Automatic detection of impact craters on Al foils from the Stardust interstellar dust collector using convolutional neural networks
This addresses the challenge of locating sparse, sub-micrometer craters for analyzing interstellar dust samples, which is incremental as it applies an existing method to a specific domain.
The paper tackled the problem of automatically detecting tiny impact craters on aluminum foils from the Stardust mission, achieving high specificity and sensitivity using a convolutional neural network based on VGG16.
NASA's Stardust mission utilized a sample collector composed of aerogel and aluminum foil to return cometary and interstellar particles to Earth. Analysis of the aluminum foil begins with locating craters produced by hypervelocity impacts of cometary and interstellar dust. Interstellar dust craters are typically less than one micrometer in size and are sparsely distributed, making them difficult to find. In this paper, we describe a convolutional neural network based on the VGG16 architecture that achieves high specificity and sensitivity in locating impact craters in the Stardust interstellar collector foils. We evaluate its implications for current and future analyses of Stardust samples.