Deblurring Barcodes Images Based On L₀-Regularization

With the development of digital technology and the popularity of mobile devices,barcodes has been applied to all aspects of our lives.However,for most reality scenarios,barcode images usually have different degrees of blur due to the factor of circumstances and image acquisition equipment,which seriously affects the recognition efficiency of the barcode recognition system.Relying solely on the anti-blur characteristics of the barcode itself,it is difficult to satisfy the needs of current bar code applications in a variety of circumstances.In order to ensure the recognition accuracy rate and improve the robustness in various circumstances,a general and efficient blur removal algorithm is required to blurred barcode before recognizing them.In this paper,by analyzing the gradient and gray histogram of blurred barcode image-and clear barcode image,we present a barcode image deblurring algorithm based on L0 regularization according to the sparseness of clear barcode gradient.Our algorithm is divided into three parts:image estimation,kernel estimation and edge enhancement processing.These three parts iterate alternately and gradually obtain clear images.In the part of image estimation,the L0 regularization constraint on the image gradient is adopted,and the resolution of the clear barcode image is completed by introducing the auxiliary variable transformation energy function;In the part of kernel estimation,the L2 regularization term of kernel is added,and the conjugate gradient descent method is used to solve the kernel;The edge enhancement part uses the bilateral filter and the shock filter to enhance the edge of the blurred image,which improves the accuracy of the kernel estimation.In addition,considering the coding characteristics of ID barcodes,we converted 1D barcode images into ID digital signals to speed up our algorithm.The experimental results show that our deblurring algorithm works well on barcode images and the recognition rate of barcode recognition system has been greatly improved compared to other algorithms.