Research On Algorithm Of Characteristic Acquisition And Recognition Of Pulse Manifestation

The research of this article is mainly based on the subject "the client equipment's development of mobile telemedicine support system". Along with the development of computer and sensor technology, how to process Pulse signals in terms of Chinese Medicine is becoming more and more necessary. It is significant both in theory and practical application to study algorithms for characteristics acquisition and recognition of human pulse signal.There have been a lot of reports about the usage of wavelet transform in carrying on the filter pretreatment and signal detecting. There are also a lot of works done focused on how to use neural network to classify and recognize the pulse manifestations. But these techniques are far from perfect. In this paper, based on summarizing others' researches the following works have been done.Firstly, based on comparing different objective algorithms of human pulse manifestations, considering that most traditional methods are apt to obtain the information only from time-domain and can not preserve all the original information of pulse, Wavelet Transform (WT) is adopted to carry on the filter pretreatment, analyze aperiodic pulse and acquire its new characteristics that are the energy in different scales of pulse wavelet transform.Secondly, artificial neural network (ANN) has a strong self-adaptability and self-learning ability as well as excellent robustness and tolerance ability. By the network, an optional nonlinear input-output mapping relationship can be realized. In view of the shortage of Newton algorithm and conjugate grads algorithm, which is the convergence speed and convergence accuracy are not independent to initial weight value, using BP network which is based on Levenberg-Marquardt algorithm, a discerning system has been created successfully.In clinical area, the classification has some thing of fuzzy character. And fuzzy theory is more and more used in different fields. In this paper, a discerning system with a fuzzy neural network has been designed and it works well with nine different kinds of human pulses.Finally, the algorithms designed above are used in a system of pulse recognition. They are proved to be feasible and effective in pulse discerning. The research shows that the exactness ratio of both algorithms is high, the exactness ratio of BP network based on Levenberg-Marquardt algorithm is 83. 3% and the exactness ratio of fuzzy neural network is 88. 9%. As the algorithm implemented in this paper are in accordance with characteristics of pulse and Chinese pulse manifestations. Hence, it could be used for reference in objective research and discerning of human pulse and other pattern recognition systems.