Spatial Vision Processes: From the Optical Image to the Symbolic Structures of Contour Information

By National Aeronautics and Space Administration

The significance of machine and natural vision is discussed together with the need for a general approach to image acquisition and processing aimed at recognition. An exploratory scheme is proposed which encompasses the definition of spatial primitives, intrinsic image properties and sampling, 2-D edge detection at the smallest scale, the construction of spatial primitives from edges, and the isolation of contour information from textural information. Concepts drawn from or suggested by natural vision at both perceptual and physiological levels are relied upon heavily to guide the development of the overall scheme. The scheme is intended to provide a larger context in which to place the emerging technology of detector array focal-plane processors. The approach differs from many recent efforts in edge detection and image coding by emphasizing smallest scale edge detection as a foundation for multi-scale symbolic processing while diminishing somewhat the importance of image convolutions with multi-scale edge operators. Cursory treatments of information theory illustrate that the direct application of this theory to structural information in images could not be realized. Jobson, Daniel J. Langley Research Center RTOP 549-02-21-01...

• Format: Paperback
• Language: English
• ISBN: 1729009441
• ISBN13: 9781729009444
• Release Date: January 1
• Publisher: Independently Published
• Length: 32 Pages
• Weight: 0.22 lbs.
• Dimensions: 0.1" x 8.5" x 11.0"