采用GPU并行计算与图像匹配的工件条码识别算法  

Workpiece Barcode Identification Algorithm Using GPU Parallel Computing and Image Matching

作　　者：董夙慧[1,2] 孙中廷[1,2] 徐永刚[1] 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221008 [2]江苏联合职业技术学院徐州财经分院,江苏徐州221011

出　　处：《华侨大学学报（自然科学版）》2017年第1期80-85,共6页Journal of Huaqiao University(Natural Science)

基　　金：江苏省自然科学基金资助项目(BD2011129)

摘　　要：设计基于图形处理器(GPU)并行计算与图像匹配的条码识别算法.首先,设计基于归一化协方差的图像匹配算子,定位每个条码的位置.然后,根据条码面积、周长、形状因子,确定条码种类.最后,采用GPU并行计算方式完成条码解析.实验数据显示:与当前条码识别技术相比,所提算法能够同步识别多种条码,且在面对光照强度较暗或过高时,该算法具有更高的稳定性与抗干扰性,其仍具有更高的识别精度与效率.The workpiece barcode synchronization identification algorithm based on graphics processing unit （GPU） parallel computing and image matching was proposed in this paper. Firstly, the position of each bar-code was located by designing image matching operator based on normalized covariance. Then the type of bar-code was determined according to the barcode area, perimeter,and shape factor. Finally, the analysis of the barcode was finished by GPU parallel computing method. Experimental data shows that： compared with the current barcode recognition technique, this algorithm can identify a variety of barcode, this algorithm had a higher stability under the light intensity is dark or too high, and it had a higher recognition accuracy and efficiency.

关 键 词：条码识别 图形处理器 图像匹配 并行计算 归-化协方差 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]