Laser optical field reconstruction using deep learning and phase diversity

Wang, JikaiJikaiWangRavi, Sonam SmithaSonam SmithaRaviSchwartz, ManuelManuelSchwartzScharun, MichaelMichaelScharunDannecker, BenjaminBenjaminDanneckerBauer, DominikDominikBauerRominger, VolkerVolkerRomingerFlamm, DanielDanielFlammNolte, StefanStefanNolte2024-10-082024-10-082024https://publica.fraunhofer.de/handle/publica/47694910.1117/12.30143262-s2.0-85199011588We report on the recording of the near and far field intensity beam profiles to train a convolutional neural network, which is aimed to online detect system aberrations of an ultrafast laser amplifier. We extend the state of the art by implementing a spiral phase plate to use the concept of phase diversity. It is found that the underlying optical field in amplitude and phase can be accurately revealed.endeep learningField reconstructionlaser beam characterizationphase diversityLaser optical field reconstruction using deep learning and phase diversityconference paper