A Detective Platform For Identity Classification Based On Biological Features Of ECG Signals

As a kind of humanity electronic data,ecg signal has a great importance on the area of health protection,disease detection,identification system and so on.Apart from common complex ecg signal detection devices in hospital,there is a popular usage in the field of wearable devices.These devices feature in smart,lower injection,high demand of lower power consumption.However,the traditional wearable devices exist of fatal drawbacks that poor time of the battery consumption.In this paper,a kind of lower power consumption ecg signal and identification detection system based on the theory compressed sensing is presented.Also this system has a great adapter property and expanded performance.The MCU of STM32F103C8T6 is applied in this ecg signal detection system and the mature AD8232 integration module is used as the ecg signal sample font end.Also single leader link paten is used to sample ecg data.Bluetooth HM06 is applied for the wireless transport module.The procedure of this system works as follows: the MCU controls the 12-bit AD module to sample the ecg data which is from the ecg signal sample font end.And then the MCU transports the compressed ecg data to PC client via the Bluetooth module.When PC client received the ecg signal data,a Stair Random Binary Matrix based on the compressed sensing theory roles as sensing matrix to compress the original data,then matlab tool kits are used to reconstruct the origin ecg signal based the recovery matrix and Orthogonal Matching Pursuit algorithm.Also the process of wave filter and denoising is conducted for preparation of features extraction for identification with libSVM tools based support vector machine theory.Experimental results show that stair random binary matrix this work proposed has more perfect performance compared with other six measurement matrix and this system has higher compressed radio at 40%,acceptable accuracy at 87.5% via train data and test data and reconstruction error at 0.0668.So this system satisfies the required performance of wearable devices.