Eastman-kodak Company

Hewlett-packard Development Company, L.p.

Purdue Research Foundation

Rochester Institute of Technology

Peter G Anderson

Charles Addison Bouman

Roger S Gaborski

Christopher T Asbury

Changmeng Liu

David G Tilley

Methods, devices and systems for compressing images are provided. One method includes creating halftone mask structures and applying compression coding techniques to arrayed pixels sorted using the halftone mask structures in order to convert an image to a compressed multi-level, halftoned image.

Methods, devices, and systems for creating and compressing multi-level halftones

Methods, devices and systems for compressing images are provided. One method includes creating halftone mask structures and applying compression coding techniques to arrayed pixels sorted using the halftone mask structures in order to convert an image to a compressed bi-level, halftoned image.

Methods, devices and systems for compressing images

A half-tone mask is characterized by an N.times.M array of mask value locations associated with at least a portion of the pixels of the digital image. The size of the mask array may contain a relatively small number of pixel locations, which is applied spatially to mutually contiguous subsets or sub-arrays of pixels of a relatively high resolution digital image. Although the mask may also be made large enough to encompass the entire image of interest, respective threshold values of the mask may be computed `on the fly`, so as to obviate the need to prestore all of the threshold values of the mask. Each of the locations of the mask array is assigned a respectively different grey scale/half-tone conversion threshold value. These threshold values are distributed among the locations of the mask array such that as one proceeds through the mask array, the next sequential location is specified in accordance with fractional parts of integral multiples of a two-dimensional analog of the Golden Mean. Each successive mask location is assigned a sequentially larger grey scale to half-tone conversion threshold. These threshold values are successive integers 0 through K, where K is one less than M.times.N. The use of a two-dimensional analog of the Golden Mean to algebraically distribute the threshold values disperses successive threshold value integers, so that any half-tone image obtained by the mask is not contaminated with low-frequency, visually disturbing artifacts.

Half-tone conversion mask for digital images

A system that classifies or identifies hand printed or machine printed alphanumeric characters using an iterated least squares polynomial discriminant method is disclosed. Super-sampling of labeled training characters is performed prior to feature extraction. Feature extraction subsamples individual character images. The feature vectors and associated labelings are used to create a weight matrix. During iteration, the weight matrix, to be subsequently used for identification, is modified by determining which characters are incorrectly classified, or classified with too small a confidence, and replicating those characters during training to strengthen the correct classification. The correct classification is also strengthened by using negative feedback to inhibit incorrect classification by using negative values in the target vectors. The speed of the learning process is enhanced by maintaining intermediate matrices and step wise increasing the amount of each feature vector and the size of the sample set used during training. The memory efficiency utilization is enhanced by maintaining and modifying intermediate matrices and compressing the sparse binary features vectors.

Supervised training augmented polynomial method and apparatus for

Methods and apparatus for constructing a classification weights matrix for a pattern recognition system are provided which enable large system feature sets to be reduced and yield at least the same level of performance achieved using the large feature set. Methods and apparatus are also described for determining (evaluating) the classification efficiency of selected subsets of a given feature set. Further aspects of the invention are directed to: (a) methods and apparatus for constructing reduced element classification weights matrices utilizing a genetic search process to find the subset having a maximum classification efficiency; and (b) pattern recognition systems (including, in particular, character identification systems), which utilize classifiers constructed in accordance with the aforementioned aspects of the invention to actually perform pattern recognition.

Growing community of inventors

Pittsford, NY, United States of America

Peter G Anderson