In order to solve the problem of bar code failure because of inadequate protection, pollution and other factors in the process of production, presented a new multi-information fusion technology used of residual bar code information, shape, texture and the weight information to identify the tool information in the database and recovery the tool identification. Experimental results show that the proposed scheme has better performance in accuracy, robustness and efficiency, can fully satisfy the actual requirement in production.

A mosaic method of based on rotational scan sequence cylindrical barcode which is suitable for metal parts is put forward, which combined the hardware and the software. Firstly, extract the feature points of each serial image based on scale-interaction of Marr wavelets. Then, get the optimal match of feature points with improved image local entropy. Lastly, complete the natural splice of the defect 2D bar code images through the algorithm of three interpolation and multi-resolution spline. The experimental results show that the algorithm could extract the feature points of consistent relative position and quantity after the processes of rotation, brightness, blur and nose, at the same time, assure the splicing efficiency and measurement precision of image mosaic. The method can also better finish the defect bar code recovery caused by curvature deformation.

基于配套光源的优化,提出了基于随机共振算法的图像溢出信息恢复方法.该方法利用噪声激发饱和系统的随机共振,扩大输出图像的动态范围,挽回高光溢出的信息.分析了不同类型和不同强度的噪声对恢复信息可分辨性的影响.仿真结果表明,不同类型的噪声都有对应的特定强度,使系统输出信息的可分辨性达到极值.最后,通过局部过曝的实例,验证了基于随机共振的图像信息恢复方法能达到抑制高光和提高图像对比度的目的.

To resolve the problem of the distortion of the 2-D barcode on the cylindrical surface,surface distortion model is setted up to correct the barcodes based on the automatic reading system of the two-dimensional barcode in the surface of tools.Moreover,considering some barcodes cannot display in one image,the missing barcode image based on image mosaic is recovered.First,the overlapped areas with related phase as the range of the feature collecting is estimated.Second,the SIFT points of related areas are collectted and the SIFT features with the algorithm of approximate nearest neighbor are matched.Third,the original images are revovered by using the step-in-step-out method.The experiments show that the algorithm improves the efficiency of the 2-D barcode reading.