基于视觉传感的服装样片位图矢量化  

Vectorization of clothing pattern bitmap based on vision sensor

作　　者：王琪[1] 冯远静[1] 董辉[1] 刘连杰[1] 俞立[1] 

机构地区：[1]浙江工业大学信息工程学院,浙江杭州310032

出　　处：《浙江工业大学学报》2010年第3期273-277,共5页Journal of Zhejiang University of Technology

基　　金：浙江省自然科学基金资助项目(Y106660);浙江省科技厅重大专项基金资助项目(2007C11090)

摘　　要：为了提高服装CAD之放缩部分基本服装纸样的输入效率,本文提出了一种基于位图矢量化的新型服装样片矢量图输入方法.该方法首先通过视觉传感器采集多个服装样片位图,并输入到计算机内,经图像分割后得到样片轮廓的数字序列表示,然后再通过间接提取轮廓特征点并对特征点之间的曲线进行分段递归拟合,最终生成服装CAD系统能够识别编辑的矢量图文件.实验结果表明,该矢量化方法能通过终止条件很好地控制拟合的精度,满足了服装CAD系统矢量化的精度要求,并且与传统的数字化仪录入方式相比,效率高,成本低,易于操作.A novel input method for vector graph of new clothes patterns based on vectorization of bitmap is proposed to increase the input efficiency of the basic patterns in the zoom part of apparel CAD. The pattern images are collected by a vision sensor and then are transmitted to a computer. The pattern edge represented by the digital sequence is obtained after the image is segmented. The feature points are extracted indirectly and curves between the feature points are recursively sub-fitted. Finally, the vector graph which can be identified and edited by apparel CAD is generated. The result demonstrates that the vectorization algorithm can well control fitting precision through ending condition. It can fulfill the precision demand of apparel CAD. Comparing to traditional inputting method on digital equipment, it has the advantages such as high efficiency, low cost and good maneuverability.

关 键 词：视觉传感器 矢量化 曲线拟合 图像分割 服装CAD 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]