Detection of Signal Signatures of Cartographic Features,
Abstract
Currently photo interpretation and terrain image identification for military applications is performed manually. This work is time consuming and costly. The work described below is a first step to aiding or semi-automating some of these labor intensive tasks. A semi-automated technique for extracting a selected set of cartographic features such as, road intersections, straight line roads, and rectangular objects from aerial photographs was recently developed at the U.S. Army Engineer Topographic Laboratories (USAETL) using the Walsh transform. The discrete function (Walsh) transform was chosen because of its simplicity (Walsh functions are only two-valved) resulting in simple implementation, and because Walsh functions conform to the selected set of the cartographic features. Since the Walsh transform coefficients were produced by using Walsh functions having alternate magnitudes (either +1 or -1) at different sequences, the Walsh transform coefficients are also decomposed spectral components of the signal signature of the input aerial photographic images. The technique was investigated in two ways as follows: (1) Using a 32-by 32-element solid state sensor array to convert aerial imagery into an electronic signal which was processed in a minicomputer to yield Walsh transforms of the image. (2) A prototype image spectrum analyzer (PISA) was developed which utilizes a large size plasma discharge device (8.5-by 8.5-inch illuminating area with 512 electrodes each in both x and y directions) to generate two-dimensional Walsh function patterns, and produce 512 by 512 Walsh coefficients in 14 seconds.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1980
- Accession Number
- ADA090375
Entities
People
- Pi-fuay Chen
- William W. Seemuller
Organizations
- Geospatial Research Laboratory