Measurement of blood flow velocity is key to understanding physiology and pathology in vivo. While most measurements are performed at the middle of the blood vessel, little research has been done on characterizing the instantaneous blood flow velocity distribution. This is mainly due to the lack of measurement technology with high spatial and temporal resolution. Here, we tackle this problem with our recently developed dual-wavelength line-scan third-harmonic generation (THG) imaging technology. Simultaneous acquisition of dual-wavelength THG line-scanning signals enables measurement of blood flow velocities at two radially symmetric positions in both venules and arterioles in mouse brain in vivo. Our results clearly show that the instantaneous blood flow velocity is not symmetric under general conditions.

针对传统城墙病害检测采用人工勘测方法，检测效率较低且易受到主观因素干扰等问题，提出一种基于卷积神经网络的城墙多光谱成像病害无损检测方法，利用最小噪声分离方法对城墙多光谱成像数据进行预处理，降低数据维度的同时保留原始数据特征，减少数据噪声；为解决城墙不同病害类型的像素混杂多样造成分类准确率较低的问题，利用卷积操作对城墙病害进行特征提取，保留最重要的特征并去除无关特征，稀疏网络模型；通过全连接层对提取到的特征进行整合梳理和分类，并加入两次dropout防止过拟合问题的出现。最后在城墙多光谱数据集上，使用训练后的卷积神经网络分类模型对城墙病害进行像素级分类检测，并将预测结果进行可视化展示。实验结果表明：总体精度和Kappa系数分别为93.28%和0.91，表明所提方法是有效的，该方法对提高城墙病害检测准确率、掌握城墙病害分布具有重要意义。

利用分数涡旋光束进行信息传输，可以大大提高通信系统的容量。但在接收端准确测定分数涡旋光束的模态，尤其在信道中存在湍流的情况下，存在一定的困难。本文提出了改进的残差网络（I-ResNet）以提高模态检测的正确率。实验结果表明，本文构建的网络能够准确地识别光束模态，且具有较好的泛化性。在传输距离为1500 m、弱湍流（Cn2=10-16m-2/3）、模态分辨率Δl≥0.05时，准确率可以达到100%；强湍流（Cn2=10-14m-2/3）、Δl=0.15时，准确率可以达到96.5%。随着湍流强度或传输距离的增加，正确识别率下降。这些结果对自由空间光通信系统设计具有一定的指导意义。

The skin is heterogeneous and exerts strong scattering and aberration onto excitation light in multiphoton microscopy (MPM). Shifting to longer excitation wavelengths may help reduce skin scattering and aberration, potentially enabling larger imaging depths. However, previous demonstrations of skin MPM employ excitation wavelengths only up to the 1700nm window, leaving an open question as to whether longer excitation wavelengths are suitable for deep-skin MPM. Here, in order to explore the longer-wavelength territory, first, we demonstrate characterization of the broadband transmittance of excised mouse skin, revealing a high transmittance window at 2200nm. Then, we demonstrate third-harmonic generation (THG) imaging in mouse skin in vivo excited at this window. With 9mW optical power on the skin surface operating at 1MHz repetition rate, we can get THG signals of 250μm below the skin surface. Comparative THG imaging excited at the 1700nm window shows that as imaging depth increases, THG signals decay even faster than those excited at 2200nm. Our results thus uncover the 2200nm window as a new, promising excitation window potential for deep-skin MPM.

汉长安城遗址出土骨签数量庞大，基于数字图像技术的骨签尺寸自动测量方法能够提高工作效率。针对骨签纹理繁多、边缘对比度低导致测量精度不高的问题，提出一种基于分区域自适应阈值Zernike矩的骨签图像尺寸测量方法。首先，利用Canny算子进行像素级定位；然后，对称设定边缘有效分区域，以欧氏距离为比例系数计算各个分区域像素点与中心点的加权灰度值，作为各区域Zernike矩提取边缘的判定阈值，提取骨签亚像素级边缘；其次，引入边缘轮廓判别条件剔除无效纹理，获取骨签图像精确轮廓；最后，使用最小外接矩算法计算骨签不规则轮廓，结合相机标定完成骨签尺寸测量。实验结果表明，与其他方法相比，该方法测量的骨签长宽的均方根误差分别降低1.3 mm和1.2 mm，平均相对误差分别降低3%和6.3%，平均绝对误差分别降低1.23 mm和1.08 mm，能够有效提高复杂边缘轮廓的骨签尺寸测量精度。