Abstract: Direct Part Marking (DPM) technology is an important means to achieve product identification, and 2D bar code technology is one of the keys. DPM logo of this paper made metal as a background and marked 2D with a laser above it. Compared to image recognition in 2D bar code based on print background, the image recognition in 2D bar code based on metal background is more complex. This paper improved the traditional identification methods, integrated use of the largest connected components extraction, the improved Hough transform, the code division based on maximum degree of matching and the gray image based on non-destructive method of information extraction to achieve the rough positioning, precise positioning, calibration, bar code segmentation and data extraction of the bar code images. The results show that this program has a strong anti-interference for the wear, light pollution, distortion and uneven illumination 2D barcode on the metal. Finally, reliable results were obtained.

Key words: two-dimension barcode, Direct Part Marking (DPM), metal background, image positioning, anti-interference

摘要： 直接零件标识(DPM)技术是实现产品标识的重要手段，二维条码技术是其关键技术之一。DPM标识是以金属作为背景用激光在其上面打上二维条码，与印刷品上的二维条码识读相比，基于金属背景下的二维条码识读更为复杂。改进了传统的识别方法，综合运用最大连通域提取、改进的霍夫变换定位边界、基于最大匹配度的网格化分和基于灰度图像的无损信息提取等方法来实现条码图像的粗定位、精定位、校正、条码分割和数据提取。实验表明，本方案对于磨损、轻度污染、畸变及光照不均的金属上的二维条码图像的识读具有较强的抗干扰性，并取得可靠的识读结果。

关键词: 二维条码, DPM, 金属背景, 图像定位, 抗干扰性