Research On Human Pulse Signal Extraction Based On Digital Speckle Correlation Method

As an important physiological health signal of human body,pulse plays an important role in cardiovascular disease diagnosis and other medical processes.At this stage,contact medical equipment is usually used to measure the pulse wave accurately.As a new non-contact optical detection method,the digital speckle correlation method(DSCM)has been widely used in many industries such as industrial measurement and biomedicine because of its simple operation and accurate measurement.The DSCM records the digital speckle images of the moving object to compute the surface displacement and estimate the whole field strain according to the correlation of the gray features of the speckle image.In this paper,the digital speckle correlation method is applied to the field of human pulse detection,which realizes the non-contact and non-invasive detection of the human pulse.It is especially suited for patients with extensive skin burn or others cannot be directly contacted,which greatly increase the means of diagnosis and treatment to ameliorate the experience of patients.The main research of this thesis focuses on the realization of human pulse detection based on digital speckle correlation method and the performance improvement of digital speckle correlation method.The specific contents are as follows:Firstly,this thesis expounds on the research significance of digital speckle correlation method,the research status of theory and application,the characteristics of pulse wave and the source of noise.As for the digital speckle correlation method,its basic principle and related theories are widely studied,including speckle image preparation for experimental analysis,the composition of image acquisition system,the selection of shape function and correlation function,displacement search algorithm and so on.An optimized coarse fine search algorithm based on correlation function threshold is proposed to solve the common local extremum problem and ensure the search accuracy to 100%,and still keep a high search speed.Secondly,in order to improve the reliability of subset correlation matching,the accuracy of subpixel displacement search algorithm and the efficiency of displacement search in the measurement process of digital speckle correlation method.The subset size selection rules are given based on the sum of square of subset intensity gradients(SSSIG)and time limit of correlation search algorithm.In addition,the optimal speckle radius is studied.The experimental results show that there is an optimal speckle radius to minimize the total measurement error for different subset sizes.When the speckle size is not the optimal speckle radius,the subset size can be adjusted to make the speckle size of the current image conform to the optimal speckle radius,so as to make the measurement error reach the best level,which also provides an idea for the subset size selection.Finally,after experiments of speaker vibration detection and motor vibration detection are carried out to verify the detection principle and the experimental devices,the human pulse signal is successfully measured.A fusion filtering technique based on ensemble empirical mode decomposition(EEMD)and wavelet reconstruction is proposed to comprehensively filter out the light interference,baseline wander and motion artifacts in the extracted pulse signal to identify the pulse wave better.A graphic user interface program about digital speckle correlation method is also written which covers many functions to be easily used.