Based On The Texture Of Paper Count

Paper-counting is very common in industrial production. Traditionally, it is carried out by counting machines. The shortage of this traditional technology is obvious, including costing expensively, taking up too much room and tending to be incorrect, so that it is especially necessary to improve it.From the image analysis'point of view, this article regards the stack-up paper margin as texture image, and makes research on adjusting the slant texture of stack-up paper, and also detects the gradient using frequency detection, projection features and Hough method. According to the massive relevant resources, it is advisable to use Gabor filter and one dimension one line method for the low quantity texture of paper image. Furthermore it analyzes the reliability of the one dimension one line method in the condition that main frequency point is caused by partial texture fault, and introduces the morphology analysis to analyze the texture joint and fracture due to uneven exposal. Then, it utilizes the improved Ostu method to detect the edge and segmentation. Finally, it detects the projection of Binarization image, and attains reliable results of paper-counting.This article is based on the achievements of former scholars. It analyzes the entire flow-line of the texture image and modifies the traditional analyzing methods of it. Moreover, it puts relevant theories of paper-counting into the practice and accesses the parameters of each implementing step with statistics methods.The algorithms in this article are realized on vc IDE or matlab platform, The set of processing is accurate in paper-counting, and strong in robustness, as a result it enhances the feasibility of paper-counting based on texture analysis.