The high-power mode-programmable orbital angular momentum (OAM) beam has attracted significant attention in a wide range of applications, such as long-distance optical communication, nonlinear frequency conversion, and beam shaping. Coherent beam combining (CBC) of an optical phased array (OPA) can offer a promising solution for both generating the high-power OAM beam and rapidly switching the OAM modes. However, achieving real-time phase noise locking and formation of desired phase structures in a high-power CBC system faces significant challenges. Here, an internal phase-sensing technique was utilized to generate the high-power OAM beam, which effectively mitigated thermal effects and eliminated the need for large optical devices. An OPA with six elements was employed for experimental demonstration. The first effective generation of over 1.5 kW mode-programmable OAM beam in a continuous-wave domain was presented. Moreover, the results demonstrated that the generated OAM beam could be modulated with multiple dimensions. The topological charge can be switched in real time from -1 to -2. Notably, this OAM beam emitter could function as an OAM beam copier by easily transforming a single OAM beam into an OAM beam array. More importantly, a comprehensive analysis was conducted on power scaling, mode switching speed, and expansion of OAM modes. Additionally, the system’s compact design enabled it to function as a packageable OAM beam emitter. Owing to the advantages of having high power and programmable modes with multiple dimension modulation in phase structures and intensity distribution, this work can pave the way for producing high-power structured light beams and advancing their applications.

提出了一种针对病理切片图像的端到端语义分割方法--边缘感知网络（BPNet），以提高病理图像分割精度。BPNet网络首先在解码器阶段增加边缘感知模块，改善网络对于病理图像边缘的特征信息提取能力。然后，采用自适应通道注意力模块弥补不同层次特征间的语义差距，进一步加强网络的特征聚合能力。在此基础上，设计了一种基于结构和边缘的联合损失函数，以实现最佳的病理图像分割结果。在GlaS和MoNuSeg两个公开病理数据集上的分割实验结果表明，所提方法的Dice系数得分在两个数据集上分别达到92.21%和81.18%，有效提升了病理图像的分割精度。