Light field high-resolution reconstruction method based on radiation field representation and neural field parameterization

The invention discloses a light field high-resolution reconstruction method based on radiation field representation and neural field parameterization, and the method comprises the steps: 1, constructing a dual-plane light field used for representing a scene, and representing the dual-plane light field as a radiation field; 2, performing neural field parameterization on the radiation field by using a multi-layer sensor, and establishing a neural radiation field of a biplanar light field; step 3, using high-frequency mapping coding and voxel rendering to optimize and train a neural radiation field of the biplanar light field; and 4, performing high-resolution reconstruction on the light field data by using the optimized neural radiation field of the biplanar light field. According to the invention, angle super-resolution can be carried out on a light field scene with complex and unfavorable conditions. 本发明公开了一种基于辐射场表示和神经场参数化的光场高分辨重建方法，其包括：步骤1，构建用于表示场景的双平面光场，再将双平面光场表示为辐射场；步骤2，利用多层感知器对辐射场进行神经场参数化，建立双平面光场的神经辐射场；步骤