Research On Pulse Monitoring Method Based On Superpixel Edge Detection

The development of society is accompanied by the increase of chronic diseases,and people want to know their physiological conditions conveniently at all times in daily life.Especially for young people with heavy work pressure and elderly people with physical problems,it is very important to realize daily heart rate monitoring.However,at present,the pulse detection technology in the market mostly adopts contact equipment,but it is not conducive to daily monitoring due to cost and technical reasons.Therefore,non-contact detection method has gradually become the mainstream direction of current research.In this thesis,based on the theory of superpixel segmentation and edge detection in computer vision,the wrist ulnar artery is taken as the research target for non-contact measurement,and the pulse wave signal extraction method based on image vision processing is studied to solve the problems of low precision,poor real-time performance and complex post-processing in skinbased pulse detection.Using simple linear iterative clustering algorithm,the image is divided into several superpixel blocks with good boundary conditions,and the waveform signal is extracted according to the change of the gray value of the edge of the superpixel.Because the pixel blocks formed by the supepixel algorithm may be over-divided or under-divided,the gray value obtained later will be null and can not provide edge information,so the edge detection accuracy of the pulse wave signal is improved by combining the superpixel algorithm with the edge detection algorithm.Compared with only using superpixel algorithm,the technology that combines superpixel with edge detection can filter out some noise and unnecessary tiny details.After the formation of the superpixel sub-block,according to the change of the gray value of the superpixel edge of each frame of the target area image in the video set,the pulse wave waveform of the ulnar artery of the wrist is extracted,and the high and low frequency noise signals and motion artifacts are removed by Butterworth filter filtering,wavelet denoising and empirical mode decomposition,so as to obtain a smooth and complete waveform,and then the waveform is analyzed to monitor the heart rate.According to the comparative experimental study,the heart rate value extracted by this method is compared with the heart rate value extracted by the non-contact measurement method which is widely used at present and the heart rate value measured by the portable detection instrument which is more popular in the market.Qualitative analysis and quantitative analysis are used to analyze the experimental results,and the results show that the heart rate results extracted by this method have high accuracy and consistency.