INTRODUCTION 



The concept of constructing a seafloor contour 

 map by means of stereo-sonar imagery has been inves- 

 tigated by Mittleman and Malloy of the Civil 

 Engineering Laboratory, Naval Construction 

 Battalion Center, Port Hueneme, California," [1]. 

 This investigation was mainly theoretical but included 

 some qualitative results of scanning the seafloor with 

 a pair of laterally separated, side-looking, sonar fish. 

 Because of virtually no precision in the estimate of 

 fish separation, it was not possible to compute rela- 

 tive target elevations. However, direct stereoscopic 

 viewing of some of the sonar-image pairs did result in 

 the blending of seafloor features into a three- 

 dimensional illusion. From these results it was 

 concluded that stereo-sonar techniques could be used 

 for contour mapping of the seafloor in the same way 

 that stereo-photo techniques are used to map land 

 areas. 



In June 1973, further sea trials were conducted 

 by CEL in an attempt to study quantitatively the 

 concept of stereo-sonar mapping. Two basic improve- 

 ments in stereo-sonar technology were made for these 

 sea trials: (1) a side-scan system was used having a 

 sonar frequency lower by an order of magnitude than 

 the frequency of the system used in the investigation 

 of Reference 1; and (2) the lateral separation of the 

 two sonar fish was doubled. Lowering the sonar 

 frequency allowed coverage of a relatively large sea- 

 floor area, yielding, in turn, more data for a given 

 operational period. It was assumed that increasing the 

 lateral separation of the two sonar fish and towing at 

 relatively short cable lengths would maintain a con- 

 stant horizontal separation of the two fish; and, 

 hence, target elevation errors would be less than 3 

 feet in 100 feet. 



The June 1973 work described in this report was 

 performed to produce sonar-image pairs which, not 

 only would blend into a three-dimensional picture 



when viewed stereoscopically, but also would allow 

 digital computation of seafloor contour data by inter- 

 facing the computer with the settings made by the 

 stereo operator. 



THE STEREO-SONAR PROBLEM 



Stereo plotting to construct land contour maps 

 from stereo-photo pairs is used because a human 

 observer with his two eyes can pick out points of 

 equal elevation with high precision. He can do this 

 more efficiently than a machine and at lower cost. A 

 machine subsystem can then be used to take the 

 stereo observer's adjustments and produce numerical 

 values for the contours. 



A basic problem in side-looking stereo-sonar 

 mapping is the requirement that the seafloor be 

 scanned a line at a time, instead of an area at a time 

 as in aerial stereo-photo mapping of land topography. 

 This requirement imposes a severe restriction on the 

 allowable magnitude of system errors because in the 

 line-scanning system, information is obtained at a 

 much slower rate than in the area-scanning system. 



Another problem in stereo-sonar is that sonar 

 measures range instead of angle as in optical imaging. 

 Whereas the eye-brain system directly yields precise 

 information on elevation differences if the image pair 

 is obtained optically, in the stereo viewing of sonar 

 image pairs, the eye-brain system yields false eleva- 

 tion differences. Hence, it becomes necessary to 

 transform the false elevations to true elevations by 

 using a set of equations. As outlined in Reference 2, 

 this has been done for radar-stereo imagery. Side- 

 looking airborne radar measures range in the same 

 way that side-scan sonar does; that is, by measuring 

 the travel time of the signal. 



Further problems arise when an attempt is made 

 to view stereo-sonar imagery in a stereoscope and to 

 create a three-dimensional illusion. The result is 



Designation prior to January 1, 1974: Naval Civil Engineering Laboratory, 

 Port Hueneme, California. 



