Study On Extraction Algorithm Of Palm - To - Length Ratio

As the occurrence of breast cancer has been greatly risen, early detection and treatment has became the key to improve the survival rate of breast cancer patients. Currently gene detection is a significant susceptibility estimate approach for breast cancer, which have a certain effect on early detection of breast cancer, but this method shows some drawbacks, such as high cost, trauma, low efficiency and the low detection rate.Biological research notes that human finger length ratio is related to breast cancer. Therefore if it is possible to extract susceptibility characteristics of breast cancer from palm, it will provide a noninvasive, simple method for susceptibility estimate approach of breast cancer and improve the cure rate of the breast cancer patients. Currently, conventional method of extracting finger length ratio uses electronic vernier to manually caliper hand features then obtains the required information through statistics and calculation. Manual feature acquisition has some problems such as low efficiency, low accuracy and poor repeatability.In order to extract the digit ratio accurately and efficientively, this thesis accomplishes researches in the following aspects.1、An improved plate-rotating algorithm is proposed to obtain the finger length ratio of human hand. Firstly, hand shape images are collected through a digital camera by using non-physical-contact mode. Secondly, these hand shape images are pre-processed, such as color to gray conversion, binarization, the median filter and the corrosion expansion. Thirdly, the outline of the hand shape image is extracted. Since the improved plate-rotating algorithm is used for absolutely single pixel contour so the absolutely single pixel of the hand shape contour is realized by tracking the outline of hand shape. Fourth, the absolutely single pixel contour is traversed and the intersection angle is calculated simultaneously. Finally, the feathers of hand shape images (finger tip, finger valley) are located by analyzing the relevant angles as well as the right hand and left hand can be identified.2、A hardware system to obtain the hand shape images is designed, which hand shape images are collected through a digital camera by using non-physical-contact mode.3、A software of correlation analysis is designed. Firstly, the interface is connected with the database, so the extracted palm features can be saved in the database. Secondly, the finger length ratio and the result of left and right hand identification can be displayed. Thirdly, the other required information such as ATD angle, fingerprint, palm print, symmetries and so on can also be displayed in the future study through the extended interface. Finally, the required information which is already extracted can be saved to the database.4、Preliminary study on the relation of finger length ratio and breast cancer has been down.The proposed method can accurately locate the finger tip, finger valley and finger root from the obtained hand shape images by using the improved plate-rotating algorithm, and finally calculate the finger length ratio. I complete analysis on the relation of finger length ratio and breast cancer and draw conclusion that digit ratio can be a susceptibility characteristics of breast cancer.