Title: An Ultrafast Image Recovery and Recognition System Implemented With Nanomagnets Possessing Biaxial Magnetocrystalline Anisotropy
Abstract: A circular magnetic disk with biaxial magnetocrystalline anisotropy has four stable magnetization states which can be used to encode a pixel's shade in a black/gray/white image. By solving the Landau–Lifshitz–Gilbert equation, we show that if moderate noise deflects the magnetization slightly from a stable state, it always returns to the original state, thereby automatically denoising the corrupted image. The same system can compare a noisy input image with a stored image and make a matching decision using magnetic tunnel junctions. These tasks are executed at ultrahigh speeds ( <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$\sim$</tex></formula> 2 ns for a <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$512 \times 512$</tex></formula> pixel image).