Pulse Motion Analysis Methods Based On Dynamic Image

Pulse diagnosis plays an important role in Traditional Chinese Medicine (TCM), as the method of non-invasive, pulse diagnosis has been applied to clinical diagnosis. But pulse theory is not systematic, and the standard of distinguishing symptom is not clear, so as a result, it is difficult for TCM to popularize in the practice.Along with the development of the technology of sensors and computer technology, there has been a great evolution in the domain of the research about pulse acquisition sensors. People look forward to using modern technology in the TCM diagnose to realize the reality and character of pulse more scientifically. How to acquire multidimensional pulse information based on image is now a quite new research area. In this paper it is proposed to use one kind of image pulse information collecting device made by ourselves to obtain multidimensional, dynamic, overall pulse information. This method can not only promote the realistic study of TCM, but also create the condition for automatic pulse diagnosis. It has broad prospect of clinical application.In this paper we first analyze the mechanism about the generation of the pulse. From the analysis we may know that, the formation of pulse is very complicated for it is related with many kinds of being tissues and organs. So it is not enough to only collect the information of pulse pressure. It is necessary to implement 3-dimensional and dynamic related information collecting.Starting from bionics and using the self-made pulse acquisition device to collect 3-dimensional and dynamic pulse information at radial artery. Comparing with previous pulse acquisition sensors, the proposed sensor can collect multidimensional image, and then make analyzing pulse information image possible. The collected dynamic video stream can be changed into single frame images, so we can extract the image characters by enhancing the contrast, smoothing the pictures, detecting the edges and so on.Subsequently, in the estimation of three-dimensional movement, the move tracks of some specific points can be captured by building movement model and parameter estimation. After that, the optical-flow fields are computed respectively by using standard optical-flow method and constructing pyramid-style multi-resolution structure. As a result of the experiment, thesemethods can reflect the movement of pixels on film pushed by pulse power quite clearly, as well as detect the broadcasting characteristic of pulse wave. The methods provide definitely scientific evidence for realistic pulse diagnosis.