Abstract: A method and system of signature verification that includes receiving a signature-in-question, extracting at least one of an image of the signature-in-question and a trajectory of the signature-in-question from the received signature-in-question, accessing a reference signature, extracting at least one of a reference image and a reference trajectory from the accessed reference signature, comparing the at least one of the image of the signature-in-question and trajectory of the signature-in-question to the at least one of the reference image and reference trajectory, and determining whether signature-in-question is authentic based on the comparison.

Abstract: Systems and methods herein can compensate for higher demand on computing processing functions by changing required error rates for error-controlled processes. Thus, the systems and methods maintain an average error rate over a time division. By adjusting the error rate and throughput within different time subintervals of the time division, the systems and methods can maintain an average error rate for the complete time division but adjust the throughput during smaller portions the time division. In other words, while maintaining the amount of processing capacity, the length of the time division, and the average error rate over the time division, but changing the error rate in smaller portions of the time division, the systems and methods can adjust the amount of total throughput during high demand periods for the processing function.

Abstract: Systems and methods herein can compensate for higher demand on computing processing functions by changing required error rates for error-controlled processes. Thus, the systems and methods maintain an average error rate over a time division. By adjusting the error rate and throughput within different time subintervals of the time division, the systems and methods can maintain an average error rate for the complete time division but adjust the throughput during smaller portions the time division. In other words, while maintaining the amount of processing capacity, the length of the time division, and the average error rate over the time division, but changing the error rate in smaller portions of the time division, the systems and methods can adjust the amount of total throughput during high demand periods for the processing function.

Abstract: Methods and systems are disclosed to aid in the detection of areas of interest in an image. Multiple image recognition processes analyze the image and identify areas of interest. The identified areas of interest are compared to determine confidence values for each identified area of interest using a voting process. The confidence values may be used in determining areas of increased interest which are highlighted on the image. In embodiments, identified areas of interest meeting a certain threshold requirement are selected as areas of increased interest. In other embodiments, new areas of increased interest are created by combining areas of interest. Embodiments of the disclosed methods and system may be used to aid in the detection of cancer in mammogram images.

Abstract: Methods and systems are disclosed to aid in the detection of cancer or lesion in a mammogram images. Two mammogram images are input into an application that aids in determining the probability of a cancer or lesion being present in one or both of the images. The images are divided into different nodes and labels are applied to the nodes. The first node is compared to different variants of corresponding nodes on the second image as well as neighboring nodes on the first image. Based upon the comparisons, a unary and binary potential is calculated for the label that is applied to the node. The process is repeated for every possible label and for every node. Once the unary and binary potentials have been calculated, the potentials are input into a Conditional Random Field model to determine the probability of cancer for each node of the images.

Abstract: Methods and systems are to facilitate the interaction between a user and a computer aided detection (“CAD”) system when identifying areas of interest in an image. The CAD system receives input from a user indicating areas of interest that the user perceives in the image. The input may include location information as well as additional information about the area of interest. The CAD system may then perform a voting process using the user provided areas of interest and one or more sets of areas of interest identified by at least one image recognition process to produce a final set of areas of interest in the image. The voting process is useful in eliminating false positives and false negatives that may have been identified by the user and/or the image recognition processes. Further embodiments disclosed herein provide methods and systems for training a voting process to complement a particular user.

Abstract: Methods and systems are disclosed to aid in the detection of areas of interest in an image. Multiple image recognition processes analyze the image and identify areas of interest. The identified areas of interest are compared to determine confidence values for each identified area of interest using a voting process. The confidence values may be used in determining areas of increased interest which are highlighted on the image. In embodiments, identified areas of interest meeting a certain threshold requirement are selected as areas of increased interest. In other embodiments, new areas of increased interest are created by combining areas of interest. Embodiments of the disclosed methods and system may be used to aid in the detection of cancer in mammogram images.

