应用轮廓链序列的符号M阵列结构光解码  

Decoding of symbol M-array using contour chain sequence for structured light

作　　者：贾小军[1] 孙亚新 方玫[1,2] 曾丹 

机构地区：[1]嘉兴学院数理与信息工程学院,浙江嘉兴314001 [2]上海大学特种光纤与光接入网重点实验室,上海200072

出　　处：《光电子．激光》2017年第10期1108-1117,共10页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61572307)资助项目

摘　　要：利用符号M阵列形成单帧模式作为投射图像,提出对符号编码结构光进行解码的方案。首先,建立符号的游历算法,生成符号轮廓链序列,根据链角实现符号轮廓的波形特征表述;其次,采用基于符号矢量方向及内部有向链角,进行符号定位及识别;利用符号矢量方向可以将符号分成4类,每类2个符号,根据符号内部链角的波形特征,区分每类中的两个符号;最后,利用子窗体全局唯一性创建子窗体定位条件,实现子窗体匹配,并根据投票方式进行符号对应,提取符号实现关键点三维信息。实验结果表明,提出的解码方法在应用于不同的检测对象时,均能有效地解码符号,实现三维重建,具有较强的鲁棒性。Using symbol M-array to produce a one-shot pattern as a projection image, a decoding scheme is proposed to solve the symbols encoding for structured light. Firstly,a symbol contour chain sequence is generated by adopting symbol travel algorithm. According to the chain angle,waveform characteristics of symbol contour can be expresse＆ Secondly, based on the symbol＇s vector direction and inner chain angle, the location and recognition of symbol are accomplished. All symbols can be divided into four categories by using symbol vector direction, and each category includes two symbols. According to the waveform characteristics of inner chain angle of symbols,the two symbols in each category can he distinguished. Fi- nally,using the uniqueness of sub-window in global M-array to generate sub-window positioning condi- tions, the corresponding sub-windows can be obtained. The corresponding symbols using the voting algo- rithm can help to extract three-dimensional information of key points of symbol. The experimental results show that the proposed decoding method can effectively decode symbols, realize the three dimensional reconstruction when it is applied to different detection objects, and has strong robustness.

关 键 词：轮廓链序列 符号M阵列 结构光 解码 三维重建 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]