Lakshman C Prasad

US Patent:

6937765, Aug 30, 2005

Filed:

Mar 14, 2003

Appl. No.:

10/388680

Inventors:

Alexei N. Skourikhine - Los Alamos NM, US
Lakshman Prasad - Los Alamos NM, US

Assignee:

The Regents of the University of California - Los Alamos NM

International Classification:

G06K009/48
G06T011/20

US Classification:

382199, 345442

Abstract:

Contours are extracted for representing a pixelated object in a background pixel field. An object pixel is located that is the start of a new contour for the object and identifying that pixel as the first pixel of the new contour. A first contour point is then located on the mid-point of a transition edge of the first pixel. A tracing direction from the first contour point is determined for tracing the new contour. Contour points on mid-points of pixel transition edges are sequentially located along the tracing direction until the first contour point is again encountered to complete tracing the new contour. The new contour is then added to a list of extracted contours that represent the object. The contour extraction process associates regions and contours by labeling all the contours belonging to the same object with the same label.

US Patent:

7127104, Oct 24, 2006

Filed:

Jul 7, 2004

Appl. No.:

10/887424

Inventors:

Lakshman Prasad - Los Alamos NM, US
Alexei N. Skourikhine - Los Alamos NM, US

Assignee:

The Regents of the University of California - Los Alamos NM

International Classification:

G06K 9/00
G06K 9/34

US Classification:

382164, 382173, 382197, 382199, 382241, 358538

Abstract:

A computer implemented method transforms an image comprised of pixels into a vectorized image specified by a plurality of polygons that can be subsequently used to aid in image processing and understanding. The pixelated image is processed to extract edge pixels that separate different colors and a constrained Delaunay triangulation of the edge pixels forms a plurality of triangles having edges that cover the pixelated image. A color for each one of the plurality of triangles is determined from the color pixels within each triangle. A filter is formed with a set of grouping rules related to features of the pixelated image and applied to the plurality of triangle edges to merge adjacent triangles consistent with the filter into polygons having a plurality of vertices. The pixelated image may be then reformed into an array of the polygons, that can be represented collectively and efficiently by standard vector image.

US Patent:

8428354, Apr 23, 2013

Filed:

Jun 23, 2010

Appl. No.:

12/822059

Inventors:

Lakshman Prasad - Los Alamos NM, US
Sriram Swaminarayan - Los Alamos NM, US

Assignee:

Los Alamos National Security, LLC - Los Alamos NM

International Classification:

G06K 9/34

US Classification:

382173

Abstract:

A method for rapid hierarchical image segmentation based on perceptually driven contour completion and scene statistics is disclosed. The method begins with an initial fine-scale segmentation of an image, such as obtained by perceptual completion of partial contours into polygonal regions using region-contour correspondences established by Delaunay triangulation of edge pixels as implemented in VISTA. The resulting polygons are analyzed with respect to their size and color/intensity distributions and the structural properties of their boundaries. Statistical estimates of granularity of size, similarity of color, texture, and saliency of intervening boundaries are computed and formulated into logical (Boolean) predicates. The combined satisfiability of these Boolean predicates by a pair of adjacent polygons at a given segmentation level qualifies them for merging into a larger polygon representing a coarser, larger-scale feature of the pixel image and collectively obtains the next level of polygonal segments in a hierarchy of fine-to-coarse segmentations. The iterative application of this process precipitates textured regions as polygons with highly convolved boundaries and helps distinguish them from objects which typically have more regular boundaries.

US Patent:

6393159, May 21, 2002

Filed:

Jun 29, 1999

Appl. No.:

09/345599

Inventors:

Lakshman Prasad - Los Alamos NM
Ramana Rao - Sunnyvale CA

Assignee:

The Regents of the University of California - Los Alamos CA

International Classification:

G06K 936

US Classification:

382259, 382181, 382203, 382204, 382206, 382241, 382256, 382257, 382308, 345420, 345423

Abstract:

An adaptive multiscale method approximates shapes with continuous or uniformly and densely sampled contours, with the purpose of sparsely and nonuniformly discretizing the boundaries of shapes at any prescribed resolution, while at the same time retaining the salient shape features at that resolution. In another aspect, a fundamental geometric filtering scheme using the Constrained Delaunay Triangulation (CDT) of polygonized shapes creates an efficient parsing of shapes into components that have semantic significance dependent only on the shapes structure and not on their representations per se. A shape skeletonization process generalizes to sparsely discretized shapes, with the additional benefit of prunability to filter out irrelevant and morphologically insignificant features. The skeletal representation of characters of varying thickness and the elimination of insignificant and noisy spurs and branches from the skeleton greatly increases the robustness, reliability and recognition rates of character recognition algorithms.