Research On Visualization Of Palatal Rugae Recognition

As a new biological feature,palatal rugae is one of human body recognition tools in forensic identification because of its heredity,uniqueness and stability.At present,the identification method based on palatal rugae information has not been systematically proposed.Therefore,it is necessary to conduct a further research on palatal image analysis,feature extraction and recognition algorithms.In this paper,according to the features of palatal rugae images,the study focuses on the preprocessing,segmentation,feature extraction and identification of palatal rugae images to search for the best possible method in order to achieve the purpose of identification.1.Palatal rugae image preprocessing: in this paper,the features of palatal images that obtained by different acquisition methods are compared in order to find the best ones.According to the results,the images obtained by making oral palate plaster models are selected as experiment pictures.Then they are tailored in the same way and all changed into gray images.Finally these images are smoothed with different filters which are expected to maintain the edges and details of images as well as reduce the noise.As the result,the bilateral filter showed a better performance.2.Palatal rugae image segmentation: after smoothed by bilateral filter,palatal rugae images are firstly segmented using several frequently-used edge detection operators and classical threshold segmentation algorithms.Results show the edge detection operator and simple threshold segmentation algorithm could not effectively isolate the palate wrinkles.Therefore,a fuzzy C-means clustering algorithm combined with threshold segmentation is proposed.The palatal rugae image processed with the proposed method is clear without other impurities in the background,which meets the requirement of feature extraction of palatal rugae.3.Palatal rugae feature extraction and recognition: firstly,the palatal rugae region of the segmented image is filled using seed filling method based on a scanning line.So it's easy to calculate the eigenvalue of the palatal wrinkle.Then the basic features of the images and the eigenvalue calculation method are briefly introduced.The basic characteristics of the palatal rugae in the binary image are extracted and calculated,including the number of palatal rugae lines and the duty cycle.Finally,we design the decision tree classifier to classify and code palatal rugaes.With the code,palatal rugae can be recognized.When designing the palatal rugae recognition system,we can add the individual palatal rugae code and personal information to the database at the same time.Palatal rugae code is used to match the samples in database,then the forensic medicine identification can completed correctly.