神经环路的宽视场高分辨率功能重建一直是脑科学研究的热点和难点问题。受制于数据通量的约束，现有光学显微成像技术在研究大尺度神经环路功能连接时，无法重建完整功能环路。本项目拟开展基于计算照明的光场显微成像研究：针对成像速度和信噪比的固有矛盾，构建基于计算摄像学的结构光场成像模型，提出基于机器学习的结构照明智能评估机制，揭示结构照明优化策略的工作机理；继而建立超短脉冲傅里叶相位调制与三维光场映射关系，提出超短脉冲高效相位逆重建算法，突破传统三维光场生成的迭代算法速度和精度局限；最后探索环路中神经细胞功能联系，提出多尺度聚合优化算法，搭建基于计算照明的光场显微成像系统平台，最终实现鲜活鼠脑切片、视网膜组织等生命科学功能的动态神经环路的高分辨率重建，为人工智能提供结构和功能指引。

High-resolution reconstruction of the neural functional circuit is a heated issue in brain science research. Limited to the low data throughput, current microscopic techniques cannot reconstruct large-scale functional circuits at high resolution. This proposal aims to develop light-field microscopy based on computational illumination. In order to overcome the inherent contradiction between speed and signal-to-noise ratio, we construct a structured-illuminated light-field imaging model based on computational photography, and propose an intelligent evaluation mechanism, revealing the mechanism of optimization strategy. Next, we establish the relationship between ultrashort pulse phases in the Fourier domain phase and its intensity distribution in the space domain and propose an efficient inverse reconstruction algorithm to break through the speed and accuracy limitations in conventional iterative algorithms. Finally, we explore the neural circuit connections, propose a multi-scale optimization algorithm, and build a light-field microscopic platform based on computational illumination, to realize functional imaging on biological specimens, such as fresh brain slices and retinal tissue. High-resolution reconstruction of dynamic neural circuits will provide insights into the structural and functional of the brain, which facilitate the development of artificial intelligence.