【摘要】 运用基于动态图像序列的运动目标跟踪技术, 结合图像边缘提取理论中的检测算子模板, 提出了一种快速提取啤酒瓶凸性字符的方法: 首先采用单个面阵CCD得到啤酒瓶字符的运动图像序列, 然后利用背景在运动图像序列中的镜头突变性和出现的必然性, 检测视频流中的背景帧, 最后采用基于边缘检测的背景配准方法, 得到运动图像和校正背景图像的差分图像, 恢复出真正的运动物体 (字符). 实验结果表明, 该方法比现有的字符图像提取方法具有更好的抗噪性和实用性, 加快了字符提取的速度, 满足了在线检测的要求.更多还原

【Abstract】 A fast extraction method of beer bottle rasied character is introduced in this paper. It is the technique combination of moving target tracking based on dynamic image sequence with edge detection algorithm for speeding up the processing of character extraction to meet the demand of on-line detection in industry.First dynamic image sequence is obtained with an area CCD camera, and the background frames are detected based on the coherence between the background and character image from the video and inevitability of background frames emergence. Finally, the moving character is extracted by differentiation with dynamic image and registration background image. Experimental results indicate that the method is more efficient than the present method of character extraction.更多还原