The Design And Implementation Of Food Anti-fake Label Image Recognition Based On Android Platform

Food safety has profound influence on the development of harmonious society.Recenty, events on food safety occured all the time, so the security and traceabil ity of food is imminent. As an inseparable part of food, label can provide the reli able information carrier on security and traceability. Based on the design of a sec urity food label in project “research and demonstration on electronic Traceability Technology of food safety”,this paper do some research and exploration for the d esign and development of automatic recognition using the labeling. The main resea rch are character segmentation,character recognition and label authentication under the environment of mixed text with english and chinese, and implement the food traceability function based on tag identification.the recognition of label image includes image preprocessing, character recognition,p hysical feature extraction and traceable information acquisition. image preprocessing is t he base of label image recognition, by comparing and summarizing the advantages an d disadvantages of different ways from experiments, this paper analyzes existing approac h of image preprocessing and therefore proposes prelimilary processing scheme complian t with the application in question. Using the character recognition engine directly to carr y out the process fails to achieve its effects owing to the mixed arrangement of Chinese-E nglish characters in the label's text area. Under the circumstance,through studying the la yout structure of characters, this paper raises an optimal project based on the reco gnition rate determined by character types so as to better solve the problems of t he Chinese characters' false segmentation and significantly improve the accuracy of its recognition.Considering the variety of physical characteristics, the application's c ounterfeit module adopts hot deployment plan and analyzes class loading mechanis m of Android platform to fulfill dynamic loading method of physical characteristic s' extraction algorithm, completing feature adaptive extraction while at the same ti me reducing memories occupied by the application. Furthermore, echoing the high concurrency requirements of traceability, this paper designs load balancing strategy for the entire application system, making it more stable and reliable.By exploiting technology of Android NDK and open source project——Open CV, this paper has brought about the above algorithm on an Android mobile phon e, accomplishing related software development. After testing and experimenting, th e software obtains preferable recognition effect on label images.