RAVEGRID: RASTER-TO-VECTOR GRAPHICS FOR IMAGE DATA
Sriram Swaminarayan, Lakshman Prasad · OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) · 2007
The Internet has revolutionized our world by vastly increasing our means and ability of acquisition, representation, and communication of information.The ubiquity of image data on the Internet, as well as in the wide range of devices, from big screen TV s to hand-held cellular phones that support image display, calls for a scalable and more manipulable representation of imagery.Moreover, with the growing need for automating image-based search, object recognition, and image understanding, it is desirable to represent image content at a semantically higher level by means of tokens that support computer vision tasks.RaveGrid is new software for rapid conversion of raster digital images made up of pixels into vector graphical images comprised of polygons.This affords scalability of images for rendering on various display platforms and screen sizes, economy of representation, and enables object detection, extraction, and recognition .RaveGrid is a novel technology that has applications in Web graphics, image search engines for the Internet, computer vision, and computer art.RaveGrid is compatible with the new scalable vector graphics (SVG) standard of the World Wide Web Consortium, as well as with the well-established and widely used encapsulated postscript (EPS) format. FeaturesRaveGrid (RAster to VEctor GRaphics for Image Data) is software that decomposes a digital image consisting of pixels into polygons that correspond to visual features in the image.The polygonal decomposition can be varied from a high resolution, for high visual fidelity, to a low resolution, for detection and extraction of salient image features.Unlike most image processing software that operate on all the pixels in an image to achieve their goals, RaveGrid uses only a small subset of image pixels, namely edge pixels, to achieve polygonal image decomposition.This makes RaveGrid highly efficient and fast enough for real-time applications .Using geometry and principles of human visual perception, RaveGrid constructs polygons by relating salient image edges that bound each visual feature at any desired resolution .This gives RaveGrid its ability to obtain visually meaningful polygonal decompositions of images.Since the description of image regions in terms of polygons is, in general, more efficient than in terms of pixels, RaveGrid typically reduces the size of an image leading to image compression.The description of images in terms of polygons, termed vectorization, enables the scaling of an image to any arbitrary size, thus accommodating to the display screen of any device, large or small.RaveGrid' s characterization of an image in terms of features rather than pixels allows the querying of images in terms of shape, structure, and color, giving RaveGrid object detection and recognition capabilities.Vector images produced by RaveGrid can be written as text files describing polygons in terms of their vertex coordinates and a fill color.This allows easy reading of high-level structural and color information in images into other programs as well as makes the images available for a)