comprising one or more dimensions to extract color and texture information from an information signal, e.g., an image, selected by a user. The system then generates a distance measure characterizing the relationship of the selected image to another image stored in a database, by applying a grammar, comprising a set of predetermined rules, to the color and texture information extracted from the selected image and corresponding color and texture information associated with the stored image. The vocabulary may include dimensions such as overall color, directionality and orientation, regularity and placement, color purity, and pattern complexity and heaviness.

Abstract: The invention provides a perceptually-based system for pattern retrieval and matching, suitable for use in a wide variety of information processing applications. An illustrative embodiment of the system uses a predetermined vocabulary comprising one or more dimensions to extract color and texture information from an information signal, e.g., an image, selected by a user. The system then generates a distance measure characterizing the relationship of the selected image to another image stored in a database, by applying a grammar, comprising a set of predetermined rules, to the color and texture information extracted from the selected image and corresponding color and texture information associated with the stored image. The vocabulary may include dimensions such as overall color, directionality and orientation, regularity and placement, color purity, and pattern complexity and heaviness.

Abstract: A method for comparing a ranking scheme for an object, such as a document image, with a known standard, or ideal ranking scheme for a comparative assessment of the performance and reliability of the two ranking schemes, is taught. The same set of objects are processed and ranked by both ranking schemes. The ranking scheme being tested is examined for all subject objects ranked lower than its ranking by the ideal ranking scheme. The test ranking scheme is charged for each swapped object which it ranked higher than the subject object, which the ideal ranking scheme ranked lower.

Abstract: A technique analyzing loosely constrained text blocks, such as e-mail signature blocks by performing a two-dimensional geometrical analysis and a one-dimensional language analysis in order to classify sub-blocks at the loosely constrained text block into particular functional classes. The present technique may also be utilized to identify a personal name from a user name in a loosely constrained text block, such as an e-mail signature block.

Abstract: A technique for segmenting a loosely constrained text block, such as an e-mail signature block into sub-blocks by performing line segment extraction and connected component analysis on the foreground characters and background characters and recursively repeating connected component analysis on both the foreground and background characters and line segment extraction on the background characters until a text output includes no mixed reading blocks. A technique for correcting over segmentation errors in a line of text from a loosely constrained text block which has undergone geometrical analysis.

Abstract: Handwriting recognition which is invariant with respect to translation, rotation and scale is achieved with a new feature signal, ratio of tangents, and a new application of the normalized curvature feature. Use of these features is optimized by augmenting the ratio of tangents with the sign of local curvature and weighing each feature signal with its relative discriminative power.

Abstract: A signature verification method involves segmenting a smoothed and normalized signature and, for each segment, evaluating at least one local feature, thereby to obtain a feature-value vector. The method further includes computing the likelihood that the segments having the observed sequence of feature values were generated by a particular hidden Markov model, such model comprising a sequence of states having duration probabilities and feature-value histograms. In one aspect, the invention involves explicit state-duration modeling to account for duration in the respective states underlying the unknown signature. In another aspect, the invention involves computing a measure of signature complexity, and setting the number of states of the hidden Markov model according to such complexity measure.

Abstract: A method of handwriting recognition is provided that combines the efficiency of a point oriented system and the shape information of a segment oriented system in an HMM based handwriting recognition system. The partial segmentation hypotheses obtained using the point oriented features in a conventional Viterbi search are augmented with scores based on segmental shape measurements made on the hypothesized segments. In addition, a method for segmentation of a handwriting sample is accomplished by incorporating a feature to distinguish between continuous and not continuous strokes and influencing the indicated hypothesis score by adding a penalty to the contraindicated hypothesis score.

Abstract: Methods and systems for performing handwriting recognition which include, in part, application of stochastic modeling techniques in conjunction with language modeling. Handwriting recognition is performed on a received data set, which is representative of a handwriting sample comprised of one or more symbols. Recognition is performed by selectively segmenting the data set into one or more strokes utilizing an evolution grammar for identifying each one of the strokes among one or more alternatives. Each one of the strokes represents a segment of the handwriting sample. The identified strokes are evaluated as a stroke sequence, representative of one or more of the handwriting sample's symbols, to identify the handwriting sample.

Abstract: Handwriting recognition which is invariant with respect to translation, rotation and scale is achieved with a new feature signal, ratio of tangents, and a new application of the normalized curvature feature. Use of these features is optimized by augmenting the ratio of tangents with the sign of local curvature and weighing each feature signal with its relative discriminative power.

Abstract: Methods and systems for performing handwriting recognition which include, in part, application of stochastic modeling techniques in conjunction with language modeling. Handwriting recognition is performed on a received data set, which is representative of a handwriting sample comprised of one or more symbols. Recognition is performed by representing the data set as a sequence of features and then processing the features utilizing stochastic modeling in conjunction with an evolutional grammer for performing stroke identification, to identify the handwriting sample.