قالب وردپرس درنا توس
Home / WorldTech / Lunar craters help in the search for unexploded bombs from the Vietnam War

Lunar craters help in the search for unexploded bombs from the Vietnam War



There is still no fully safe and surefire way to locate unexploded ordinances after the war, but Ohio State University (1

9459008) researchers have found a way to use machine learning-powered image processing algorithms to study Satellite imagery and location of focus where UXO is likely to be.

Researchers focused their efforts on a 100 square kilometer area near Kampong Trabaek, Cambodia, the target of carpet bombing missions by the United States was the United States Air Force during the Vietnam War. The team was given access to released military data showing that 3,205 bombs had been dropped in the area between 1970 and 1973. Determining how many of these bombs did not detonate became increasingly difficult as nature slowed six decades later. The most affected areas in the country have been recaptured, hiding and hiding the craters counted and used for accurate estimates.

The OSU study used a two-step process to get a more accurate estimate of how many bombs were left in the area. Initially, algorithms were used that were previously developed to automatically detect and count craters on the moon and other objects caused by meteorite impacts. It was a start, but meteor craters are much larger and more defined than those left by carpet bombing campaigns because the moon's satellite imagery is rich in contrast and not obscured by nature or erosion.

  A comparison of the satellite images with actual bomb creators and false alarms generated by ponds, buildings and even leaf changes.

A comparison of satellite images with actual bomb creators and false alarms generated by ponds, buildings, and even leaf variations.
Photo : Ohio State University

The satellite imagery of the Cambodian region was in favor of these algorithms due to 60 years of plant growth, erosion and other natural phenomena have left the country much more difficult to analyze bomb craters that have nothing to do with the meteor craters on the moon. It was initially able to identify potential craters, but the researchers had to go a step further and train a neural network in a database of known satellite crater images to create an additional algorithm that takes variations in size, shape, color, and texture into account could and other opaque features.

The model was ultimately able to correctly identify 152 out of 177 known bomb craters with an accuracy rate of approximately 86 percent, while at the same time correctly eliminating over 1,000 false alarms that would otherwise have wasted efforts at UXO. The researchers say these results would increase bomb detection efforts by over 160 percent compared to current methods. With a more accurate count of the number of bombs actually detonated during the war, it is now estimated that there may still be over 1,600 unexploded bombs in the area under investigation, which is more than half of the bombs dropped decades ago.


Source link