a. inflating a cuff communicated with a first pressure sensor which has been zeroed, and then stopping inflation after a cuff pressure reaches a preset value;

b. deflating the cuff at a preset speed and detecting whether a pulse signal occurs;

c. after the pulse signal is detected, processing data with an algorithm for recovering a trend envelope of an oscillating pulse wave, and displaying the measured blood pressure;

wherein during the step a, the first pressure sensor is zeroed in a state of communicating with atmosphere; and

during the step c, the algorithm for recovering the trend envelope of the oscillating pulse wave is a nonlinear fitting algorithm, wherein data processing during the step c includes the following steps:

C1: sampling during the deflation process, and thereby obtaining a cuff pressure X and a pressure pulse wave amplitude Y of each sampling point;

C2: taking a cuff pressure X0 and a pressure pulse wave amplitude Y0 of a first sampling point among the sampling points as initial values, and selecting a pressure step Δ;

C3: selecting at least three adjacent sampling points while sequentially moving point-by-point on the basis of the sampling points, and then performing data processing as follows for the sampling points until the last sampling point;

piecewise performing trend term fitting by constructing a multiple-order curve with said at least three adjacent sampling points, generating a cuff pressure Xm and a pressure pulse wave amplitude Ym of an interpolation point between two adjacent sampling points of said at least three adjacent sampling points and then saving Xm and Ym, wherein Xm is within a cuff pressure interval of said two adjacent sampling points, Xm=X0−mΔ, while each of n and m is a natural number which increases by one during such fitting process;

C4: performing smooth processing by taking said pressure pulse wave amplitude of each saved interpolation point as an intermediate value, averaging the amplitude of each interpolation point with amplitudes of at least two adjacent interpolation points while moving point-by-point, and thereby obtain mg the smoothed amplitude of each interpolation point and then saving it; and

C5: searching the maximum of pulse wave amplitude among the saved and smoothed amplitudes, constructing a multiple-order curve with the data of the point corresponding to the maximum of pulse wave amplitude and at least one point on either side thereof, and calculating the maximum of pulse wave amplitude of the multiple-order curve as the extreme value of trend envelope curve of the oscillating pulse wave, wherein a cuff pressure corresponding to the extreme value is an average pressure.