Palm Diagnosis System Algorithm Based On Multispectral Palmprint Image

In recent years,the pattern recognition technology has obtained rapid development and wide application,especially in the field of identity recognition based on the palmprint features.However,the research on the application of the automatic palm diagnosis technology based on Traditional Chinese Medicine(TCM)is still at an infancy stage.The realization and application of the automatic palm diagnosis technology can help people aware the underlying disease and prevent it from getting worse in time.It is also playing an important role in pushing forward the development of TCM.When the special shapes of the palmprint appear on some regions of hands,it indicates that the corresponding organs of our bodies may have certain health problems.The automatic palm diagnosis technology,which combining with the digital picture processing technique,can use the computer to recognize the abnormal pathological palmprint of hands automatically.The key procedures of the automatic palm diagnosis technology include the partition acquisition of the palm image,the image preprocessing,and the detail extraction of palmprint,the recognition of pathological wrinkles.Above all,the acquisition and recognition of the pathological palmprint are two essential steps.How to pick up the complicated tiny wrinkles effectively,and recognize the underlying pathological palmprint accurately is still a difficult and challenge task.With the introduction of the multispectral technology,the current study is mainly focus on the setup and use of the automatic palm diagnosis system.The main contributions of this work are as follows:(1)A multispectral palmprint acquisition instrument is designed.For the purpose of collecting the palm multispectral images of different parts on hands much more clearly,the multispectral palmprint acquisition instrument designed in this paper can obtain six different spectral magnification images without touching the palm.Thereafter,a multispectral palmprint images database also can be established after finishing the preprocessing works,such as ROI extraction,image enhancement also can be proceeded by this acquisition instrument.(2)The extraction of tiny wrinkles.The extraction methods proposed in this work is based on the characteristics of multispectral palmprint images.An optimum selection of the multispectral palmprint images wavelengths is be made first according to the objective indices.In addition,a fine wrinkles extraction method is proposed based on the Nonsubsample Shearlet transformation of multispectral information fusion and morphological operations.Finally,the superiority and effectiveness of this extraction method are verified by series of comparative experiments.(3)The extraction and recognition of closed pathologic palmprint.A closed loop searching algorithm is designed firstly based on the degree characteristics of pixels.This algorithm can make a selection rules in accordance with the actual closed wrinkle features which can be used to diagnose the disease.Through calculating the area of the minimum circumscribed rectangle,the corresponding rectangle degree can be obtained.Therefore,a method based on the rectangle degree and eccentricity ratio is developed to recognize the six different closed pathologic palmprint.Three objective evaluation standards are set and several comparative and recognition experiments are conducted at last.The experiments results demonstrated that the extraction and recognition algorithm proposed in this paper can recognize the pathologic palmprint much more accurately.(4)The establishment of automatic palm diagnosis system.The automatic palm diagnose system is visualized by MATLAB GUI interface.It contains many viewable functions of reading and storage the palmprint images,preprocessing,the extraction and fusion of the wrinkle features,pathologic palmprint recognition,the disease indication based on TCM database.