Laminated Paper Counting Algorithm Based On Compressive Sensing And Hough Transform

Laminated paper counting has widely used in paper manufacturer, packing, paperprinting and other industries. A fast and accurate laminated paper counting system canimprove the production efficiency in these industries. The accuracy of the count resultcalculated by traditional laminated paper counting machines cannot be ensured, which isaffected by various factors, such as the sensitivity of the measurement equipment, thesurroundings and the nature of the laminated paper etc. And the laminated paper countingmachines have other disadvantage, such as big floor occupation, high power consume,operating complexly, maintaining difficultly and bring errors easily etc. According to theproblem that conventional laminated paper counting machines have some unavoidableshortcomings, domestic scholars have tried to apply the technology of machine vision intothe laminated paper counting system. The laminated paper counting algorithm is the coreand difficulty of the laminated paper counting system based on machine vision.Result shows that the traditional laminated paper counting algorithm based on textureanalysis does not take advantage of the sparsity of the laminated paper image signal.Compressive sensing theory breaks the bottleneck of the traditional Nyquist samplingtheorem, it more fully uses the sparsity of signals, and greatly reduces the signal samplingand transmission costs. In this paper, a laminated paper counting algorithm based oncompressive sensing and Hough transform is proposed for solving the shortage of thetraditional laminated paper counting algorithm based on texture analysis. The main worksare as follows:Firstly, in this article, on the basis of the Hough transform theory, we created aover-complete dictionary based on straight lines. We proved that the over-completedictionary is irrelevant to some common random measurement matrixes and the laminatedpaper image signal can exhibit sparsity through sparse decomposing on the over-completedictionary. Therefore, we have proved the feasibility of applying the compressive sensingtheory to the laminated paper counting algorithm.Secondly, we establish a compressive sensing model of the laminated paper counting.On the basis of the traditional laminated paper counting algorithm based on textureanalysis, we propose a laminated paper counting algorithm based on compressive sensingand Hough transform. Finally, the MATLAB simulation is performed to demonstrate the feasibility of theproposed algorithm, and demonstrates that the proposed algorithm need less sample pointto accomplish the laminated paper counting task compared with the traditional laminatedpaper counting algorithm based on texture analysis while guarantee the accuracy of thecount result.