Research On Paper Counting System Based On Image Processing

Paper and sheet counting is widely used in life and production, then mechanical flip-style paper counter as an improvement of production efficiency of the equipment used in the printing industry and office occasions. But the traditional mechanical paper counter also exist disadvantages such as easy to cause paper wear, high cost, counting process with noise and so on. Therefore, it is important to develop an efficient non-contact paper counting device based on image processing.There is no relevant mature based on the image processing of non contact type of paper counting devices in the domestic and foreign markets. Based on the above situation, by comparing and analyzing the existing paper counting techniques and algorithms, in this paper, the research of image acquisition and paper counting algorithm for paper counting system is developed. The main work is as follows:Through the comparison and analysis of the texture image acquisition method and counting method of the stack of paper edge at the present stage, the digital microscope is proposed as an image acquisition device, system scheme based on the combination of image mosaic and texture image processing, software and hardware platform is built based on image processing and counting.Moving digital microscope for texture image acquisition, there is no less than fifty percent overlapping area between collected data. A reference consisting of disorder points is added to the test paper to keep the registration between the image feature points in the overlapping area more accurate.The image mosaic method based on SURF(Speeded Up Robust Feature) algorithm is realized. Firstly, SURF algorithm is used to detect and describe the feature points. Secondly, the nearest neighbor method is used to match feature points, and RANSAC(Random Sampling Consensus) algorithm is used to eliminate the error matching. And then, the gradual approach to the image fusion method is used to make the border of the image mosaic transition naturally, and mosaic with the next texture image acquired with reference as the way of image to the mosaic image. Finally, a full picture of the stack of paper to be measured is got.The mosaic of paper texture image processing count is realized. Firstly, making the target area the of the paper texture image preprocessed. Image preprocessing includes image resizing, image enhancement and tilt correction. Then there are three methods used to process the image. The three methods are the way based on judging the black stripes, the way based on morphological filtering and the way based on gabor filter. Finally, determining the counting method based on gabor filter. Aiming at the important steps in image texture processing such as image enhancement and thresholding, the results are verified by different algorithms.In this paper, the paper counting system based on the image processing has achieved the paper non-contact counting function, it has a good ability to develop the market.