Research On Paper Counting System Based On Image Recognition

At present, there is a lot of paper to be counted in the production of paper industry or printing industry. The paper count is important for printing industry. Traditional mechanical paper counter had some inherent disadvantages, such as: the size of the paper material requirements, counting process noise, speed is limited, paper wear, counting accuracy is not good. These bring much inconvenience to the management of production. Therefore, the significance of developing a highly effective non-contact paper counter is significant.In the domestic and foreign markets there are few the paper non-contact automatic counting products. The paper has carried out the research on the image collection and the paper counting algorithm for the paper counting system, after analyzing the existing paper counting techniques and algorithms. The main contributions of this dissertation are as follows:(1) By comparing the present paper counting method, to determine the counting system use visual technology. The overall scheme of the system is proposed, and the hardware and software platform based on visual counts is constructed.(2) According to the proposed overall system scheme, the hardware and software platform of visual inspection is constructed. Through the experiment, the advantages and disadvantages of various platforms are compared, and the design scheme of the non-contact counting system is determined.(3) According to the content of twain protocol, communication between PC and image acquisition equipment is realized. Through the interface, the motion control of linear CCD equipment and the setting of the acquisition parameters are realized.(4) The texture detection and counting paper is refined as the preprocessing and texture detection, aiming at the important steps, the different algorithm is adopted to verify the effect of the process. By this experiment, the reliability and validity of the method of the paper counting system is verified. Further, the count algorithm of multi class stack texture is studied. A texture segmentation algorithm based on FCM and SVM is proposed. The algorithm avoids the tedious of the traditional SVM artificial selection sample, and has the independent choice sample, completes the training, realizes the more class image segmentation.This paper designs and develops the paper counting system which realizes the function of non-contact counting of the paper.The counting system has the characteristics of high efficiency and stability for the ordinary printing paper.The design of the hardware and software of the system takes into account the promotion of the practical application of the project.