多光谱颜色成像 / Multispectral Color Imaging

多光谱相机通过增加光谱波段的成像通道数量，与普通RGB相机相比，能获得所拍摄场景更多的光谱信息，因而在很多领域具有重要应用。实验室历时多年，搭建了基于积分球与滤光片轮的多光谱成像颜色测量（ICM）系统，处于国际领先地位，在纺织及成衣工业成功应用，获2013年瑞士日内瓦国际发明展览会金奖。

A multispectral camera acquires spectral images by splitting the visible spectrum into many bands. It is of wide applications including textile color measurement, object recognition, medical imaging, remote sensing, etc. We have successfully developed the world-leading multispectral imaging color measurement (ICM) system for the textile and apparel industry, under the support of Hong Kong Innovation and Technology Fund. The system won the Gold Medal on the 41st International Exhibition of Inventions of Geneva, Switzerland, April 2013.

颜色是人眼感知外部世界的主要信息，在工业及消费电子中得到越来越多的重视。实验室开展成像设备（手机、相机、扫描仪、显示器、打印机等）的颜色特性化、视觉色差建模等方面的研究，并通过与多光谱成像技术相结合，为相关工业界提供全面的颜色测量和管理解决方案。

As a major perception of the world, color information is important to the related industrial applications and consumer electronics. We conduct R & D work on imaging device (i.e., mobile phone, digital cameras, scanners, displays, and color printers) characterization, color communication, and visual color difference modeling. Based on multispectral imaging and color technology, we work toward the total color measurement and management solutions for the industry.

图像处理 / Image Processing

面向行业应用，开展织物图像分割、重着色等方面的算法。针对多光谱图像特性，研究包括图像配准、去噪、去模糊、修复（inpainting）、超分辨率、融合等在内的各类图像恢复方法，并开展多光谱视频重建研究。上述研究可广泛用于普通RGB图像与多模态图像，以及移动设备（例如手机）图像的恢复与增强。

We work on the fabric image segmentation and re-coloring algorithms for the textile industry. Based on the characteristics of multispectral image, we work on image restoration techniques including image registration, denoising, deblurring, inpainting, super-resolution, and image fusion. We also work on multispectral video reconstruction methods. The developed techniques are applicable to traditional RGB images and multimodal images, as well as the photo captured by smart phones.

人体红外热图像能很好地反映人体的健康状况。实验室开展红外热图像处理与分析，包括图像拼接与分割、人体温度建模、异常检测、红外图像与三维模型融合等方面内容，在此基础上研发人体健康辅助诊断系统。

Infrared thermal image has a high correlation to human’s health status. We work on infrared image processing, including image stitching, human temperature modeling, abnormal detection, infrared image / 3D fusion, etc. The aim is to develop a computer aided health diagnosis system based on the research work.

计算机视觉 / Computer Vision

利用计算机视觉开展三维立体重建方面的研究与应用。光度立体视觉方面，利用数学优化与机器学习，从多幅光影图像中可靠恢复非郎伯表面（具有高光、阴影等复杂反射现象）的形状信息。通过与物体表面反射特性的结合，可用于虚拟现实中物体外观的高度真实感渲染。

We work on the 3D reconstruction of object surfaces in computer vision. In photometric stereo, we estimate the surface normal of various surfaces from multiple images, by using mathematical optimization and machine learning. When combined with surface reflectance, the technique is applicable to the realistic rendering of virtual scenes.

单目/双目立体方面，研究三维场景感知与重建，特别是将相关技术应用于无人机视觉避障、路径规划、导航、地图构建等方面，增强无人系统智能与自主性。在计算机视觉方面还开展目标识别与跟踪等方面的技术研究。

In monocular / binocular stereo, we work on the 3D environment perception and reconstruction. Specially, the techniques are applied to the obstacle avoidance, path planning, navigation, and SLAM of unmanned air vehicle (UAV). We also work in the fields such as defect detection, object recognition, and visual tracking.

机器学习 / Machine Learning

开展各类机器学习相关算法（例如深度学习）研究，将其与上述图像处理、计算机视觉算法进行深度融合，促进相关技术进步与应用。

We work on machine learning (such as deep learning) and apply these techniques in the image processing and computer vision fields.

表面材质采集 / BTF & BRDF Acquisition

客观世界中的物体表面具有各类材质特性。实验室设计并搭建了双向反射率函数（BRDF）与双向纹理函数（BTF）数据采集系统，并在此基础上开展了材质数据稀疏采集、超分辨率重建、光谱BTF数据重建等方面的研究。实验室将BTF数据用于共享，以促进相关领域的研究。

We developed an imaging system that captures the bidirectional reflectance distribution function (BRDF) and bidirectional texture function (BTF) data. We work on the sparse acquisition of BTF data, as well as the super-resolution and spectral reconstruction of BTF data. Our BRDF/BTF dataset are shared for research purpose. Please visit here for the dataset.