Research Of Paper Diseases Detection Algorithm Based On Machine Vision

In actual production process, subject to a number of external and internal factors,disadvantages of paper outlook such as dirt,dot,holes,lines and winkles will showappearance. Presently, quantities of enterprises still take manual procedure which hasbeen existed long before to detect and allocate these paper diseases, which has shown itsshortage on lay off the production efficiency and high probability of miss detection andlow detection accuracy. Therefore, basing on machine visual, the paper diseasesdetection system has shown its capacity on not only better realizing the visualization,granular and automation in paper diseases detection but reducing production cost andraising detection efficiency, which has in hance has promoted enterprises’ profit.This passage has analyzed background, developing process and relativeresearching of online paper diseases detection system and calculation method basing onmachine visual. And detaliledly described the paper diseases distributed detectionsystem based on embedded processor.This essay also has briefly introduced the photograph processing technologies suchas image denoising and separating and takes some of their calculation method intoprogramming realization. What is more, this passage also has studied the detectioncalculating method of complicated paper diseases. As complicated paper diseasesmainly appear as line features in photographs, line detection algorithm is frequentlyapplied into this kind of paper diseases, among which the most usual one in Houghtransformation. Although Hough transformation has strong robustness and high capacityof anti-disturbance when applied into detecting complicated paper diseases, its largecalculation has make the instantaneity of production difficult to meet. Therefore, in thisessay, I tried to make out a way out of this, by applying transformation of probability ofHough. It is to process fitting a straight line by way of Hough transformation, which hastaken randomly extraction when choosing fitting points. What is more, it adds thecalculator. It output a line where randomly extracted points are satisfying the presetThreshold requirements. In this algorithm, Points of space mapping and straight linedetection alternates. Points are randomly extracted in point set of corresponding space.By which, less calculation can be realized and the same as the shortening of algorithmprocessing period.Moreover, In order to mature paper diseases detection algorithm, this essay also has worked on a survey into the application of ultrasonic wavelets in paper diseasedetection, which, however, is still in preliminary stage which needs more maturing inmany areas.