Abstract: In a combined holistic and analytic recognition system, a holistic recognition module recognizes an input word or phrase image by matching an input string of character features for the whole word or phrase against a string of prototype features for a plurality of reference words in a lexicon. This yields a holistic answer list of candidates for the input word or phrase. Based on each answer in the holistic answer list, the holistic recognition module creates segmentation hypotheses for each character in an answer. The analytical recognition module uses segmentation hypotheses from the holistic answer list to cut the image of the input string of characters into individual character images and to produce an analytical answer list. The holistic answer list and the analytic answer list are compared to find the best answer from the two lists.

Abstract: Methods and systems are disclosed for combining the results from multiple fraud detection devices into a single, ordered list of suspect items. The confidence value for each suspect item identified by each fraud detection process may be calculated. Calculated confidence values may be refined by resealing the confidence value. The confidence value of items from multiple lists may then be further refined using a voting process. Use of a voting process takes into account the determinations of each of the fraud detection processes in order to calculate or refine the confidence values associated with each suspect item. After conducting the voting process, the suspect items may be combined into a single suspect list. The new suspect list may be ordered by confidence values. Additionally, the expected loss of each item may be calculated for each suspect item. Suspect lists may also be ordered according to expected loss.

Abstract: Methods for displaying various user interfaces are disclosed for use in conjunction with image analysis software. User interfaces disclosed identify areas of interest on an analyzed image. The identified areas of interest are correlated with a confidence value displayed on the image as well, giving the user an idea of the likelihood that an identified area of interest corresponds to an actual area of interest on the image. The disclosed user interfaces may also provide a tool to users which allow the user to set and reset a threshold value or values used in analyzing the image. The displayed areas of interest may change due to the user setting a custom threshold value.

Abstract: Expert decision-making operations are trained to emulate expert behavior based on an history of behaviors by experts in a variety of observed situations. A history of behaviors is built up from observations of actions taken by experts in analyzing a plurality of situations. The observations are captured, and behaviors from the observations are constructed. The behaviors indicate an association between situation features and methods with parameter for solving the situations. The training operations capture observations of behavior by experts. The observations include situation data about multiple situations and actions by the experts. The actions are associated with the situations. Subject knowledge information is loaded from the observations; the subject knowledge information has a features library, a method library and a parameters library.

Abstract: An expert decision-making method is emulated based on a history of behaviors by experts in a variety of observed situations. The history of behaviors is built up from observations of actions taken by experts in analyzing a plurality of situations. Situation data representative of a situation to be processed is received, and situation features are extracted from the situation data. Each situation feature is associated with an expert behavior method used to process the situation. A behavior method is recognized from a pattern of situation features. Recognizing a behavior method is based on feature/method separation data in multidimensional space of features into areas with each area associated with a method used by experts. Parameter values for parameters in the recognized behavior method are calculated based on the situation features. The calculation of parameter values is accomplished by recognizing parameter calculation rules and calculating the parameter values using the rules.

Abstract: A check processing system for recognizing text on a check and processing the check using an interpretation for the recognized text is disclosed. The check processing system recognizes text on a check using a text recognition method wherein the text is compared to information entries in a customized recognition list stored in a restrictive lexicon. Thus, a process of creating restricted lexicons containing only customized recognition lists is disclosed. Like the text on the check, the information entries may be strings of characters, such as, without limitation, strings of numerical characters, alphabetic characters or a combination of alphabetic and numerical characters. Furthermore, the strings of characters for the text, as well as the information entries, may be a word or a phrase of words, such as payee names or memo categories. After the text is recognized, the check processing system processes the check using the interpretation by linking the check to the interpretation on an account statement.

Abstract: In a combined holistic and analytic recognition system, the holistic recognition module will recognize an input word or phrase image by matching an input string of character features for the whole word or phrase against a string of prototype features for a plurality of reference words in a lexicon. This will yield a holistic answer list of recognized word or phrase candidates for the input word or phrase along with a confidence value for each answer on the list. At the same time based on each answer in the answer list, the holistic recognition modules will generate a list of character features and segment the character features into sets for each character in an answer. The analytical recognition module uses segmentation hypotheses from the segmented character feature sets to cut the image of the input string of characters into individual character images.

