There ’s still no completely good and surefire method for locating undischarged ordinance after a state of war is over , but research worker atOhio State Universityhave found a way to harness mental image processing algorithms , power by simple machine learning , to study satellite mental imagery and situate hot smudge where UXO are potential to be located .
The researchers focused their exertion on a 100 - satisfying - klick expanse near Kampong Trabaek , Cambodia , which was the target of carpet - bombing missions carried out by the United States Air Force during the Vietnam War . The team was leave access to declassified military data that revealed that 3,205 bombs had been dropped in the area between 1970 and 1973 . Determining exactly how many of those bombs did n’t explode has dumbfound harder and harder as , six 10 later , nature has slowly reclaim the country ’s heaviest hit areas , hiding and obscuring the craters that are counted and used to make precise estimates .
TheOSU studyused a two - tone process to total up with a more accurate estimation of how many bombs were still left in the area . It started by leverage algorithmic program that had previously been developed for automatically greet and count volcanic crater on the lunation and other objects produce by meteor impacts . It was a start , but meteor craters are much with child than those pull up stakes by carpet - bombing campaigns , and more defined give that satellite imaging of the moon is gamey - line and not hide by nature or erosion .
Photo: (Ohio State University)
The orbiter imagery of the Kampuchean realm was much harder for those algorithmic rule to analyze , thanks to 60 geezerhood of plant spirit growth , erosion , and other innate phenomenon that have result the bomb crater looking nothing like the meteoroid craters on the moon . It was ab initio capable to key out possible crater , but the research worker had to go one stride further and civilise a neural web on a database of known orbiter volcanic crater images to make an additional algorithm that could take into score variations in sizing , form , color , textures , and other obscuring features .
The model was eventually able to correctly identify 152 of 177 known bomb craters , for an truth pace of about 86 pct , while right eliminating over 1,000 simulated positive that would have otherwise pine away UXO disposal efforts . The researchers say those results would further dud detection drive by over 160 percentage , compare to current methods . With a more precise reckoning of how many bomb actually explode during the war , it ’s now estimated there could still be over 1,600 undischarged turkey in the area study , accounting for upwards of half the bomb dropped decades ago .
satellite imagery
A comparison of satellite imagery featuring actual bomb creators and false positives created by ponds, buildings, and even variations in foliage.Photo: (Ohio State University)
Daily Newsletter
Get the undecomposed tech , scientific discipline , and finish word in your inbox daily .
News from the futurity , delivered to your present tense .
You May Also Like
