采用多角度成像技术的纺织品三维轮廓重建算法  

Algorithm for three-dimensional reconstruction of textile contour based on multi-directional imaging

作　　者：孟想 辛斌杰[1] 李佳平 MENG Xiang;XIN Binjie;LI Jiaping(School of Fashion Technology, Shanghai University of Engineering and Science, Shanghai 201620, China)

机构地区：[1]上海工程技术大学服装学院,上海201620

出　　处：《纺织学报》2018年第4期144-150,共7页Journal of Textile Research

基　　金：国家自然科学基金项目(F010402)

摘　　要：为更全面地对织物表面信息进行表征,方便对织物进行结构分析、外观评估、质量检测和逆向工程等测试,设计了1套可实现数字化图像采集、相机校准和三维重建的纺织品三维轮廓重建方法及系统。自行搭建了1套针对纺织品多角度图像采集平台,同时根据成像参数优化、算法优化深入探讨二者对三维重建效果的影响,逐步优化实验流程和拍摄环境,以提高纺织品三维重建系统的精度。结果表明,在现有硬件的基础条件下,该平台可完成低密度针织物三维轮廓的高精度重建,重建所得模型携带织物表面大量的纹理信息、空间位置关系及颜色信息,特征匹配的特征点均超过1 000个。In order to more fully characterize the surface information of the fabric to facilitate the subsequent fabric structure analysis,appearance evaluation,quality inspection and reverse engineering and other applications,a three-dimensional contour reconstruction method and system were developed for realizing the digital image acquisition,camera calibration and three-dimensional reconstruction.A multiangle image acquisition platform for textiles was set,the influence of the imaging parameters optimization and the algorithm optimization on the three-dimensional reconstruction effect was investigated,and the experimental process and shooting environment was gradually optimized to improve the system accuracy of the three-dimensional reconstruction of textiles.The results show that based on existing hardware,the platform can complete high-precision reconstruction of the three-dimensional contour of low density textiles,and the reconstructed model can carry a large number of texture information,spatial position relation and color information on the surface of the fabric,and the feature point number is above 1 000.

关 键 词：纺织品检测 相机标定 特征点匹配 点云融合 三维重建 

分 类 号：TS101.8[轻工技术与工程—纺织工程]