Abstract: Images of handwritten cursive records are extracted, and an automated search on the images of the cursive records is performed based on an ASCII query for a record. A cursive equivalent of the ASCII query is matched to the images of the cursive records, and a similarity value is generated to indicate the extent of match between features of the cursive equivalent of the ASCII query and features of each cursive record. The records are sorted based upon their similarity value determined in the matching process. This provides a candidate list of cursive record images to be manually examined by a user for the purpose of making a final determination as to whether any of the cursive records on the candidate list satisfy the query.

Abstract: In a combined holistic and analytic recognition system, the holistic recognition module will recognize an input word or phrase image by matching an input string of character features for the whole word or phrase against a string of prototype features for a plurality of reference words in a lexicon. This will yield a holistic answer list of recognized word or phrase candidates for the input word or phrase along with a confidence value for each answer on the list. At the same time based on each answer in the answer list, the holistic recognition modules will generate a list of character features and segment the character features into sets for each character in an answer. The analytical recognition module uses segmentation hypotheses from the segmented character feature sets to cut the image of the input string of characters into individual character images.

Abstract: The invention improves the appearance of freehand drawn lines and shapes in an electronic document by first recognizing freehand drawn lines and shapes and generating a line made up of sequential straight line segments for the freehand drawn line when the line does not form a closed line and generating a multiple straight-line-segment shape when the line forms a closed line. If a multiple segment shape is being reshaped, a basic shape is selected from reference ideal shapes as the basic shape of the multiple segment shape. The basic shape is adjusted to provide a specific shape as an improved shape for the freehand drawn shape. The recognition of the freehand drawn lines and shapes is accomplished by comparing source segments of a source freehand drawn line to a straight line and substituting a straight line segment for a source segment if the deviation between the source segment and the straight line is below a predetermined value.

Abstract: Character model graphs are created, and the parameters of the model graphs are adjusted to optimize character recognition performed with the model graphs. In effect the character recognizer using the model graphs is trained. The model graphs are created in three stages. First, a vector quantization process is used on a set of raw samples of handwriting symbols to create a smaller set of generalized reference characters or symbols. Second, a character reference model graph structure is created by merging each generalized form model graph of the same character into a single character reference model graph. The merging is based on weighted Euclidian distance between parts of trajectory assigned to graph edges. As a last part of this second stage “type-similarity” vectors are assigned to model edges to describe similarities of given model edge to each shape and to each possible quantized value of other input graph edge parameters.

Abstract: A character recognizer recognizes a handwritten input character. A sequence of points in two dimensional space representative of a stroke trajectory forming the input character is gathered. An input Directed Acyclic Graph is built with nodes representative of singular points at the beginning, end, and along the trajectory of the input character and with edges between nodes representative of an edge trajectory formed by the sequence of points of the input character between the singular points. Each edge in the input graph is described based on the shape, orientation and pen lift of the edge trajectory that the edge represents. The input graph is evaluated against model graphs, which are also Directed Acyclic Graphs, for all possible characters to find a path through a model graph that produces a best path similarity score with a corresponding path through the input graph. The input character is identified as an answer character represented by the model graph producing the best path similarity score.

Abstract: The invention improves the appearance of freehand drawn lines and shapes in an electronic document by first recognizing freehand drawn lines and shapes and generating a line made up of sequential straight line segments for the freehand drawn line when the line does not form a closed line and generating a multiple straight-line-segment shape when the line forms a closed line. If a multiple segment shape is being reshaped, a basic shape is selected from reference ideal shapes as the basic shape of the multiple segment shape. The basic shape is adjusted to provide a specific shape as an improved shape for the freehand drawn shape. The recognition of the freehand drawn lines and shapes is accomplished by comparing source segments of a source freehand drawn line to a straight line and substituting a straight line segment for a source segment if the deviation between the source segment and the straight line is below a predetermined value.