The Pretreatment Research Of Shopping Receipts Image Recognition

Shopping receipts, as the supermarket store receipts, are sales retained credentials to theuser by Shopping malls or other commercial institutions when consumers are shopping. Now,in addition to the few handwritten receipts, most of them are from the printing of the cashregister. Store receipts is a true record of the economy behavior of the consumer, through theanalysis of its can service for marketing and economic development strategies. Throughanalysis of store receipts, we were able to get a lot of information, such as consumer spendinglevels, spending habits, favorite of some commodities. More importantly, businesses throughthe analysis of shopping receipts can get more commercial information, and the consumer’sfamily and friends through the analysis of them to understand consumers living conditions.Combination of market demand and the needs of the people, this paper do a certaindegree of pretreatment research of image recognition of shopping receipts,the work mainly inthe following several parts:Classifies the collected image of the shopping receipt, and because there may be apostmark or scratches on the part of the s shopping receipts,we need processing it before wemake imagetreatment of the gray-scale image, binarization, smooth it and tilt correction. Atthe same time, compare the image processing algorithm of each stage.After pre-treatment,we make layout analysis of shopping receipts images. Based on thecharacteristics of shopping receipts this paper selects the layout analysis algorithm based onthe expansion algorithm to split the shopping receipt layout.Do character segmentation processing for shopping receipts,and proposed two methodsof segmentation which are based on the maximum height and width of the character andbased on connected domain. The experimental results are satisfactory.Do normalization and thinning processing of the segmented character. This article uses arefinement method based on mathematical morphology, and improves the method，comparedthe pros and cons of the two algorithms. The experiments show that，for the improvedthinning algorithm，the treatment effect is better in the intersection point, and strokes aresmoother.