摘要:
A feature regulation application method for hierarchical decision learning systems receives feature regulation training data and invokes a plurality of hierarchical decision learning to create feature subset information output. The method receives learning data and uses the feature subset information and the learning data to create feature subset learning data output. The hierarchical decision learning method uses the feature subset learning data to create hierarchical decision output. The feature regulation method also outputs feature ranking information that can be used to create hierarchical decision output. The invention provides a computationally feasible method for feature selection that considers the hierarchical decision learning systems used for decision making.
摘要:
Rotation and scale invariant profiles are generated for matching to a pattern template using a fast regular shaped pattern construction method. The invention achieves rotation invariant matching, rotation and scale invariant matching, and/or scale invariant matching. Invariant profiles are used to perform fast rotation, rotation and scale, or scale invariant search for initial detection of match candidates. The rotation invariant contours of this invention approximate circular contours through use of regular shaped patterns such as octagon or multiple rotated octagons. Rotation invariant search does not depend on rotation angles and is very fast.
摘要:
A fast matching method performs pattern decomposition and synthesis learning to create a pattern search recipe that is used by an invariant pattern search and synthesis method to generate the match result. The pattern search recipe includes template component invariant profiles, component weights, and allowable partial pattern configurations. The invariant matching method supports partial pattern match. This invention decomposes a template into multiple compact shaped components and performs search using separate rotation and scale invariant profiles for each component. It then synthesizes the search results for the complete template or partial template using the component search results.
摘要:
An accumulation method for fast pattern search can accurately locate regular shaped patterns of interest. The method can be used for invariant search to match patterns of interest in images where the searched pattern varies in size or orientation or aspect ratio, when pattern appearance is degraded, and even when the pattern is partially occluded, where the searched image is large, multidimensional, or very high resolution, or where the pattern size is large. The computation requirement is independent of the size of the pattern region.
摘要:
A systematic way of linking structure constraints of components of a common object and checking and resolving their inconsistency is used to improve detection results in image-based decision systems. A multilevel Chain-And-Tree (CAT) model is used to direct processing using both forward and backward scans through the related components. Since components occur as parts of an object, the context (relational structure) in which the component appears can be used to reduce noise and variation affects. In the method, object knowledge is translated into constraints between components. The constraints are used to enhance feature detection, defect detection, and measurement accuracy and consistency.
摘要:
A structure-guided automatic learning system for image feature enhancement uses a learning image together with an application domain structure and detection target specification to produce a feature enhancement image processing recipe. An enhancement goodness measure is used to select between alternatives in the learning process. The feature enhancement recipe is used in an application module to process input images and produce a feature enhanced image output. Calipers are used for application domain structure and detection target specification. To unify the processing steps for all caliper specifications, a non-directional box caliper defined region such as a circle caliper or an arc caliper or other connected structures can be converted into a directional box caliper defined region so that a directional box caliper based feature enhancement method can be applied. The process can be inverted to convert a converted directional box caliper region back to the original format.
摘要:
A robust biological specimen classifier. An automated microscopy system obtains features from an image of a biological specimen slide. A computer system in the automated microscopy system computes feature variations. Clusters are created that comprise samples of similar characteristics. A cluster membership generator generates membership values for each cluster. Classifiers specialized to certain feature sets are used to provide independent outputs. These outputs are offset and biased by the output of the membership generator. The output of the adjusted classification values are summed to create a slide score output.
摘要:
An automated biological screening system obtains biological and procedural data from a slide set of a selected clinical laboratory. The integration system tests the data on standardized criteria and passes and fails the data in selected categories. The results of the assessment are used to make process adjustment recommendations based on the results of a laboratory process adjustment procedure. Assessment and adjustment may continue until data from a slide set from the selected clinical laboratory passes in each category. The integration system then sets up, calibrates and installs the automated biological screening system. During operation, the integration system continuously monitors biological data generated by the automated biological screening system. The biological data may also be stored in a central product/service database for additional monitoring. The integration system also serves as an objective standard for reviewing and improving laboratory practices.
摘要:
A teachable object contour mapping method for region partition receives an object boundary and a teaching image. An object contour mapping recipe creation is performed using the object boundary and the teaching image to generate object contour mapping recipe output. An object contour mapping is applied to an application image using the object contour mapping recipe and the application image to generate object contour map output. An object region partition using the object contour map to generate object region partition output. An updateable object contour mapping method receives a contour mapping recipe and a validation image. An object contour mapping is performed using the object contour mapping recipe and the validation image to generate validation contour map output. An object region partition receives a region mask to generate validation object region partition output. A boundary correction is performed using the validation object region partition to generate corrected object boundary output. An update contour mapping is performed using the corrected object boundary, the validation image and the contour mapping recipe to generate updated contour mapping recipe output.
摘要:
A directed pattern enhancement method receives a learning image and pattern enhancement directive. Pattern enhancement learning is performed using the learning image and the pattern enhancement directive to generate pattern enhancement recipe. An application image is received and a pattern enhancement application is performed using the application image and the pattern enhancement recipe to generate pattern enhanced image. A recognition thresholding is performed using the pattern enhanced image to generate recognition result. The pattern enhancement directive consists of background directive, patterns to enhance directive, and patterns to suppress directive. An update learning method performs pattern enhancement progressive update learning.