Research On Accurate And Rapid Method To Identify Precious Metal Jewelry Images

With the development of economy and the improvement of living standards,the demand for jewelry has increased significantly.However,the classification of jewelry is completed manually with high labor intensity,low efficiency and error easily at present.With the development of image processing and recognition technology,image recognition technology has been widely applied in production and life.This paper studies the use of digital image processing technology to achieve accurate and rapid identification of jewelry images.The paper takes jewelry as the recognition object and uses camera to collect jewelry images.Digital image processing technology is used to process the image,including preprocessing,feature extraction and recognition.By extracting the features of area,quality,color and distance from jewelry,two sets of recognition schemes based on color feature and distance feature are developed and compared.On the software platform of Visual C++,using C++programming to develop jewelry recognition system and design recognition interface has great theoretical value and practical application value.The main contents of this paper are as follows:(1)Jewelry image acquisition system is designed.The parameters of camera,electronic scale,light source type and lighting form are selected and tested.The connection and data transmission between hardware devices are realized.Flow chart of jewelry recognition is designed.Jewelry collection background is selected,and collect jewelry images and intercept regions of interest.(2)The preprocessing of jewelry images includes selection and extraction of color components,the selection of threshold of image segmentation,graying,binarization,noise removal and edge extraction.G plus R combined with the threshold are used to get binary image.Template filtering is used to remove noise at one time.The edge of binary image is extracted by the 2×2 template combined with the threshold.(3)Quality feature and area feature of jewelry are obtained firstly.The centroid coordinates of jewelry images are extracted,the autocorrelation values of the color feature are calculated by the recognition scheme based on the color feature,and the autocorrelation values of the distance feature are calculated by the recognition scheme based on the distance feature.These features are stored at the end of the file of image data in JPEG format,and images with feature information form an image database.(4)Features data from the pending jewelry and sample jewelry is processed,scale constraint are used in the quality and area features,and the distance and color features are graphically constrained.The data from distance and color features are processed by autocorrelation and cross-correlation algorithm,and the correlation coefficient is used to judge whether types of jewelry.(5)Jewelry recognition system is developed based on platform of Visual C++.The structure,process and man-machine interface of the system are designed,and the operation steps of the system are explained.The main functions include sample image acquisition and name input,real-time video display,selection of sample file storage path and recognition result.The system is evaluated from two aspects:accuracy and time consumption.The experimental results show that the accuracy of the recognition system is 96%,and the average recognition time is 3 seconds.