Image Filtering In The Application Of Bar Code Recognition

Two-dimensional code is no longer use the linear sweep method to identifica-ion its information, that identify its information must recognize the two-dimensi-onal code’s image, but in the process of image acquisition and preservation maybe impulse pepper and salt noise, result in the system does not recognize the barcode. In order to improve the rate of recognition, it is necessary to remove thenoise before decoding.This thesis is focus on remove the image’s noise. The QR code wholedecoding process generally includes graying, median filtering, image positioning,rotation, segmentation and decoding. This thesis main work is that：according tothe two-dimensional bar code has more dramatic features and other details,presented a new algorithm which is median filter base on minimum variance andmean filter.The algorithm first improved filtering template, select the minimum variancetemplate combined with the improvement of the mean filter for image denoising,Better protection in the filter bar code image noise, image detail and edgeinformation to enhance the recognition rate of the image; in-depth study ofimpulse noise based on the regional difference filtering algorithm based on theprinciple of regional the algorithm first regional difference between the filteredimage, and then after application of median filtering, filter out the moremoderate image of the gray-scale changes in the salt and pepper noise, andachieved good results; according to the median filter denoising principle, forbetter filtering algorithm output value close to the original pixel values of pixelsin this paper two in the median filtering algorithm for image filtering, comparedwith the median filtering algorithm improvements.meanwhile according to theprinciple of area different presented the median filtering combined with areadifferent filtering algorithms. The experimental results show that the newalgorithm can effectively remove the bar code image noise meanwhile canpreserve more edge and detail features, certain to enhance the recognition rate ofthe image. To demonstrate the performance of algorithm,we apply SNR and PSNR to doan objective evaluation of the algorithm,results show that the algorithm hasbetter performance.