Research On The Method Of Face Image Quality Analysis In Auto 4S Shop

With the help of face recognition system,the sales staff of auto 4S shop can quickly obtain information about the customers entering the store,and thus provide targeted services to them.However,in practical applications,the collected face images often suffer from degradation problems.Through face image quality analysis,the low-quality face image will be effectively limit into the system,and then improve the system’s recognition performance.Because the process of subjective quality analysis of face images is cumbersome and difficult to meet the needs of real-time system,the study of objective analysis methods is particularly important.Illumination and sharpness distortion have a great influence on the quality of face images,and the existing method of face image quality analysis lacks integrity in the investigation of light and sharpness.According to analyses above,this thesis takes the auto 4S shop face recognition system as the basic research framework,and carries out related research on illumination analysis and sharpness analysis of face image in auto 4S shop.The main research contents of this thesis are as follows:1.This thesis theoretically researched the method of face image quality analysis.Firstly,some subjective and objective methods of face image quality analysis were learned in this thesis.Secondly,their advantages and disadvantages were summarized and compared.Then,various factors that make the quality of face images decrease were generalized,and difficulties in face image quality analysis were discussed.Finally,the thesis simply predicts the future of this research direction.2.This thesis researched the illumination analysis method of face images in auto 4S shop.Firstly,the influence of illumination changes on the face images was analyzed in this thesis.Secondly,a common non-reference image quality index is improved in combination with the face symmetry.In case that the lighting symmetry index can’t accurately reflect the illumination quality,the illumination intensity index is further studied in this thesis.Finally,the lighting symmetry index with the illumination intensity index were merged by the product method to form the standard of illumination quality analysis,and completely achieved illumination quality analysis of face images in auto 4S shop.In the end,experiments on the face database of auto 4S shop and Yale B verify that the method of this thesis has improved the recognition performance of auto 4S shop face recognition system.3.This thesis researched the sharpness analysis method of face images in auto 4S shop.Firstly,the influence of sharpness changes on the face images was studied in this thesis.Secondly,the spectral characteristics of blurred images was analyzed.In order to quantitatively analyze the low-frequency components of spectrum map that reflect the degree of blurring,measures of spectrum binarization are further taken on the basis of spectrum shifting.Finally,the sharpness quality analysis standard of this thesis was formed,and achieved illumination quality analysis of face images in auto 4S shop.In the end,experiments on the face database of auto 4S shop and Yale B verify that this method can improve the recognition performance of auto 4S shop face recognition system to some extent.4.Firstly,this thesis effectively integrated the illumination and sharpness quality analysis methods.And then a method of face images quality analysis in auto 4S shop that can comprehensively investigate the illumination,sharpness,and background interference is formed.Finally,experimental results on the face database in auto 4S shop verify the effectiveness of research method studied in this thesis.