Method And Implementation Of Document Vision And Seal Defect Detection Based On Machine Vision

With the development of machine vision technology and automation technology,machine vision technology has been widely used to various defects detection.Using the knowledge of machine vision technology and image processing,this paper detects the defect of certificate printing and seal for online detection system,which mainly includes the following contents:(1)Proposing a fast calculation method of document tilt angle based on edge points rotation.This method fistly down-samples the document image,and uses ShenJun operator to extract the edges of the downsampled image.Then,a 'Rough' and 'Accurate' tilt angle calculation method is adopted to rorate the edge points.After that,calculate the variance of the projection curve.Since the rotation angle corresponding to the maximum variance is most likely the inclination angle of the document image,the inclination angle is calculated accordingly.Experimental result shows that this method can quickly calculate the tilt angle of the document images.(2)Proposing a method of detecting the lack of ink in documents based on histogram peak analysis.The method first removes the influence of the grayscale transition region around the character stroke.Then,using sliding window to traverse the image,calculate the image histogram in the sliding window and judge whether the window is lack of ink according to the distribution of the peak of the histogram.Finally,count the number of sliding windows which is lack of ink by column.Judges image to be lack of ink if there are more than T sliding windows which are judged to be lack of ink.The experimental results show that this method can make an accurate judgment on the image of ink shortage.(3)Presenting a method to detect the defect of seal based on residual image.This method firstly locates the seal,then registers the seal to be detected with the standard seal,aftter calculating the residual image of the seal,the morphological filter is applied to the residual image.Finally,measure the perimeter and area of the target in the residual image,according to the circumference and the size of the one target to determine whether there are defects int the seal.In the seal location,this paper locates the position of the seal according to the symmetry features of the circular seal,and this method can quickly locate the seal location with high robustness.In the seal registration,this paper uses the ORB algorithm to extract the feature point of the seal to be measured and the standard seal,with matching the feature points,and then uses the voting strategy to calculate the deflection angle between the seal to be measured and the standard seal.The method can effectively calculate the deflection angle between the seal to be measured and the standard seal to achieve registration.(4)Realizing the lack of ink detection based on gray histogram and SVM.According to the characteristics of the text image,this paper selects the gray histogram feature,and uses the SVM classifier to detect the image in the sliding window.The experimental results show that the method can effectively judge whether it is lack of ink or not in the sliding window.(5)Introducing a certificate defect detection system for measurement verification test center in the end of this paper,which mainly introduces its hardware system and software system.