Sonar Map Stitching Lior Alezra Advisor: Dr. Gera Weiss Main - PowerPoint PPT Presentation

Sonar Map Stitching Lior Alezra Advisor: Dr. Gera Weiss

Main Purpose: š Build a desktop application that creates a map, which represent the submarine scan mission area, and display all the objects in the scanning path on the map.

Means: š The Front Looking Sonar (FLS) of the unmanned submarine (Hydro Camel) generates data during mission. š Data contains grayscale images, locations (Latitude and Longitude), azimuth, pitch ,roll angles etc.

Map creator Application Walkthrough_1 Mains screen Select .son Image file Settings

Map creator Application Walkthrough_2 Control the sonar image color: v Image Settings JET, BONE, COOL, COOPER, GREEN, HOT. Change resolution size. v Fix rotation feature – align the map into v the average azimuth of all sonar data Manually and Auto Intensity calibration. v NC – Noise cancellation feature. v DIST – min distance in meters of samples v Start range and length of sonar image v calibration

Map creator Application Walkthrough_3 Sonar Map Create map Open image

Map create algorithm š Move all over sonar data and fix boundaries for the map size: š Find min and max for lat lon values š Convert cm into pixels (depends on resolution selected for map) š Create new ppm file (image) with the boundaries calculated above š Move over the sonar data again and print sonar image on the map. š Convert sonar data into ppm format. š Rotate image according the azimuth value. š Print rotate image into the map file with print function – detailed in the end.

Real Data Example š THE OBJECT TARGET : The Detected Object: š identify a single object that watched in a mission scan: š The object detected at pings 1309-1315.

Real Data Example Map Output š THE FIRST TRY: š The Print function takes the average color of all overlaps pixels. š The object is inside the red circle š The object can be barely detected.

Real Data Example Map Output š THE SECOND TRY: Normal mode š The Print function takes the brightest pixel of all overlaps pixels. š The object is inside the red circle š Output map is very noisy, still the object can be detected better then the first try.

Real Data Example Map Output š THE THIRD TRY: Noise Cancelation mode š The Print function takes a mix of the two prev methods. The final pixel value is 90% of the second method+10% of the first method. š The object is inside the red circle š The output map is better then the prev maps, still let the user to select between the normal mode (method 2) or noise calculation method (method 3).

Real Data Example : Map Colors Color Bone Cool Cooper Green Hot Mode NC Normal