The Authenticity Identification Of Anti-counterfeiting Code Based On Mobile Terminal And Image Micro-features

The phenomenon of counterfeit and shoddy products seriously harms the interests and reputation of consumers and businesses,and disrupts the operation of the entire market economy.As an effective way to solve this problem,anti-counterfeiting technology has always been concerned by everyone.The current anti-counterfeiting technology performs well,but there are still many problems such as short anticounterfeiting period,the incapability of pre-sale anti-counterfeit and the difficulty for consumers to distinguish.With the rapid development of mobile terminal equipment and applications,combining anti-counterfeiting technology with digital image processing can achieve fast and intelligent anti-counterfeiting function.Therefore,in this paper,the texture feature introduced by the random diffusion of ink on paper is combined with the QR code to construct the anti-counterfeiting code,which is used as a unique identification of the commodity.This paper has proposed and solved practical problems based on the instability of the mobile terminal,and has verified the authenticity of the anti-counterfeiting code by studying the fast and effective image processing authentication algorithm.This paper researches on extraction and quality evaluation of the anticounterfeiting code.According to the characteristic that the anti-counterfeiting code is a fine texture,this paper use combination morphology to bind it into a large white block that can be distinguished form the background.The corner points obtained by fitting the polygon are screened through the smallest circumscribed rectangular corner points to obtain the final four corner points of the anti-counterfeiting code.The extraction result is obtained by normalizing the target of the perspective transformation.The image quality evaluation of anti-counterfeiting codes includes three situations:reflective,defocused blur and motion blur.In this paper,the reflective images are detected according to the shape of the end of the histogram.In view of the problem that it is difficult to formulate a unified standard due to different defocusing environment and mobile phone hardware,this paper argues that the position of the double peak of the histogram represents the brightness and contrast of the image to a certain extent.By dividing the histogram into three segments for linear adjustment,the adjusted image is convolved with the Laplace operator,and finally the image information entropy is calculated and used to determine whether there is defocus blur.Aiming at motion blur,according to gradient symmetry,the anti-counterfeit codes are convolved with the sobel operators in four directions to calculate the gray-scale average ratio of the convolved image in the symmetric direction,which can detect motion blur very easily and quickly.The texture thickness in the anti-counterfeiting area varies,and obviously the narrower the area is,the more difficult it is to be imitated.Therefore,this paper proposes an authentication algorithm based on key points.First,the skeleton is extracted from the foreground area of the binarized image,and the bone width is calculated according to the distance transformation,and the threshold is set to obtain the key points.By calculating Surf feature descriptors of key points and using Euclidean distance and K nearest neighbor algorithm,the key points are matched and the wrong points pair are initially removed.The matching point pairs have similar position information,so the erroneous matching point pairs can be further eliminated.Finally,by calculating the number of correct matching points and comparing the algorithm,the experiments prove the superiority of the algorithm in this paper,with a precision rate of 98% and the recall rate of 91%.For the texture study of key areas of anti-counterfeiting codes,this paper constructs key areas based on key points and further divides them according to the anticounterfeiting capabilities of key areas.First,the statistical characteristics of the gray level co-occurrence matrix are normalized to between 0 and 1 to facilitate the similarity measurement.Owing to the fact that the gray level co-occurrence matrix lacks texture information of local details,this paper proposes that the edge region can be determined based on the completed local binary amplitude information,so as to calculate the edgeregion completed local binary feature histogram.Fuse it with the normalized gray level co-occurrence matrix statistical feature to calculate the Euclidean distance between sample image’s and the target image’s feature vectors.Finally,the appropriate gray level co-occurrence matrix parameters are selected through experiments,and it is verified that the proposed algorithm is more accurate and stable than other texture algorithms in authenticity identification,with a precision rate of 98% and a recall rate of 94%.